abstract
stringlengths 383
2.62k
| TLDR
stringlengths 4
250
|
|---|---|
We present a Lipschitz continuous Transformer, called LipsFormer, to pursue training stability both theoretically and empirically for Transformer-based models. In contrast to previous practical tricks that address training instability by learning rate warmup, layer normalization, attention formulation, and weight initialization, we show that Lipschitz continuity is a more essential property to ensure training stability. In LipsFormer, we replace unstable Transformer component modules with Lipschitz continuous counterparts: CenterNorm instead of LayerNorm, spectral initialization instead of Xavier initialization, scaled cosine similarity attention instead of dot-product attention, and weighted residual shortcut. We prove that these introduced modules are Lipschitz continuous and derive an upper bound on the Lipschitz constant of LipsFormer. Our experiments show that LipsFormer allows stable training of deep Transformer architectures without the need of careful learning rate tuning such as warmup, yielding a faster convergence and better generalization. As a result, on the ImageNet 1K dataset, LipsFormer-Tiny training for 100 epochs without learning rate warmup attains a top-1 accuracy of 81.6\% which is higher than Swin Transformer-Tiny training for 300 epochs with warmup. Moreover, LipsFormer-Tiny training for 300 epochs achieves a top-1 accuracy of 83.5\% with 4.7G FLOPs and 24M parameters. | We propose a Lipschitz continuous Transformer. |
Recent studies on SGC, PageRank and S\textsuperscript{2}GC have demonstrated that several graph diffusion techniques are straightforward, quick, and effective for tasks in the graph domain like node classification. Even though these techniques do not even need labels, they can nevertheless produce more discriminating features than raw attributes for downstream tasks with different classifiers. These methods are data-independent and thus primarily rely on some empirical parameters on polynomial bases (e.g., Monomial and Chebyshev), which ignore the homophily of graphs and the attribute distribution. They are more insensitive to heterophilous graphs due to the low-pass filtering. Although there are many approaches focusing on GNNs based on heterophilous graphs, these approaches are dependent on label information to learn model parameters. In this paper, we study the question: are labels a necessity for GNNs with heterophilous graphs? Based on this question, we propose a framework of self-representation on graphs related to the Least Squares problem. Specifically, we use Generalized Minimum RESidual (GMRES) method, which finds the least squares solution over Krylov subspaces. In theoretical analysis, without label information, we enjoy better features with graph convolution.
The proposed method, like previous data-independent methods, is not a deep model and is, therefore, quick, scalable, and simple. We also show performance guarantees for models on real and synthetic data. On a benchmark of real-world datasets, empirically, our method is competitive with existing deep models for node classification. | We define a learnable and unsupervised graph convolution framework as self-representation on graph. |
Biological neural networks are characterised by a high degree of neural diversity, a trait that artificial neural networks (ANNs) generally lack.
Additionally, learning in ANNs is typically synonymous with only modifying the strengths of connection weights.
However, there is much evidence from neuroscience that different classes of neurons each have crucial roles in the information processing done by the network. In nature, each neuron is a dynamical system that is a powerful information processor in its own right. In this paper we ask the question, how well can ANNs learn to perform reinforcement learning tasks only through the optimization of neural activation functions, without any weight optimization?
We demonstrate the viability of the method and show that the neural parameters are expressive enough to allow learning three different continuous control tasks without weight optimization.
These results open up for more possibilities for synergies between synaptic and neural optimization in ANNs in the future. | We optimize parameterized activation functions in fixed random networks to solve reinforcement learning tasks. |
This paper proposes a new parameter-efficient method for fine-tuning, AoT P-Tuning. This method adds input-dependent biases before evaluating the Transformer layer, reducing the required evaluation time when compared to P-Tuning. Same as P-Tuning, AoT P-Tuning allows multi-task inference with a single backbone model for evaluating different tasks in a single batch.
We experimented with the proposed method on the GLUE and SuperGLUE benchmarking datasets using RoBERTa-Base, RoBERTa-Large, and DeBERTa-XL backbone models. Our observations show that AoT P-tuning performed on par with or better than P-Tuning v2 while being up to $1.3\times$ times faster during inference. | A novel method for parameter efficient fine-tuning. Can perform multi-task inference like P-Tuning, but up to 1.3x times faster than it. |
Long sequence modeling of Transformer gains prevalence in fields involving long texts and high-resolution images and videos but suffers from quadratic memory complexity. Existing work investigates low-complexity variants or parallel methods to handle it. The former attempts to approximate full attention and is limited by a single device's capacity. The latter struggles to manage quadratic memory of attention maps, leading to insufficient sequence scalability. In this work, we propose a novel parallel method named $\textbf{M}$icro-$\textbf{Q}$uery $\textbf{S}$equence $\textbf{P}$arallelism. MQSP slices sequences across devices and projects local queries, keys, and values in self-attention. For communication and memory efficiency, MQSP all-gathers the queries while keys and values remain locally to acquire the local attention map, on which a distributed softmax gets conducted to amortize memory by column. Meanwhile, the queries get further partitioned as Micro-Q to divide the computation and recycle the attention map by row, jointly decomposing the quadratic memory to achieve linear scalability. The evaluation result shows that MQSP scales up sequence length linearly and achieves 4.5$\times$ sequence length of ColossalAI's sequence parallelism and 4.3$\times$ of Megatron-LM3, enabling training BERT-large of 78848 sequence length on 32 A100 GPUs. MQSP can reduce up to 78.6$\%$ of memory occupation and achieve up to 3.3$\times$ throughput when training on 17408 sequence length. The convergence quality experiment proves that MQSP provides the means for long sequences with guaranteed convergence, bringing the potential for the Transformer to explore longer sequences. | MQSP is a novel sequence parallelism that linearly scales long sequence Transformers through all-gathering Micro-Q. |
Unsupervised representation learning leverages large unlabeled datasets and is competitive with supervised learning. But non-robust encoders may affect downstream task robustness. Recently, robust representation encoders have become of interest. Still, all prior work evaluates robustness using a downstream classification task. Instead, we propose a family of unsupervised robustness measures, which are model- and task-agnostic and label-free. We benchmark state-of-the-art representation encoders and show that none dominates the rest. We offer unsupervised extensions to the FGSM and PGD attacks. When used in adversarial training, they improve most unsupervised robustness measures, including certified robustness. We validate our results against a linear probe and show that, for MOCOv2, adversarial training results in 3 times higher certified accuracy, a 2-fold decrease in impersonation attack success rate and considerable improvements in certified robustness. | We provide a framework for robustness evaluation and adversarial training of representation encoders without the need for labelled data. |
Model-based reinforcement learning (MBRL) applies a single-shot dynamics model to imagined actions to select those with best expected outcome. The dynamics model is an unfaithful representation of the environment physics, and its capacity to predict the outcome of a future action varies as it is trained iteratively. An experience replay buffer collects the outcomes of all actions executed in the environment and is used to iteratively train the dynamics model. With growing experience, it is expected that the model becomes more accurate at predicting the outcome and expected reward of imagined actions. However, training times and memory requirements drastically increase with the growing collection of experiences.
Indeed, it would be preferable to retain only those experiences that could not be anticipated by the model while interacting with the environment.
We argue that doing so results in a lean replay buffer with diverse experiences that correspond directly to the model's predictive weaknesses at a given point in time.
We propose strategies for: i) determining reliable predictions of the dynamics model with respect to the imagined actions, ii) retaining only the unimaginable experiences in the replay buffer, and iii) training further only when sufficient novel experience has been acquired.
We show that these contributions lead to lower training times, drastic reduction of the replay buffer size, fewer updates to the dynamics model and reduction of catastrophic forgetting. All of which enable the effective implementation of continual-learning agents using MBRL. | This paper proposes techniques to add only the most relevant experiences in the replay buffer, using model uncertainty as selection criterion. |
Visually-situated language is ubiquitous---sources range from textbooks with diagrams to web pages with images and tables, to mobile apps with buttons and forms. Perhaps due to this diversity, previous work has typically relied on domain-specific recipes with limited sharing of the underlying data, model architectures, and objectives. We present Pix2Struct, a pretrained image-to-text model for purely visual language understanding, which can be finetuned on tasks containing visually-situated language. Pix2Struct is pretrained by learning to parse masked screenshots of web pages into simplified HTML. The web, with its richness of visual elements cleanly reflected in the HTML structure, provides a large source of pretraining data well suited to the diversity of downstream tasks. Intuitively, this objective subsumes common pretraining signals such as OCR, language modeling, image captioning. In addition to the novel pretraining strategy, we introduce a variable-resolution input representation and a more flexible integration of language and vision inputs, where language prompts such as questions are rendered directly on top of the input image. For the first time, we show that a single pretrained model can achieve state-of-the-art results in six out of nine tasks across four domains: documents, illustrations, user interfaces, and natural images. | We propose general-purpose pixel-to-text models that can be finetuned on tasks with visually-situated language, such as UIs, charts, figures, tables, documents, etc. |
Standard inference and training with transformer based architectures scale quadratically with input sequence length. This is prohibitively large for a variety of applications especially in web-page translation, query-answering etc. Consequently, several approaches have been developed recently to speedup attention computation by enforcing different attention structures such as sparsity, low-rank, approximating attention using kernels. In this work, we view attention computation as that of nearest neighbor retrieval, and use decision tree based hierarchical navigation to reduce the retrieval cost per query token from linear in sequence length to nearly logarithmic. Based on such hierarchical navigation, we design Treeformer which can use one of two efficient attention layers -- TF-Attention and TC-Attention. TF-Attention computes the attention in a fine-grained style, while TC-Attention is a coarse attention layer which also ensures that the gradients are "dense". To optimize such challenging discrete layers, we propose a two-level bootstrapped training method. Using extensive experiments on standard NLP benchmarks, especially for long-sequences, we demonstrate that our Treeformer architecture can be almost as accurate as baseline Transformer while using 30x lesser FLOPs in the attention layer. Compared to Linformer, the accuracy can be as much as 12% higher while using similar FLOPs in the attention layer. | Efficient Decision Tree based attention computation to reduce FLOPs for self-attention |
Reinforcement learning (RL) provides a powerful framework for decision-making, but its application in practice often requires a carefully designed reward function. Adversarial Imitation Learning (AIL) sheds light on automatic policy acquisition without access to the reward signal from the environment. In this work, we propose Auto-Encoding Adversarial Imitation Learning (AEAIL), a robust and scalable AIL framework. To induce expert policies from demonstrations, AEAIL utilizes the reconstruction error of an auto-encoder as a reward signal, which provides more information for optimizing policies than the prior discriminator-based ones. Subsequently, we use the derived objective functions to train the auto-encoder and the agent policy. Experiments show that our AEAIL performs superior compared to state-of-the-art methods in the MuJoCo environments. More importantly, AEAIL shows much better robustness when the expert demonstrations are noisy. Specifically, our method achieves $11\%$ and $50.7\%$ relative improvement overall compared to the best baseline GAIL and PWIL on clean and noisy expert data, respectively. Video results, open-source code and dataset are available in supplementary materials. | this paper presents a new adversarial imitation learning method based on auto-encoding |
We address the task of open-world class-agnostic object detection, i.e., detecting every object in an image by learning from a limited number of base object classes. State-of-the-art RGB-based models suffer from overfitting the training classes and often fail at detecting novel-looking objects. This is because RGB-based models primarily rely on appearance similarity to detect novel objects and are also prone to overfitting short-cut cues such as textures and discriminative parts. To address these shortcomings of RGB-based object detectors, we propose incorporating geometric cues such as depth and normals, predicted by general-purpose monocular estimators. Specifically, we use the geometric cues to train an object proposal network for pseudo-labeling unannotated novel objects in the training set. Our resulting Geometry-guided Open-world Object Detector (GOOD) significantly improves detection recall for novel object categories and already performs well with only a few training classes. Using a single ``person'' class for training on the COCO dataset, GOOD surpasses SOTA methods by 5.0% AR@100, a relative improvement of 24%. The code has been made available at https://github.com/autonomousvision/good. | We propose incorporating geometric cues into open-world object detector training and make significant improvements on various benchmarks. |
This paper presents a multi-agent reinforcement learning (MARL) scheme for proactive Multi-Camera Collaboration in 3D Human Pose Estimation in dynamic human crowds. Traditional fixed-viewpoint multi-camera solutions for human motion capture (MoCap) are limited in capture space and susceptible to dynamic occlusions. Active camera approaches proactively control camera poses to find optimal viewpoints for 3D reconstruction. However, current methods still face challenges with credit assignment and environment dynamics. To address these issues, our proposed method introduces a novel Collaborative Triangulation Contribution Reward (CTCR) that improves convergence and alleviates multi-agent credit assignment issues resulting from using 3D reconstruction accuracy as the shared reward. Additionally, we jointly train our model with multiple world dynamics learning tasks to better capture environment dynamics and encourage anticipatory behaviors for occlusion avoidance. We evaluate our proposed method in four photo-realistic UE4 environments to ensure validity and generalizability. Empirical results show that our method outperforms fixed and active baselines in various scenarios with different numbers of cameras and humans. | We propose a novel MARL framework to solve proactive multi-camrea collaborations for 3D HPE in human crowds |
We introduce CriticSMC, a new algorithm for planning as inference built from a composition of sequential Monte Carlo with learned Soft-Q function heuristic factors. These heuristic factors, obtained from parametric approximations of the marginal likelihood ahead, more effectively guide SMC towards the desired target distribution, which is particularly helpful for planning in environments with hard constraints placed sparsely in time. Compared with previous work, we modify the placement of such heuristic factors, which allows us to cheaply propose and evaluate large numbers of putative action particles, greatly increasing inference and planning efficiency. CriticSMC is compatible with informative priors, whose density function need not be known, and can be used as a model-free control algorithm. Our experiments on collision avoidance in a high-dimensional simulated driving task show that CriticSMC significantly reduces collision rates at a low computational cost while maintaining realism and diversity of driving behaviors across vehicles and environment scenarios. | We present a novel method called CriticSMC capable of being deployed in model-predictive planning and model-free online control cases within environments with hard constraints taking advantage of informative prior policies. |
Bridging geometry and topology, curvature is a powerful and expressive invariant. While the utility of curvature has been theoretically and empirically confirmed in the context of manifolds and graphs, its generalization to the emerging domain of hypergraphs has remained largely unexplored. On graphs, the Ollivier-Ricci curvature measures differences between random walks via Wasserstein distances, thus grounding a geometric concept in ideas from probability theory and optimal transport. We develop Orchid, a flexible framework generalizing Ollivier-Ricci curvature to hypergraphs, and prove that the resulting curvatures have favorable theoretical properties. Through extensive experiments on synthetic and real-world hypergraphs from different domains, we demonstrate that Orchid curvatures are both scalable and useful to perform a variety of hypergraph tasks in practice. | We introduce a flexible framework for Ollivier-Ricci curvature on hypergraphs. |
Federated Learning (FL) has been a popular approach to enable collaborative learning on multiple parties without exchanging raw data. However, the model performance of FL may degrade a lot due to non-IID data. While many FL algorithms focus on non-IID labels, FL on non-IID features has largely been overlooked. Different from typical FL approaches, the paper proposes a new learning concept called ADCOL (Adversarial Collaborative Learning) for non-IID features. Instead of adopting the widely used model-averaging scheme, ADCOL conducts training in an adversarial way: the server aims to train a discriminator to distinguish the representations of the parties, while the parties aim to generate a common representation distribution. Our experiments on three tasks show that ADCOL achieves better performance than state-of-the-art FL algorithms on non-IID features. | The paper proposes a new collaborative learning framework on non-IID features. |
Gradient-based meta-learning methods have primarily focused on classical machine learning tasks such as image classification and function regression, where they were found to perform well by recovering the underlying common representation among a set of given tasks. Recently, PDE-solving deep learning methods, such as neural operators, are starting to make an important impact on learning and predicting the response of a complex physical system directly from observational data. Since the data acquisition in this context is commonly challenging and costly, the call of utilization and transfer of existing knowledge to new and unseen physical systems is even more acute.
Herein, we propose a novel meta-learnt approach for transfer-learning knowledge between neural operators, which can be seen as transferring the knowledge of solution operators between governing (unknown) PDEs with varying parameter fields. With the key theoretical observation that the underlying parameter field can be captured in the first layer of the neural operator model, in contrast to typical final-layer transfer in existing meta-learning methods, our approach is a provably universal solution operator for multiple PDE solving tasks. As applications, we demonstrate the efficacy of our proposed approach on heterogeneous material modeling tasks, which shows that our method can handle complex and nonlinear physical response learning tasks while greatly improving the sampling efficiency in new and unseen materials. | Meta-learning method to transfer hidden physics |
Though end-to-end neural approaches have recently been dominating NLP tasks in both performance and ease-of-use, they lack interpretability and robustness. We propose Binder, a training-free neural-symbolic framework that maps the task input to a program, which (1) allows binding a unified API of language model (LM) functionalities to a programming language (e.g., SQL, Python) to extend its grammar coverage and thus tackle more diverse questions, (2) adopts an LM as both the program parser and the underlying model called by the API during execution, and (3) requires only a few in-context exemplar annotations. Specifically, we employ GPT-3 Codex as the LM. In the parsing stage, with only a few in-context exemplars, Codex is able to identify the part of the task input that cannot be answerable by the original programming language, correctly generate API calls to prompt Codex to solve the unanswerable part, and identify where to place the API calls while being compatible with the original grammar. In the execution stage, Codex can perform versatile functionalities (e.g., commonsense QA, information extraction) given proper prompts in the API calls. Binder achieves state-of-the-art results on WikiTableQuestions and TabFact datasets, with explicit output programs that benefit human debugging. Note that previous best systems are all finetuned on tens of thousands of task-specific samples, while Binder only uses dozens of annotations as in-context exemplars without any training. Our code is available at anonymized. | binding language models in symbolic languages |
Despite their remarkable ability to generalize with over-capacity networks, deep neural networks often abuse bias instead of using the actual task-related information for discriminative tasks. Since such shortcuts are only effective within the collected dataset, the resulting biased model underperforms on real-world inputs. To counteract the influence of bias, existing methods either exploit auxiliary information which is rarely obtainable in practice, or sift bias-free samples to exploit them for debiasing. However, such presumptions about the availability of the auxiliary information or bias-free samples are not always guaranteed and the existing methods could break down due to the unmet presumptions. In this paper, we propose Contrastive Debiasing via Generative Bias-transformation (CDvG) which is capable of operating without exploiting bias labels and bias-free samples explicitly. Motivated by our observation that not only discriminative models but also image translation models tend to focus on the easy-to-learn bias, CDvG employs a image translation model to transform the bias to another mode of bias while preserving task-relevant information. Through contrastive learning, we set transformed biased views against another, learning bias-invariant representations. Especially, as the bias has a stronger correlation or is easier to perceive compared to the signal, the translation model is more likely to be a bias translation model, resulting in better debiasing effect. Experimental results demonstrate that CDvG outperforms the state-of-the-arts, especially when bias-free samples are extremely scarce. | In this paper, we propose Contrastive Debiasing via Generative Bias-transformation (CDvG) which is capable of operating without exploiting bias labels and bias-free samples explicitly. |
Dynamical systems are found in innumerable forms across the physical and biological sciences, yet all these systems fall naturally into equivalence classes: conservative or dissipative, stable or unstable, compressible or incompressible. Predicting these classes from data remains an essential open challenge in computational physics on which existing time-series classification methods struggle. Here, we propose, phase2vec, an embedding method that learns high-quality, physically-meaningful representations of low-dimensional dynamical systems without supervision. Our embeddings are produced by a convolutional backbone that extracts geometric features from flow data and minimizes a physically-informed vector field reconstruction loss. The trained architecture can not only predict the equations of unseen data, but also produces embeddings that encode meaningful physical properties of input data (e.g. stability of fixed points, conservation of energy, and the incompressibility of flows) more faithfully than standard blackbox classifiers and state-of-the-art time series classification techniques. We additionally apply our embeddings to the analysis of meteorological data, showing we can detect climatically meaningful features. Collectively, our results demonstrate the viability of embedding approaches for the discovery of dynamical features in physical systems. | Unsupervised framework for learning high-quality, physically-meaningful embeddings of dynamical systems. |
We present MeGraph, a novel network architecture for graph-structured data. Given any input graph, we create multi-scale graphs using graph pooling. Then, we connect them into a mega graph by bridging inter-graph edges according to the graph pooling results. Instead of universally stacking graph convolutions over the mega graph, we apply general graph convolutions over intra-graph edges, while the convolutions over inter-graph edges follow a bidirectional pathway to deliver the information along the hierarchy for one turn. Graph convolution and graph pooling are two core elementary operations of MeGraph. In our implementation, we adopt the graph full network (GFuN) and propose the stridden edge contraction pooling (S-EdgePool) with adjustable pooling ratio, which are extended from conventional graph convolution and edge contraction pooling. The MeGraph model enables information exchange across multi-scale graphs, repeatedly, for deeper understanding of wide-range correlations in graphs. This distinguishes MeGraph from many recent hierarchical graph neural networks like Graph U-Nets. We conduct comprehensive empirical studies on tens of public datasets, in which we observe consistent performance gains comparing to baselines. Specifically, we establish 5 new graph theory benchmark tasks that require long-term inference and deduction to solve, where MeGraph demonstrates dominated performance compared with popular graph neural networks. | We present a novel graph network architechture learning on a mega graph derived by connecting multi-scale graphs. The architechture allows repeated information exchange across multiple scaled graphs. |
More and more evidence has shown that strengthening layer interactions can enhance the representation power of a deep neural network, while self-attention excels at learning interdependencies by retrieving query-activated information. Motivated by this, we devise a cross-layer attention mechanism, called multi-head recurrent layer attention (MRLA), that sends a query representation of the current layer to all previous layers to retrieve query-related information from different levels of receptive fields. A light-weighted version of MRLA is also proposed to reduce the quadratic computation cost. The proposed layer attention mechanism can enrich the representation power of many state-of-the-art vision networks, including CNNs and vision transformers. Its effectiveness has been extensively evaluated in image classification, object detection and instance segmentation tasks, where improvements can be consistently observed. For example, our MRLA can improve 1.6% Top-1 accuracy on ResNet-50, while only introducing 0.16M parameters and 0.07B FLOPs. Surprisingly, it can boost the performances by a large margin of 3-4% box AP and mask AP in dense prediction tasks. Our code is available at https://github.com/joyfang1106/MRLA. | A multi-head recurrent layer attention mechanism is proposed to retrieve query-related information from previous layers. |
This paper addresses learning end-to-end models for time series data that include a temporal alignment step via dynamic time warping (DTW). Existing approaches to differentiable DTW either differentiate through a fixed warping path or apply a differentiable relaxation to the min operator found in the recursive steps used to solve the DTW problem. We instead propose a DTW layer based around bi-level optimisation and deep declarative networks, which we name DecDTW. By formulating DTW as a continuous, inequality constrained optimisation problem, we can compute gradients for the solution of the optimal alignment (with respect to the underlying time series) using implicit differentiation. An interesting byproduct of this formulation is that DecDTW outputs the optimal warping path between two time series as opposed to a soft approximation, recoverable from Soft-DTW. We show that this property is particularly useful for applications where downstream loss functions are defined on the optimal alignment path itself. This naturally occurs, for instance, when learning to improve the accuracy of predicted alignments against ground truth alignments. We evaluate DecDTW on two such applications, namely the audio-to-score alignment task in music information retrieval and the visual place recognition task in robotics, demonstrating state-of-the-art results in both. | We introduce a novel differentiable dynamic time warping layer based on continuous time warps and implicit differentiation. |
Neural sequence models, especially transformers, exhibit a remarkable capacity for in-context learning. They can construct new predictors from sequences of labeled examples $(x, f(x))$ presented in the input without further parameter updates. We investigate the hypothesis that transformer-based in-context learners implement standard learning algorithms implicitly, by encoding context-specific parametric models in their hidden representations, and updating these implicit models as new examples appear in the context. Using linear regression as a model problem, we offer three sources of evidence for this hypothesis. First, we prove by construction that transformers can implement learning algorithms for linear models based on gradient descent and closed-form computation of regression parameters. Second, we show that trained in-context learners closely match the predictors computed by gradient descent, ridge regression, and exact least-squares regression, transitioning between different predictors as transformer depth and dataset noise vary. Third, we present preliminary evidence that in-context learners share algorithmic features with these predictors: learners' late layers encode weight vectors and moment matrices. These results suggest that in-context learning is understandable in algorithmic terms, and that (at least in the linear case) learners may work by rediscovering standard estimation algorithms. | We prove that the transformers can implement learning algorithms for linear models based e.g gradient descent, then observe they closely match the predictors of known algorithms, transitioning between different predictors as transformer depth vary. |
In this paper we introduce BO-Muse, a new approach to human-AI teaming for the optimisation of expensive blackbox functions. Inspired by the intrinsic difficulty of extracting expert knowledge and distilling it back into AI models and by observations of human behaviour in real-world experimental design, our algorithm lets the human expert take the lead in the experimental process. The human expert can use their domain expertise to its full potential, while the AI plays the role of a muse, injecting novelty and searching for areas of weakness to break the human out of over-exploitation induced by cognitive entrenchment. With mild assumptions, we show that our algorithm converges sub-linearly, at a rate faster than the AI or human alone. We validate our algorithm using synthetic data and with human experts performing real-world experiments. | A Human-AI collaborative optimisation approach using sample-efficient Bayesian optimisation |
Adversarial Training using a strong first-order adversary (PGD) is the gold standard for training Deep Neural Networks that are robust to adversarial examples. We show that, contrary to the general understanding of the method, the gradient at an optimal adversarial example may increase, rather than decrease, the adversarially robust loss. This holds independently of the learning rate. More precisely, we provide a counterexample to a corollary of Danskin's Theorem presented in the seminal paper of Madry et al. (2018) which states that a solution of the inner maximization problem can yield a descent direction for the adversarially robust loss. Based on a correct interpretation of Danskin's Theorem, we propose Danskin's Descent Direction (DDi) and we verify experimentally that it provides better directions than those obtained by a PGD adversary. Using the CIFAR10 dataset we further provide a real world example showing that our method achieves a steeper increase in robustness levels in the early stages of training, and is more stable than the PGD baseline. As a limitation, PGD training of ReLU+BatchNorm networks still performs better, but current theory is unable to explain this.
| There is a subtle bug in the theory behind PGD. We show how to correct it and that it matters in practice |
In this work, we present the Bregman Alternating Projected Gradient (BAPG) method, a single-loop algorithm that offers an approximate solution to the Gromov-Wasserstein (GW) distance.
We introduce a novel relaxation technique that balances accuracy and computational efficiency, albeit with some compromises in the feasibility of the coupling map. Our analysis is based on the observation that the GW problem satisfies the Luo-Tseng error bound condition, which relates to estimating the distance of a point to the critical point set of the GW problem based on the optimality residual.
This observation allows us to provide an approximation bound for the distance between the fixed-point set of BAPG and the critical point set of GW. Moreover, under a mild technical assumption, we can show that BAPG converges to its fixed point set.
The effectiveness of BAPG has been validated through comprehensive numerical experiments in graph alignment and partition tasks, where it outperforms existing methods in terms of both solution quality and wall-clock time. | We propose the first provable single-loop algorithm for computing the Gromov-Wasserstein (GW) distance. |
Several machine learning (ML) applications are characterized by searching for an optimal solution to a complex task. The search space for this optimal solution is often very large, so large in fact that this optimal solution is often not computable. Part of the problem is that many candidate solutions found via ML are actually infeasible and have to be discarded. Restricting the search space to only the feasible solution candidates simplifies finding an optimal solution for the tasks. Further, the set of feasible solutions could be re-used in multiple problems characterized by different tasks. In particular, we observe that complex tasks can be decomposed into subtasks and corresponding skills. We propose to learn a reusable and transferable skill by training an actor to generate all feasible actions. The trained actor can then propose feasible actions, among which an optimal one can be chosen according to a specific task. The actor is trained by interpreting the feasibility of each action as a target distribution. The training procedure minimizes a divergence of the actor's output distribution to this target. We derive the general optimization target for arbitrary f-divergences using a combination of kernel density estimates, resampling, and importance sampling. We further utilize an auxiliary critic to reduce the interactions with the environment. A preliminary comparison to related strategies shows that our approach learns to visit all the modes in the feasible action space, demonstrating the framework's potential for generating multimodal action distributions. | We propose to train a generative neural network to generate all feasible actions within an interactive environment. |
Graph neural networks have achieved great success in representing structured data and its downstream tasks such as node classification. The key idea is to recursively propagate and aggregate information along the edges of a given graph topology. However, edges in real-world graphs often have varying degrees of difficulty, and some edges may even be noisy to the downstream tasks. Therefore, existing graph neural network models may lead to suboptimal learned representations because they usually consider every edge in a given graph topology equally. On the other hand, curriculum learning, which mimics the human learning principle of learning data samples in a meaningful order, has been shown to be effective in improving the generalization ability of representation learners by gradually proceeding from easy to more difficult samples during training. Unfortunately, most existing curriculum learning strategies are designed for i.i.d data samples and cannot be trivially generalized to handle structured data with dependencies. In order to address these issues, in this paper we propose a novel curriculum learning method for structured data to leverage the various underlying difficulties of data dependencies to improve the quality of learned representations on structured data. Specifically, we design a learning strategy that gradually incorporates edges in a given graph topology into training according to their difficulty from easy to hard, where the degree of difficulty is measured by a self-supervised learning paradigm. We demonstrate the strength of our proposed method in improving the generalization ability of learned representations through extensive experiments on nine synthetic datasets and seven real-world datasets with different commonly used graph neural network models as backbone models. | We propose a novel curriculum learning strategy to improve the generalization performance of graph neural network models by gradually involving edges from well-expected to less-expected in training. |
Reinforcement learning (RL) provides a theoretical framework for continuously improving an agent’s behavior via trial and error. However, efficiently learning policies from scratch can be very difficult, particularly for tasks that present exploration challenges. In such settings, it might be desirable to initialize RL with an existing policy, offline data, or demonstrations. However, naively performing such initialization in RL often works poorly, especially for value-based methods. In this paper, we present a meta algorithm that can use offline data, demonstrations, or a pre-existing policy to initialize an RL policy, and is compatible with any RL approach. In particular, we propose Jump-Start Reinforcement Learning (JSRL), an algorithm that employs two policies to solve tasks: a guide-policy, and an exploration-policy. By using the guide-policy to form a curriculum of starting states for the exploration-policy, we are able to efficiently improve performance on a set of simulated robotic tasks. We show via experiments that it is able to significantly outperform existing imitation and reinforcement learning algorithms, particularly in the small-data regime. In addition, we provide an upper bound on the sample complexity of JSRL and show that with the help of a guide-policy, one can improve the sample complexity for non-optimism exploration methods from exponential in horizon to polynomial. | Efficiently initializing reinforcement learning policies using a prior policy. |
Temporal difference (TD) learning is a simple algorithm for policy evaluation
in reinforcement learning. The performance of TD learning is affected by high
variance and it can be naturally enhanced with variance reduction techniques, such
as the Stochastic Variance Reduced Gradient (SVRG) method. Recently, multiple
works have sought to fuse TD learning with SVRG to obtain a policy evaluation
method with a linear rate of convergence. However, the resulting convergence rate
is significantly weaker than what is achieved by SVRG in the setting of convex
optimization. In this work we utilize a recent interpretation of TD-learning as the
splitting of the gradient of an appropriately chosen function, thus simplifying the
algorithm and fusing TD with SVRG. We prove a linear convergence bound that
is identical to the convergence bound available for SVRG in the convex setting. | We prove a linear convergence time for an SVRG-inspired temporal difference method which is identical to the original convergence time bound of SVRG in the convex setting. |
Finding the best way to schedule operations in a computation graph is a classical NP-hard problem which is central to compiler optimization. However, evaluating the goodness of a schedule on the target hardware can be very time-consuming. Traditional approaches as well as previous machine learning ones typically optimize proxy metrics, which are fast to evaluate but can lead to bad schedules when tested on the target hardware. In this work, we propose a new approach to scheduling by sampling proportionally to the proxy metric using a novel GFlowNet method. We introduce a technique to control the trade-off between diversity and goodness of the proposed schedules at inference time and demonstrate empirically that the pure optimization baselines can lead to subpar performance with respect to our approach when tested on a target model. Furthermore, we show that conditioning the GFlowNet on the computation graph enables generalization to unseen scheduling problems for both synthetic and real-world compiler datasets. | We use GFlowNets for robust scheduling. |
This is paper for the smooth function approximation by neural networks (NN). Mathematical or physical functions can be replaced by NN models through regression. In this study, we get NNs that generate highly accurate and highly smooth function, which only comprised of a few weight parameters, through discussing a few topics about regression. First, we reinterpret inside of NNs for regression; consequently, we propose a new activation function--integrated sigmoid linear unit (ISLU). Then special charateristics of metadata for regression, which is different from other data like image or sound, is discussed for improving the performance of neural networks. Finally, the one of a simple hierarchical NN that generate models substituting mathematical function is presented, and the new batch concept ``meta-batch" which improves the performance of NN several times more is introduced. The new activation function, meta-batch method, features of numerical data, meta-augmentation with metaparameters, and a structure of NN generating a compact multi-layer perceptron(MLP) are essential in this study. | Smooth mathematical functions are obtained from neural networks comprised few weight parameters using a new activation function and the new batch method. |
Image copy detection and retrieval from large databases leverage two components. First, a neural network maps an image to a vector representation, that is relatively robust to various transformations of the image. Second, an efficient but approximate similarity search algorithm trades scalability (size and speed) against quality of the search, thereby introducing a source of error.
This paper improves the robustness of image copy detection with active indexing, that optimizes the interplay of these two components. We reduce the quantization loss of a given image representation by making imperceptible changes to the image before its release. The loss is back-propagated through the deep neural network back to the image, under perceptual constraints. These modifications make the image more retrievable.
Our experiments show that the retrieval and copy detection of activated images is significantly improved. For instance, activation improves by $+40\%$ the Recall1@1 on various image transformations, and for several popular indexing structures based on product quantization and locality sensitivity hashing. | In the context of image tracing, instead of watermarking an image with an ID, we slightly modify it to make its representation more indexing-friendly, which makes plain content-based indexing much more robust (62% → 100% accuracy for some settings). |
Hyperparameter optimization is an important subfield of machine learning that focuses on tuning the hyperparameters of a chosen algorithm to achieve peak performance. Recently, there has been a stream of methods that tackle the issue of hyperparameter optimization, however, most of the methods do not exploit the scaling law property of learning curves. In this work, we propose Deep Power Law (DPL), a neural network model conditioned to yield predictions that follow a power-law scaling pattern. Our model dynamically decides which configurations to pause and train incrementally by making use of multi-fidelity estimation. We compare our method against 7 state-of-the-art competitors on 3 benchmarks related to tabular, image, and NLP datasets covering 59 diverse search spaces. Our method achieves the best results across all benchmarks by obtaining the best any-time results compared to all competitors. | Multi-fidelity hyperparameter optimization with deep power laws that achieves state-of-the-art results across diverse benchmarks. |
Pruning neural networks before training has received increasing interest due to its potential to reduce training time and memory. One popular method is to prune the connections based on a certain metric, but it is not entirely clear what metric is the best choice. Recent advances in neural tangent kernel (NTK) theory suggest that the training dynamics of large enough neural networks is closely related to the spectrum of the NTK. Motivated by this finding, we propose to prune the connections that have the least influence on the spectrum of the NTK. This method can help maintain the NTK spectrum, which may help align the training dynamics to that of its dense counterpart. However, one possible issue is that the fixed-weight-NTK corresponding to a given initial point can be very different from the NTK corresponding to later iterates during the training phase. We further propose to sample multiple realizations of random weights to estimate the NTK spectrum. Note that our approach is weight-agnostic, which is different from most existing methods that are weight-dependent. In addition, we use random inputs to compute the fixed-weight-NTK, making our method data-agnostic as well. We name our foresight pruning algorithm Neural Tangent Kernel Spectrum-Aware Pruning (NTK-SAP). Empirically, our method achieves better performance than all baselines on multiple datasets. | We introduce a pruning-at-initialization method by aligning the eigenspectrum of NTK to that of the dense network. |
Guided diffusion is a technique for conditioning the output of a diffusion model at sampling time without retraining the network for each specific task. However, one drawback of diffusion models, whether they are guided or unguided, is their slow sampling process.
Recent techniques can accelerate unguided sampling by applying high-order numerical methods to the sampling process when viewed as differential equations. On the contrary, we discover that the same techniques do not work for guided sampling, and little has been explored about its acceleration. This paper explores the culprit of this problem and provides a solution based on operator splitting methods, motivated by our key finding that classical high-order numerical methods are unsuitable for the conditional function. Our proposed method can re-utilize the high-order methods for guided sampling and can generate images with the same quality as a 250-step DDIM baseline using 32-58% less sampling time on ImageNet256.
We also demonstrate usage on a wide variety of conditional generation tasks, such as text-to-image generation, colorization, inpainting, and super-resolution. | We accelerate guided diffusion sampling using splitting numerical methods. |
Graph neural networks (GNNs) continue to achieve state-of-the-art performance on many graph learning tasks, but rely on the assumption that a given graph is a sufficient approximation of the true neighborhood structure. In the presence of higher-order sequential dependencies, we show that the tendency of traditional graph representations to underfit each node's neighborhood causes existing GNNs to generalize poorly. To address this, we propose a novel Deep Graph Ensemble (DGE), which captures neighborhood variance by training an ensemble of GNNs on different neighborhood subspaces of the same node within a higher-order network structure. We show that DGE consistently outperforms existing GNNs on semisupervised and supervised tasks on six real-world data sets with known higher-order dependencies, even under a similar parameter budget. We demonstrate that learning diverse and accurate base classifiers is central to DGE's success, and discuss the implications of these findings for future work on GNNs. | We propose an ensemble of GNNs that exploits variance in the neighborhood subspaces of nodes in graphs with higher-order dependencies and consistently outperforms baselines on semisupervised and supervised learning tasks. |
From the server's perspective, federated learning (FL) learns a global model by iteratively sampling a cohort of clients and updating the global model with the sum local gradient of the cohort. We find this process is analogous to mini-batch SGD of centralized training. In mini-batch SGD, a model is learned by iteratively sampling a batch of data and updating the model with the sum gradient of the batch. In this paper, we delve into the training dynamics in FL by learning from the experience of optimization and generalization in mini-batch SGD. Specifically, we focus on two aspects: \emph{client coherence} (refers to sample coherence in mini-batch SGD) and \emph{global weight shrinking regularization} (refers to weight decay in mini-batch SGD). We find the roles of the two aspects are both determined by the aggregation weights assigned to each client during global model updating. Thus, we use aggregation weight optimization on the server as a tool to study how client heterogeneity and the number of local epochs affect the global training dynamics in FL. Besides, we propose an effective method for \textbf{Fed}erated \textbf{A}ggregation \textbf{W}eight \textbf{O}ptimization, named as \textsc{\textbf{FedAWO}}. Extensive experiments verify that our method can improve the generalization of the global model by a large margin on different datasets and models. | We provide new understandings about the training dynamics of federated learning with neural network and devise a practical tool for aggregation weight optimization, improving global model generalization. |
End-to-end learning for visual robotic manipulation is known to suffer from sample inefficiency, requiring large numbers of demonstrations. The spatial roto-translation equivariance, or the SE(3)-equivariance can be exploited to improve the sample efficiency for learning robotic manipulation. In this paper, we present SE(3)-equivariant models for visual robotic manipulation from point clouds that can be trained fully end-to-end. By utilizing the representation theory of the Lie group, we construct novel SE(3)-equivariant energy-based models that allow highly sample efficient end-to-end learning. We show that our models can learn from scratch without prior knowledge and yet are highly sample efficient (5~10 demonstrations are enough). Furthermore, we show that our models can generalize to tasks with (i) previously unseen target object poses, (ii) previously unseen target object instances of the category, and (iii) previously unseen visual distractors. We experiment with 6-DoF robotic manipulation tasks to validate our models' sample efficiency and generalizability. Codes are available at: https://github.com/tomato1mule/edf | We present SE(3)-equivariant energy based models that can learn robotic manipulation tasks end-to-end from only few demonstrations without any prior knowledge. |
Event-based cameras exhibit high dynamic range and temporal precision that could make them ideal for detecting objects with high speeds and low relative luminance. These properties have made event-based cameras especially interesting for use in space domain awareness tasks, such as detecting dim, artificial satellites with high brightness backgrounds using ground-based optical sensors; however, the asynchronous nature of event-based data presents new challenges to performing objection detection. While spiking neural networks (SNNs) have been shown to naturally complement the asynchronous and binary properties of event-based data, they also present a number of challenges in their training, such as the spike vanishing problem and the large number of timesteps required for maximizing classification and detection accuracy. Furthermore, the extremely high sampling rate of event-based sensors and the density of noisy space-based data collections can results in excessively large event streams within a short window of recording. We present a temporally-weighted spike encoding that greatly reduces the number of spikes derived from an event-based data stream, enabling the training of larger SNNs with fewer timesteps for maximal accuracy. We propose using this spike encoding with a variant of convolutional SNN trained utilizing surrogate spiking neuron gradients with backpropagation-through-time (BPTT) for both classification and object detection tasks with an emphasis on space-domain awareness. To demonstrate the efficacy of our encoding and SNN approach, we present competitive classification accuracies on benchmark datasets N-MNIST (99.7%), DVS-CIFAR10 (74.0%), and N-Caltech101 (72.8%), as well as state-of-the-art object detection performance on event-based, satellite collections. | Performing spiking neural network-based classification and object detection using a new spike encoding method for event-based vision sensors. |
Co-design of morphology and control of a Voxel-based Soft Robot (VSR) is challenging due to the notorious bi-level optimization. In this paper, we present a Curriculum-based Co-design (CuCo) method for learning to design and control VSRs through an easy-to-difficult process. Specifically, we expand the design space from a small size to the target size gradually through a predefined curriculum. At each learning stage of the curriculum, we use reinforcement learning to simultaneously train the design policy and the control policy, which is enabled by incorporating the design process into the environment and using differentiable policy representations. The converged morphology and the learned policies from last stage are inherited and then serve as the starting point for the next stage. In empirical studies, we show that CuCo is more efficient in creating larger robots with better performance by reusing the practical design and control patterns learned within each stage, in comparison to prior approaches that learn from scratch in the space of target size. | Curriculum-based Co-design of Morphology and Control of Voxel-based Soft Robots |
Many well-known Graph Neural Network (GNN) models assume the underlying graphs are homophilous, where nodes share similar features and labels with their neighbours. They rely on message passing that iteratively aggregates neighbour's features and often suffer performance degradation on non-homophilous graphs where useful information is hardly available in the local neighbourhood. In addition, earlier studies show that in some cases GNNs are even outperformed by Multi-Layer Perceptron, indicating insufficient exploitation of node feature information. Motivated by the two limitations, we propose ProtoGNN, a novel message passing framework that augments existing GNNs by effectively combining node features with structural information. ProtoGNN learns multiple class prototypes for each class from raw node features with the slot-attention mechanism. These prototype representations are then transferred onto the structural node features with explicit message passing to all non-training nodes irrespective of distance. This form of message passing, from training nodes to class prototypes to non-training nodes, also serves as a shortcut that bypasses local graph neighbourhoods and captures global information. ProtoGNN is a generic framework which can be applied onto any of the existing GNN backbones to improve node representations when node features are strong and local graph information is scarce. We demonstrate through extensive experiments that ProtoGNN brings performance improvement to various GNN backbones and achieves state-of-the-art on several non-homophilous datasets. | Class prototype-assisted message passing framework for improving node representation learning on non-homophilous graphs |
Following the investigation that protein sequence determines its structure and function, engineering protein sequences allows us to optimize the functions of proteins for specific purposes such as enhancement of catalytic activity or binding affinity maturation. In protein engineering, there are many cases where the amino acids in the middle of a protein sequence are changed while maintaining the remaining residues to avoid unwanted functional changes from remaining residues. However, existing research on protein sequence design via protein language models (PLMs) has focused on modifying suffix residues by prompting prefix residues to the model or mutating the overall sequence residues. This is unsuitable for scenarios where the residues located in the middle of the sequence are to be optimized. In this work, we suggest a PLM-based framework to solve the fill-in-middle (FIM) protein engineering tasks. To evaluate the performance of PLMs on the FIM tasks, we design a novel evaluation scheme where PLMs are tasked to generate new sequences while maintaining the secondary structures. Also, we propose a new PROTein language model specialized for the Fill-In-Middle task, ProtFIM. Experiments confirm that ProtFIM performs FIM engineering efficiently, especially for alpha-helix structures, and provides decent protein representations of sequence-function relationships. Finally, we demonstrate an artificial protein sequence design framework composed of ProtFIM and a high-quality structure predictor as a novel tool to optimize protein sequences. | We propose a new evaluation scheme and protein language model for fill-in-middle protein sequence design. |
Federated learning (FL) is an emerging distributed machine learning method that empowers in-situ model training on decentralized edge devices. However, multiple simultaneous training activities could overload resource-constrained devices. In this work, we propose a smart multi-tenant FL system, MuFL, to effectively coordinate and execute simultaneous training activities. We first formalize the problem of multi-tenant FL, define multi-tenant FL scenarios, and introduce a vanilla multi-tenant FL system that trains activities sequentially to form baselines. Then, we propose two approaches to optimize multi-tenant FL: 1) activity consolidation merges training activities into one activity with a multi-task architecture; 2) after training it for rounds, activity splitting divides it into groups by employing affinities among activities such that activities within a group have better synergy. Extensive experiments demonstrate that MuFL outperforms other methods while consuming 40% less energy. We hope this work will inspire the community to further study and optimize multi-tenant FL. | We propose a smart multi-tenant federated learning system, MuFL, to efficiently coordinate and execute simultaneous training activities under resource constraints by considering both synergies and differences among training activities. |
Traditional Dynamic Programming (DP) approaches suffer from slow backward credit-assignment (CA): only a one-step search is performed at each update. A popular solution for multi-step CA is to use multi-step Bellman operators. Unfortunately, in the control settings, existing methods typically suffer from the large variance of multi-step off-policy corrections or are biased, preventing convergence. To overcome these problems, we introduce a novel multi-step Bellman optimality equation with adaptive lookahead steps. We first derive a new multi-step Value Iteration (VI) method that converges to the optimal Value Function (VF) with an exponential contraction rate but linear computational complexity. Given some trial, our so-called Highway RL performs rapid CA, by picking a policy and a possible lookahead (up to the trial end) that maximize the near-term reward during lookahead plus a DP-based estimate of the cumulative reward for the remaining part of the trial. Highway RL does not require off-policy corrections. Under mild assumptions, it achieves better convergence rates than the traditional one-step Bellman Optimality Operator. We then derive Highway Q-Learning, a convergent multi-step off-policy variant of Q-learning. We show that our Highway algorithms significantly outperform DP approaches on toy tasks. Finally, we propose a deep function approximation variant called Highway DQN. We evaluate it on visual MinAtar Games, outperforming similar multi-step methods. | a novel adaptive multi-step Bellman Optimality Equation for efficient credit assignment that converges to the optimal value function with better contraction rate under mild assumptions |
Large pre-trained models exhibit distinct and complementary capabilities dependent on the data they are trained on. Language models such as GPT-3 are capable of textual reasoning but cannot understand visual information, while vision models such as DALL-E can generate photorealistic photos but fail to understand complex language descriptions. In this work, we propose a unified framework for composing ensembles of different pre-trained models -- combining the strengths of each individual model to solve various multimodal problems in a zero-shot manner. We use pre-trained models as "generators" or "scorers" and compose them via closed-loop iterative consensus optimization. The generator constructs proposals and the scorers iteratively provide feedback to refine the generated result. Such closed-loop communication enables models to correct errors caused by other models, significantly boosting performance on downstream tasks, e.g. improving accuracy on grade school math problems by 7.5%, without requiring any model finetuning. We demonstrate that consensus achieved by an ensemble of scorers outperforms the feedback of a single scorer, by leveraging the strengths of each expert model. Results show that the proposed method can be used as a general purpose framework for a wide range of zero-shot multimodal tasks, such as image generation, video question answering, mathematical reasoning, and robotic manipulation.
| We propose a unified framework for composing pre-trained models for a variety of zero-shot multimodal tasks through iterative consensus. |
We study the canonical statistical estimation problem of linear regression from $n$ i.i.d. examples under $(\varepsilon,\delta)$-differential privacy when a fraction of response variables are adversarially corrupted. We propose a variant of the popular differentially private stochastic gradient descent (DP-SGD) algorithm with two innovations: a full-batch gradient descent to improve sample complexity and a novel adaptive clipping to guarantee robustness. When there is no adversarial corruption, this algorithm improves upon the existing state-of-the-art approach and achieves near optimal sample complexity. Under label-corruption, this is the first efficient linear regression algorithm to provably guarantee both $(\epsilon,\delta)$-DP and robustness. Synthetic experiments confirm the superiority of our approach. | We provide a private gradient descent with adaptive clipping that achieves near optimal error rate and robustness against label noise. |
Spiking Neural Network (SNN) is widely regarded as one of the next-generation neural network infrastructures, yet it suffers from an inherent non-differentiable problem that makes the traditional backpropagation (BP) method infeasible. Surrogate gradients (SG), which are an approximation to the shape of the Dirac's $\delta$-function, can help alleviate this issue to some extent. To our knowledge, the majority of research, however, keep a fixed surrogate gradient for all layers, ignorant of the fact that there exists a trade-off between the approximation to the delta function and the effective domain of gradients under the given dataset, hence limiting the efficiency of surrogate gradients and impairing the overall model performance. To guide the shape optimization in applying surrogate gradients for training SNN, we propose an indicator $\chi$, which represents the proportion of parameters with non-zero gradients in backpropagation. Further we present a novel $\chi$-based training pipeline that adaptively makes trade-offs between the surrogate gradients' shapes and its effective domain, followed by a series of ablation experiments for verification. Our algorithm achieves 69.09\% accuracy on the ImageNet dataset using SEW-ResNet34 - a 2.05\% absolute improvement from baseline. Moreover, our method only requires extremely low external cost and can be simply integrated into the existing training procedure. | We propose a method to change the shape of surrogate gradients, which can improve the performance of spiking neural networks with low extra overhead. |
The conventional few-shot classification aims at learning a model on a large labeled base dataset and rapidly adapting to a target dataset that is from the same distribution as the base dataset. However, in practice, the base and the target datasets of few-shot classification are usually from different domains, which is the problem of cross-domain few-shot classification. We tackle this problem by making a small proportion of unlabeled images in the target domain accessible in the training stage. In this setup, even though the base data are sufficient and labeled, the large domain shift still makes transferring the knowledge from the base dataset difficult. We meticulously design a cross-level knowledge distillation method, which can strengthen the ability of the model to extract more discriminative features in the target dataset by guiding the network's shallow layers to learn higher-level information. Furthermore, in order to alleviate the overfitting in the evaluation stage, we propose a feature denoising operation which can reduce the feature redundancy and mitigate overfitting. Our approach can surpass the previous state-of-the-art method, Dynamic-Distillation, by 5.44% on 1-shot and 1.37% on 5-shot classification tasks on average in the BSCD-FSL benchmark. The implementation code will be available at https://gitee.com/mindspore/models/tree/master/research/cv/CLDFD. | We design a cross-level distillation and a feature denoising operation for handling cross-domain few-shot classification. Our approach can surpass the SOTA method by 5.44% on 1-shot and 1.37% on 5-shot classification tasks in the BSCD-FSL benchmark. |
Existing Visual Learning (VL) benchmarks often contain exploitative biases. Most former works only attempted to mitigate biases in semantically low-level and conventional visual-question-answering typed datasets like VQA and GQA. However, these methods cannot generalize to recently emerging highly semantic VL datasets like VCR and are also difficult to scale due to many severe problems like high-cost labors, drastically disrupting the data distribution\textit{, etc.}To resolve those problems and also address other unique biases on VCR-like datasets, we first conduct in-depth analysis and identify important biases in VCR dataset. We further propose a generalized solution that synthesizes counterfactual image and text data based on the original query's semantic focus while producing less distortion to the data distribution. To utilize our synthesized data, we also design an innovative intra-sample contrastive training strategy to assist QA learning in Visual Commonsense Reasoning (VCR). Moreover, our synthesized VL data also serve as a highly-semantic debiased benchmark for evaluating future VL models' robustness. Extensive experiments show that our proposed synthesized data and training strategy improve existing VL models' performances on both the original VCR dataset and our proposed debiased benchmark. | Counterfactual Vision-Language Data Synthesis with Intra-Sample Contrast Learning for Visual Commonsense Reasoning |
Although much of the success of Deep Learning builds on learning good representations, a rigorous method to evaluate their quality is lacking. In this paper, we treat the evaluation of representations as a model selection problem and propose to use the Minimum Description Length (MDL) principle to devise an evaluation metric. Contrary to the established practice of limiting the capacity of the readout model, we design a hybrid discrete and continuous-valued model space for the readout models and employ a switching strategy to combine their predictions. The MDL score takes model complexity, as well as data efficiency into account. As a result, the most appropriate model for the specific task and representation will be chosen, making it a unified measure for comparison. The proposed metric can be efficiently computed with an online method and we present results for pre-trained vision encoders of various architectures (ResNet and ViT) and objective functions (supervised and self-supervised) on a range of downstream tasks. We compare our methods with accuracy-based approaches and show that the latter are inconsistent when multiple readout models are used. Finally, we discuss important properties revealed by our evaluations such as model scaling, preferred readout model, and data efficiency. | We propose an evaluation framework that is based on MDL and model switching for evaluating representations. |
As more and more conversational and translation systems are deployed in production, it is essential to implement and develop effective control mechanisms to ensure their proper functioning and security. An essential component to ensure the safe behavior of the system is out-of-distribution (OOD) detection, which aims to detect whether an input sample is statistically far from the training distribution. While OOD detection is a widely covered topic in classification tasks, it has received much less attention in text generation. This paper addresses the problem of OOD detection for machine translation and dialog generation from an operational perspective. Our contribution includes (i) RAINPROOF a Relative informAItioN Projection Out OF distribution detection framework and (ii) a more operational evaluation setting for OOD detection. Surprisingly, we find that OOD detection is not necessarily aligned with task-specific measures. The OOD detector may filter out samples that are well processed by the model and keep samples that are not, leading to weaker performance. Our results show that RAINPROOF breaks this curse and achieve good results in OOD detection while increasing system performance. | Out of distribution detection for natural language generation |
We study dynamic algorithms robust to adaptive input generated from sources with bounded capabilities, such as sparsity or limited interaction. For example, we consider robust linear algebraic algorithms when the updates to the input are sparse but given by an adversary with access to a query oracle. We also study robust algorithms in the standard centralized setting, where an adversary queries an algorithm in an adaptive manner, but the number of interactions between the adversary and the algorithm is bounded. We first recall a unified framework of (Hassidim et al., 2020; Beimel et al., 2022; Attias et al., 2023) for answering $Q$ adaptive queries that incurs $\widetilde{\mathcal{O}}(\sqrt{Q})$ overhead in space, which is roughly a quadratic improvement over the na\"{i}ve implementation, and only incurs a logarithmic overhead in query time. Although the general framework has diverse applications in machine learning and data science, such as adaptive distance estimation, kernel density estimation, linear regression, range queries, point queries, and serves as a preliminary benchmark, we demonstrate even better algorithmic improvements for (1) reducing the pre-processing time for adaptive distance estimation and (2) permitting an unlimited number of adaptive queries for kernel density estimation. Finally, we complement our theoretical results with additional empirical evaluations. | We give algorithms robust to adaptive input from adversaries with bounded capabilities and a general framework for achieving it. |
Optimiser design influences learning speed and generalisation in training machine learning models. Several studies have attempted to learn more effective gradient-descent optimisers via solving a bi-level optimisation problem where generalisation error is minimised with respect to optimiser parameters. However, most existing neural network oriented optimiser learning methods are intuitively motivated, without clear theoretical support, and focus on learning implicit biases that improve generalisation, rather than speed of convergence. We take a different perspective starting from mirror descent rather than gradient descent, and meta-learning the corresponding Bregman divergence. Within this paradigm, we formalise a novel meta-learning objective of optimising the rate of convergence. The resulting framework, termed Meta Mirror Descent (MetaMD), learns to accelerate optimisation speed. Unlike many meta-learned neural network optimisers, it also supports convergence guarantees and uniquely does so without requiring validation data. We empirically evaluate our framework on a variety of tasks and architectures in terms of convergence rate and generalisation error and demonstrate strong performance. | We proposed a meta-learning based algorithm, learning optimisers under the mirror descent framework. |
Tensor decompositions have been successfully applied to compress neural networks. The compression algorithms using tensor decompositions commonly minimize the approximation error on the weights. Recent work assumes the approximation error on the weights is a proxy for the performance of the model to compress multiple layers and fine-tune the compressed model. Surprisingly, little research has systematically evaluated which approximation errors can be used to make choices regarding the layer, tensor decomposition method, and level of compression. To close this gap, we perform an experimental study to test if this assumption holds across different layers and types of decompositions, and what the effect of fine-tuning is. We include the approximation error on the features resulting from a compressed layer in our analysis to test if this provides a better proxy, as it explicitly takes the data into account. We find the approximation error on the weights has a positive correlation with the performance error, before as well as after fine-tuning. Basing the approximation error on the features does not improve the correlation significantly. While scaling the approximation error commonly is used to account for the different sizes of layers, the average correlation across layers is smaller than across all choices (i.e. layers, decompositions, and level of compression) before fine-tuning. When calculating the correlation across the different decompositions, the average rank correlation is larger than across all choices. This means multiple decompositions can be considered for compression and the approximation error can be used to choose between them. | We show empirically an approximation error resulting from compressing a network layer with tensor decomposition is correlated with the classification error, enabling the choice of layer, decomposition and rank to be based on the approximation error. |
We consider a novel lossy compression approach based on unconditional diffusion generative models, which we call DiffC. Unlike modern compression schemes which rely on transform coding and quantization to restrict the transmitted information, DiffC relies on the efficient communication of pixels corrupted by Gaussian noise. We implement a proof of concept and find that it works surprisingly well despite the lack of an encoder transform, outperforming the state-of-the-art generative compression method HiFiC on ImageNet 64x64. DiffC only uses a single model to encode and denoise corrupted pixels at arbitrary bitrates. The approach further provides support for progressive coding, that is, decoding from partial bit streams. We perform a rate-distortion analysis to gain a deeper understanding of its performance, providing analytical results for multivariate Gaussian data as well as theoretic bounds for general distributions. Furthermore, we prove that a flow-based reconstruction achieves a 3 dB gain over ancestral sampling at high bitrates. | Theoretical and empirical results on a novel lossy compression approach using diffusion models |
In the literature on game-theoretic equilibrium finding, focus has mainly been on solving a single game in isolation. In practice, however, strategic interactions—ranging from routing problems to online advertising auctions—evolve dynamically, thereby leading to many similar games to be solved. To address this gap, we introduce meta-learning for equilibrium finding and learning to play games. We establish the first meta-learning guarantees for a variety of fundamental and well-studied games, including two-player zero-sum games, general-sum games, Stackelberg games, and multiple extensions thereof. In particular, we obtain rates of convergence to different game-theoretic equilibria that depend on natural notions of similarity between the sequence of games encountered, while at the same time recovering the known single-game guarantees when the sequence of games is arbitrary. Along the way, we prove a number of new results in the single-game regime through a simple and unified framework, which may be of independent interest. Finally, we evaluate our meta-learning algorithms on endgames faced by the poker agent Libratus against top human professionals. The experiments show that games with varying stack sizes can be solved significantly faster using our meta-learning techniques than by solving them separately, often by an order of magnitude. | We formalize and study the problem of meta-learning across a wide range of fundamental multi-agent settings. |
Existing collaborative self-supervised learning (SSL) schemes are not suitable for cross-client applications because of their expensive computation and large local data requirements. To address these issues, we propose MocoSFL, a collaborative SSL framework based on Split Federated Learning (SFL) and Momentum Contrast (MoCo). In MocoSFL, the large backbone model is split into a small client-side model and a large server-side model, and only the small client-side model is processed locally on the client's local devices. MocoSFL has three key components: (i) vector concatenation which enables the use of small batch size and reduces computation and memory requirements by orders of magnitude; (ii) feature sharing that helps achieve high accuracy regardless of the quality and volume of local data; (iii) frequent synchronization that helps achieve better non-IID performance because of smaller local model divergence. For a 1,000-client case with non-IID data (each client only has data from 2 random classes of CIFAR-10), MocoSFL can achieve over 84% accuracy with ResNet-18 model. Next we present TAResSFL module that significantly improves the resistance to privacy threats and communication overhead with small sacrifice in accuracy for a MocoSFL system. On a Raspberry Pi 4B device, the MocoSFL-based scheme requires less than 1MB of memory and less than 40MB of communication, and consumes less than 5W power. The code is available at https://github.com/SonyAI/MocoSFL. | Existing collaborative SSL schemes are not suitable for cross-client applications because of their expensive computation and local data requirements. To address these issues, we propose MocoSFL based on Split Federated Learning and MoCo. |
While active learning (AL) improves the labeling efficiency of machine learning (by allowing models to query the labels of data samples), a major problem is that compute efficiency is decreased since models are typically retrained from scratch at each query round. In this work, we develop a new framework that circumvents this problem by biasing further training towards the recently labeled sets, thereby complementing existing work on AL acceleration. We employ existing and novel replay-based Continual Learning (CL) algorithms that are effective at quickly learning new samples without forgetting previously learned information, especially when data comes from a shifting or evolving distribution. We call this compute-efficient active learning paradigm $\textit{``Continual Active Learning" (CAL)}$. We demonstrate that standard AL with warm starting fails, both to accelerate training, and that naive fine-tuning suffers from catastrophic forgetting due to distribution shifts over query rounds. We then show CAL achieves significant speedups using a plethora of replay schemes that use model distillation, and that select diverse/uncertain points from the history, all while maintaining performance on par with standard AL. We conduct experiments across many data domains, including natural language, vision, medical imaging, and computational biology, each with very different neural architectures (Transformers/CNNs/MLPs). CAL consistently provides a 2-6x reduction in training time, thus showing its applicability across differing modalities. | We reduce Active Learning (AL) training time with the help of replay based Continual Learning algorithms all while maintaining performance on par with standard AL. |
This work presents N-Student Learning, a pseudo-label based multi-network training setup that can be applied to nearly any supervised learning architecture in order to help combat the problem of overfitting and control the way in which a network models uncertainty in the data. The effectiveness of N-Student Learning relies on the idea that a network's predictions on unseen data are largely independent of any instance-dependent noise in the labels. In N-Student Learning, each student network is assigned a subset of the training dataset such that no data point is in every student's training subset. Unbiased pseudo-labels can thus be generated for every data point in the training set by taking the predictions of appropriate student networks. Training on these unbiased pseudo-labels minimizes the extent to which each network overfits to instance-dependent noise in the data. Furthermore, based on prior knowledge of the domain, we can control how the networks learn to model uncertainty that is present in the dataset by adjusting the way that pseudo-labels are generated. While this method is largely inspired by the general problem of overfitting, a natural application is found in the problem of classification with noisy labels — a domain where overfitting is a significant concern. After developing intuition through a toy classification task, we proceed to demonstrate that N-Student Learning performs favorably on benchmark datasets when compared to state-of-the-art methods in the problem of classification with noisy labels. | A pseudo-label based multi-network training setup to help combat the problem of overfitting. |
A popular approach to streaming speech translation is to employ a single offline model together with a \textit{wait-$k$} policy to support different latency requirements. It is a simpler alternative compared to training multiple online models with different latency constraints. However, there is an apparent mismatch in using a model trained with complete utterances on partial streaming speech during online inference. We demonstrate that there is a significant difference between the speech representations extracted at the end of a streaming input and their counterparts at the same positions when the complete utterance is available. Built upon our observation that this problem can be alleviated by introducing a few frames of future speech signals, we propose \textbf{F}uture-aware \textbf{in}ferenc\textbf{e} (FINE) for streaming speech translation with two different methods to make the model aware of the future. The first method FINE-Mask incorporates future context through a trainable masked speech model. The second method FINE-Wait simply waits for more actual future audio frames at the cost of extra latency. Experiments on the MuST-C EnDe, EnEs and EnFr benchmarks show that both methods are effective and can achieve better trade-offs between translation quality and latency than strong baselines, and a hybrid approach combining the two can achieve further improvement. Extensive analyses suggest that our methods can effectively alleviate the aforementioned mismatch problem between offline training and online inference. | Future-aware inference for streaming speech translation |
Companies have faced increasing pressure in recent years to anonymize user collected data when sharing internally or to third parties. Text data in particular contains copious amounts of personally identifiable information that has proven to be difficult to de-identify while remain useful for the party of interest. Previous works have suggested that synthetic text generation could provide a promising avenue to curate high performant and private datasets. In this paper, we introduce an approach to synthesize high utility text classification datasets by performing conditional generation through a large language model, distilGPT2, while providing measurable guarantees via differential privacy. We show that naive approaches suffer heavily from utility loss by entangling task-relevant factors in the transformer embedding space, making controlled generation more difficult. We analyze how incorporating a secondary learning objective can improve the performance of the generative model, improving utility of the generated data. | synthesis of private text classification datasets via conditional text generation through GPT-2 fine-tuned with DP-SGD |
Federated learning (FL) is an emerging distributed learning framework that collaboratively trains a shared model without transferring the local clients' data to a centralized server. Motivated by concerns stemming from extended communication and potential attacks, one-shot FL limits communication to a single round while attempting to retain performance. However, one-shot FL methods often degrade under high statistical heterogeneity, fail to promote pipeline security, or require an auxiliary public dataset. To address these limitations, we propose two novel data-free one-shot FL methods: FedCVAE-Ens and its extension FedCVAE-KD. Both approaches reframe the local learning task using a conditional variational autoencoder (CVAE) to address high statistical heterogeneity. Furthermore, FedCVAE-KD leverages knowledge distillation to compress the ensemble of client decoders into a single decoder. We propose a method that shifts the center of the CVAE prior distribution and experimentally demonstrate that this promotes security, and show how either method can incorporate heterogeneous local models. We confirm the efficacy of the proposed methods over baselines under high statistical heterogeneity using multiple benchmark datasets. In particular, at the highest levels of statistical heterogeneity, both FedCVAE-Ens and FedCVAE-KD typically more than double the accuracy of the baselines. | We vastly improve on one-shot federated learning performance under very high statistical heterogeneity by reframing the local learning task with a conditional variational autoencoder. |
In statistical learning theory, generalization bounds usually involve a complexity measure that is constrained by the considered theoretical framework. This limits the scope of such analysis, as in practical algorithms, other forms of regularization are used. Indeed, the empirical work of Jiang et al. (2019) shows that (I) common complexity measures (such as the VC-dimension) do not correlate with the generalization gap and that (ii) there exist arbitrary complexity measures that are better correlated with the generalization gap, but come without generalization guarantees. In this paper, we bridge the gap between this line of empirical works and generalization bounds of statistical learning theory. To do so, we leverage the framework of disintegrated PAC-Bayes bounds to derive a generalization bound that involves an arbitrary complexity measure. Our bound stands in probability jointly over the hypotheses and the learning sample, which allows us to improve the correlation between generalization gap and complexity, as the latter can be set to fit both the hypothesis class and the task. | We provide novel probabilistic generalization bounds able to integrate arbitrary complexity measures be leveraging the framework of disintegrated PAC-Bayes bounds |
Video representation learning has been successful in video-text pre-training for zero-shot transfer, where each sentence is trained to be close to the paired video clips in a common feature space. For long videos, given a paragraph of description where the sentences describe different segments of the video, by matching all sentence-clip pairs, the paragraph and the full video are aligned implicitly. However, such unit-level similarity measure may ignore the global temporal context over a long time span, which inevitably limits the generalization ability. In this paper, we propose a contrastive learning framework TempCLR to compare the full video and the paragraph explicitly. As the video/paragraph is formulated as a sequence of clips/sentences, under the constraint of their temporal order, we use dynamic time warping to compute the minimum cumulative cost over sentence-clip pairs as the sequence-level distance. To explore the temporal dynamics, we break the consistency of temporal order by shuffling the video clips or sentences according to the temporal granularity. In this way, we obtain the representations for clips/sentences, which perceive the temporal information and thus facilitate the sequence alignment. In addition to pre-training on the video and paragraph, our approach can also generalize on the matching between different video instances. We evaluate our approach on video retrieval, action step localization, and few-shot action recognition, and achieve consistent performance gain over all three tasks. Detailed ablation studies are provided to justify the approach design. | Global sequence matching under temporal order consistency matters in contrastive-based video-paragraph/text learning. |
Accurate delineation of fine-scale structures is a very important yet challenging problem. Existing methods use topological information as an additional training loss, but are ultimately making pixel-wise predictions. In this paper, we propose a novel deep learning based method to learn topological/structural. We use discrete Morse theory and persistent homology to construct a one-parameter family of structures as the topological/structural representation space. Furthermore, we learn a probabilistic model that can perform inference tasks in such a topological/structural representation space. Our method generates true structures rather than pixel-maps, leading to better topological integrity in automatic segmentation tasks. It also facilitates semi-automatic interactive annotation/proofreading via the sampling of structures and structure-aware uncertainty. | We use discrete Morse theory and persistent homology to construct an one-parameter family of structures as the topological/structural representation space to perform inference tasks. |
Despite the success of automated machine learning (AutoML), which aims to find the best design, including the architecture of deep networks and hyper-parameters, conventional AutoML methods are computationally expensive and hardly provide insights into the relations of different model design choices. To tackle the challenges, we propose FALCON, an efficient sample-based method to search for the optimal model design. Our key insight is to model the design space of possible model designs as a design graph, where the nodes represent design choices, and the edges denote design similarities. FALCON features 1) a task-agnostic module, which performs message passing on the design graph via a Graph Neural Network (GNN), and 2) a task-specific module, which conducts label propagation of the known model performance information on the design graph. Both modules are combined to predict the design performances in the design space, navigating the search direction. We conduct extensive experiments on 27 node and graph classification tasks from various application domains, and an image classification task on the CIFAR-10 dataset. We empirically show that FALCON can efficiently obtain the well-performing designs for each task using only 30 explored nodes. Specifically, FALCON has a comparable time cost with the one-shot approaches while achieving an average improvement of 3.3% compared with the best baselines. | We propose FALCON, an efficient AutoML method that searches for the optimal model design on design graphs. |
Scaled dot-product attention has become the essence of state-of-the-art deep neural networks for various machine learning tasks. Though its ubiquitous accomplishments, it is inefficient for long sequence tasks and problematic for tasks requiring memory states such as compositional generalization. We propose a novel perspective of the attention mechanism by reinventing it as a memory architecture for neural networks, namely Neural Attention Memory (NAM). NAM follows the same query-key-value structure by constructing a memory matrix while reducing its computational complexity from quadratic to linear to the sequence length. NAM writes a memory matrix via the sum of outer products of value and unit key vectors, and reads it by multiplying the matrix with a unit query vector. Indeed, we show that our normalized outer-product attention mechanism is mathematically equivalent to the conventional attention mechanism. Then, we evaluate a NAM-based Transformer on long-range arena tasks and demonstrate its efficiency and efficacy. Finally, we propose two NAM-based memory-augmented neural networks, namely Long Short-Term Attention Memory (LSAM) and NAM Turing Machine (NAM-TM), and test their compositional generalization capability using four different tasks. LSAM replaces LSTM's long-term cell state with NAM memory matrix and NAM-TM implements a Turing tape data structure using NAM read/write primitives. The experimental results show that the proposed models outperform traditional Transformer and LSTM, as well as DNC. NAM opens up possibilities in diverse machine learning research problems, including hierarchical data modeling, efficient edge inference, and few-shot learning. | Neural attention memory is a differentiable NN memory architecture based on attention which is efficient and powerful. |
We present 3DiM (pronounced "three-dim"), a diffusion model for 3D novel view synthesis from as few as a single image. The core of 3DiM is an image-to-image diffusion model -- 3DiM takes a single reference view and their poses as inputs, and generates a novel view via diffusion. 3DiM can then generate a full 3D consistent scene following our novel stochastic conditioning sampler: the output frames of the scene are generated autoregressively, and during the reverse diffusion process of each individual frame, we select a random conditioning frame from the set of previous frames at each denoising step. We demonstrate that stochastic conditioning yields much more 3D consistent results compared to the naive sampling process which only conditions on a single previous frame. We compare 3DiMs to prior work on the SRN ShapeNet dataset, demonstrating that 3DiM's generated videos from a single view achieve much higher fidelity while being approximately 3D consistent. We also introduce a new evaluation methodology, 3D consistency scoring, to measure the 3D consistency of a generated object by training a neural field on the model's output views. 3DiMs are geometry free, do not rely on hyper-networks or test-time optimization for novel view synthesis, and allow a single model to easily scale to a large number of scenes. | Novel View Synthesis with diffusion models from as few a single image |
$1$-parameter persistent homology, a cornerstone in Topological Data Analysis (TDA), studies the evolution of topological features such as cycle basis hidden in data. It has found its application in strengthening the representation power of deep learning models like Graph Neural Networks (GNN). To enrich the representations of topological features, here we propose to study $2$-parameter persistence modules induced by bi-filtration functions. In order to incorporate these representations into machine learning models, we introduce a novel vectorization on $2$-parameter persistence modules called Generalized Rank Invariant Landscape {\textsc{Gril}}. We show that this vector representation is stable and differentiable with respect to underlying filtration functions and can be easily integrated into machine learning models to augment encoding topological features. We present an algorithm to compute the vectorization and its gradients. We also test our methods on synthetic graph datasets and compare the results with some popular graph neural networks. | A differentiable topological layer based on a novel vector representation on $2$-parameter persistence modules. |
Random masks define surprisingly effective sparse neural network models, as has been shown empirically. The resulting Erd\"os-R\'enyi (ER) random graphs can often compete with dense architectures and state-of-the-art lottery ticket pruning algorithms struggle to outperform them, even though the random baselines do not rely on computationally expensive pruning-training iterations but can be drawn initially without significant computational overhead. We offer a theoretical explanation of how such ER masks can approximate arbitrary target networks if they are wider by a logarithmic factor in the inverse sparsity $1 / \log(1/\text{sparsity})$. While we are the first to show theoretically and experimentally that random ER source networks contain strong lottery tickets, we also prove the existence of weak lottery tickets that require a lower degree of overparametrization than strong lottery tickets. These unusual results are based on the observation that ER masks are well trainable in practice, which we verify in experiments with varied choices of random masks. Some of these data-free choices outperform previously proposed random approaches on standard image classification benchmark datasets. | We prove that random networks contain lottery tickets with high probability. |
Machine learning (ML) may be oblivious to human bias but it is not immune to its perpetuation. Marginalisation and iniquitous group representation are often traceable in the very data used for training, and may be reflected or even enhanced by the learning models.
In the present work, we aim at clarifying the role played by data geometry in the emergence of ML bias. We introduce an exactly solvable high-dimensional model of data imbalance, where parametric control over the many bias-inducing factors allows for an extensive exploration of the bias inheritance mechanism.Through the tools of statistical physics, we analytically characterise the typical properties of learning models trained in this synthetic framework and obtain exact predictions for the observables that are commonly employed for fairness assessment.
Despite the simplicity of the data model, we retrace and unpack typical unfairness behaviour observed on real-world datasets.
We also obtain a detailed analytical characterisation of a class of bias mitigation strategies. We first consider a basic loss-reweighing scheme, which allows for an implicit minimisation of different unfairness metrics, and quantify the incompatibilities between some existing fairness criteria. Then, we consider a novel mitigation strategy based on a matched inference approach, consisting in the introduction of coupled learning models. Our theoretical analysis of this approach shows that the coupled strategy can strike superior fairness-accuracy trade-offs. | We propose a generative model, exactly solvable using statistical physics, which emphasize the impact of data geometry in inducing bias in classification. |
This paper is about increasing the computational efficiency of clustering algorithms. Many clustering algorithms are based on properties of relative locations of points, globally or locally, e.g., interpoint distances and nearest neighbor distances. This amounts to using a lower dimensional space than the full dimensionality $D$ of the space in which the points are embedded. We present a clustering algorithm, Divide-and-Cluster (DAC), which detects local clusters in small neighborhoods obtained by recursive tessellation of space, and then merges them hierarchically, following the Divide-and-Conquer paradigm. This significantly reduces computation time which may otherwise grow nonlinearly number $n$ of points. We define locality as hypercubical neighborhoods in a recursive hypercubical decomposition of space, represented by a tree. Clusters are detected within each hypercube, and merged with those from neighboring hypercubes while traversing up the tree. We expect DAC to perform better than many other algorithms because (a) as clusters merge into larger clusters (components), their number steadily decreases vs the number of points, and (b) we cluster only neighboring components. The ordering of component appearances also simultaneously yields a cluster hierarchy (tree). Further, our use of small neighborhoods allows piecewise uniform approximation of large, nonuniform, arbitrary shaped clusters, thus avoiding the need for global cluster models. We experimentally verify the correctness of detected clusters on a variety of datasets, posing a variety of challenges, as well as show that DAC’s runtime is significantly better than representative algorithms of other types, particularly for increasing values of $n$.
| This paper clusters n points located in a D-dimensional space by detecting their mutual clustering affinity within local neighborhoods, using more efficient local computations, and then hierarchically growing the local clusters outward. |
Subgraph representation learning based on Graph Neural Network (GNN) has exhibited broad applications in scientific advancements, such as predictions of molecular structure-property relationships and collective cellular function. In particular, graph augmentation techniques have shown promising results in improving graph-based and node-based classification tasks. Still, they have rarely been explored in the existing GNN-based subgraph representation learning studies. In this study, we develop a novel multi-view augmentation mechanism to improve subgraph representation learning models and thus the accuracy of downstream prediction tasks. Our augmentation technique creates multiple variants of subgraphs and embeds these variants into the original graph to achieve highly improved training efficiency, scalability, and accuracy. Benchmark experiments on several real-world biological and physiological datasets demonstrate the superiority of our proposed multi-view augmentation techniques in subgraph representation learning. | We develop a novel multi-view augmentation mechanism to improve subgraph representation learning models and thus the accuracy of downstream prediction tasks. |
Topic models have been prevalent for decades with various applications like automatic text analysis due to their effectiveness and interpretability. However, existing topic models commonly suffer from the notorious topic collapsing issue: the discovered topics semantically collapse towards each other, leading to highly repetitive topics, insufficient topic discovery, and damaged model interpretability. In this paper, we propose a new neural topic model, Embedding Clustering Regularization Topic Model (ECRTM), to solve the topic collapsing issue. In addition to the reconstruction error of existing work, we propose a novel Embedding Clustering Regularization (ECR), which forces each topic embedding to be the center of a separately aggregated word embedding cluster in the semantic space. Instead of collapsing together, this makes topic embeddings away from each other and cover different semantics of word embeddings. Thus our ECR enables each produced topic to contain distinct word semantics, which alleviates topic collapsing. Through jointly optimizing our ECR objective and the neural topic modeling objective, ECRTM generates diverse and coherent topics together with high-quality topic distributions of documents. Extensive experiments on benchmark datasets demonstrate that ECRTM effectively addresses the topic collapsing issue and consistently surpasses state-of-the-art baselines in terms of topic quality, topic distributions of documents, and downstream classification tasks.
| We propose a neural topic model that addresses the topic collapsing issue with a novel clustering regularization on word and topic embeddings. |
Knowledge Graph Embedding (KGE) is a common method to complete real-world Knowledge Graphs (KGs) by learning the embeddings of entities and relations.
Beyond specific KGE models, previous work proposes a general framework based on group. A group has a special element identity that uniquely corresponds to the relation identity in KGs, which implies that identity should be represented uniquely. However, we find that this uniqueness cannot be modeled by bilinear based models, revealing the crack between the framework and models. To this end, we study the required conditions and propose a solution named Unit Ball Bilinear Model (UniBi). In addition to the theoretical superiority, UniBi is more robust and interpretable. Experiments demonstrate that UniBi models the uniqueness without the cost of performance and verify its a robustness and interpretability. | We scrutinize the identity relation in knowledge graphs, find that bilinear based models fail to uniquely model it, and propose a solution with other good properties. |
Neural networks have seen an explosion of usage and research in the past decade, particularly within the domains of computer vision and natural language processing. However, only recently have advancements in neural networks yielded performance improvements beyond narrow applications and translated to expanded multitask models capable of generalizing across multiple data types and modalities. Simultaneously, it has been shown that neural networks are overparameterized to a high degree, and pruning techniques have proved capable of significantly reducing the number of active weights within the network while largely preserving performance. In this work, we identify a methodology and network representational structure which allows a pruned network to employ previously unused weights to learn subsequent tasks. We employ these methodologies on well-known benchmarking datasets for testing purposes and show that networks trained using our approaches are able to learn multiple tasks, which may be related or unrelated, in parallel or in sequence without sacrificing performance on any task or exhibiting catastrophic forgetting. | In this paper, we test our "Multiple Subnetwork Hypothesis," which proposes that it is possible to train unused weights within a pruned feedforward neural network to learn subsequent tasks. |
Language models (LMs) have been instrumental for the rapid advance of natural language processing. This paper studies continual pre-training of LMs, in particular, continual domain-adaptive pre-training (or continual DAP-training). Existing research has shown that further pre-training an LM using a domain corpus to adapt the LM to the domain can improve the end-task performance in the domain. This paper proposes a novel method to continually DAP-train an LM with a sequence of unlabeled domain corpora to adapt the LM to these domains to improve their end-task performances. The key novelty of our method is a soft-masking mechanism that directly controls the update to the LM. A novel proxy is also proposed to preserve the general knowledge in the original LM. Additionally, it contrasts the representations of the previously learned domain knowledge (including the general knowledge in the pre-trained LM) and the knowledge from the current full network to achieve knowledge integration. The method not only overcomes catastrophic forgetting, but also achieves knowledge transfer to improve end-task performances. Empirical evaluation demonstrates the effectiveness of the proposed method. | This paper proposes a continual post-training method based on soft-masking to learn a sequence of unlabeled domain corpora to adapt a language model to improve the end-task performances in these domains. |
In this paper, we propose a self-supervised video representation learning (video SSL) method by taking inspiration from cognitive science and neuroscience on human visual perception. Different from previous methods that mainly start from the inherent properties of videos, we argue that humans learn to perceive the world through the self-awareness of the semantic change or consistency in the input stimuli in the absence of labels, accompanied by representation reorganization during the post-learning rest periods. To this end, we first exploit the presence of saccades as an indicator of semantic change in a contrastive learning framework to mimic the self-awareness in human representation learning, where the saccades are generated without eye-tracking data. Second, we model the semantic consistency by minimizing the prediction error between the predicted and the true state of another time point during a fixation. Third, we later incorporate prototypical contrastive learning to reorganize the learned representations such that perceptually similar representations would be associated closer. Compared to previous counterparts, our method can capture finer-grained semantics from video instances, and the associations among similar ones are further strengthened. Experiments show that the proposed bio-inspired video SSL method significantly improves the Top-1 video retrieval accuracy on UCF101 and achieves superior performance on downstream tasks such as action recognition under comparable settings. | In this paper, we propose a self-supervised video representation learning method by taking inspiration from cognitive science and neuroscience on human visual perceptionization. |
We approach the fundamental problem of obstacle avoidance for robotic systems via the lens of online learning. In contrast to prior work that either assumes worst-case realization of uncertainty in the environment or a given stochastic model of uncertainty, we propose a method that is efficient to implement and provably grants instance-optimality to perturbations of trajectories generated from an open-loop planner in the sense of minimizing worst-case regret. The resulting policy thus adapts online to realizations of uncertainty and provably compares well with the best obstacle avoidance policy in hindsight from a rich class of policies. The method is validated in simulation on a dynamical system environment and compared to baseline open-loop planning and robust Hamilton-Jacobi reachability techniques. | Regret bounds for online learning obstacle avoidance policies |
Recent works on continual learning have shown that unsupervised continual learning (UCL) methods rival or even beat supervised continual learning methods. However, most UCL methods typically adopt fixed learning strategies with pre-defined objectives and ignore the influence of the constant shift of data distributions on the newer training process. This non-adaptive paradigm tends to achieve sub-optimal performance, since the optimal update direction (to ensure the trade-off between old and new tasks) keeps changing during training over sequential tasks. In this work, we thus propose a novel UCL framework termed AUDR to adaptively rectify the update direction by a policy network (i.e., the Actor) at each training step based on the reward predicted by a value network (i.e., the Critic). Concretely, different from existing Actor-Critic based reinforcement learning works, there are three vital designs that make our AUDR applicable to the UCL setting: (1) A reward function to measure the score/value of the currently selected action, which provides the ground-truth reward to guide the Critic's predictions; (2) An action space for the Actor to select actions (i.e., update directions) according to the reward predicted by the Critic; (3) A multinomial sampling strategy with a lower-bound on the sampling probability of each action, which is designed to improve the variance of the Actor's selected actions for more diversified exploration. Extensive experiments show that our AUDR achieves state-of-the-art results under both the in-dataset and cross-dataset UCL settings. Importantly, our AUDR also shows superior performance when combined with other UCL methods, which suggests that our AUDR is highly extensible and versatile. | We propose an Actor-Critic framework with adaptive update direction rectification for unsupervised continual learning. |
AlphaZero-type algorithms may stop improving on single-player tasks in case the value network guiding the tree search is unable to approximate the outcome of an episode sufficiently well. One technique to address this problem is transforming the single-player task through self-competition. The main idea is to compute a scalar baseline from the agent’s historical performances and to reshape an episode’s reward into a binary output, indicating whether the baseline has been exceeded or not. However, this baseline only carries limited information for the agent about strategies how to improve. We leverage the idea of self-competition and directly incorporate a historical policy into the planning process instead of its scalar performance. Based on the recently introduced Gumbel AlphaZero (GAZ), we propose our algorithm GAZ ‘Play-to-Plan’ (GAZ PTP), in which the agent learns to find strong trajectories by planning against possible strategies of its past self. We show the effectiveness of our approach in two well-known combinatorial optimization problems, the Traveling Salesman Problem and the Job-Shop Scheduling Problem. With only half of the simulation budget for search, GAZ PTP consistently outperforms all selected single-player variants of GAZ. | Solving deterministic single-agent problems through self-competition by including a historical policy in the planning process of Gumbel AlphaZero. |
The technical progression of artificial intelligence (AI) research has been built on breakthroughs in fields such as computer science, statistics, and mathematics. However, in the past decade AI researchers have increasingly looked to the social sciences, turning to human interactions to solve the challenges of model development. Paying crowdsourcing workers to generate or curate data, or ‘data enrichment’, has become indispensable for many areas of AI research, from natural language processing to inverse reinforcement learning. Other fields that routinely interact with crowdsourcing workers, such as Psychology, have developed common governance requirements and norms to ensure research is undertaken ethically. This study explores how, and to what extent, comparable research ethics requirements and norms have developed for AI research and data enrichment. We focus on the approach taken by two leading AI conferences: ICLR and NeurIPS. In a longitudinal study of accepted papers, and a comparison with Springer journal articles and Psychology papers, this work finds that ICLR and NeurIPS have established protocols for human data collection which are inconsistently followed by authors. Whilst Psychology papers engaging with crowdsourcing workers frequently disclose ethics reviews, payment data, demographic data and other information, such disclosures are far less common in leading AI conferences despite similar guidance. The work concludes with hypotheses to explain these gaps in research ethics practices and considerations for its implications. | This paper shows how AI researchers engage with research ethics when employing crowdworkers. The work finds research ethics disclosures are infrequent in AI papers, inconsistently following venue publication policies. |
Understanding how convolutional neural networks (CNNs) can efficiently learn high-dimensional functions remains a fundamental challenge. A popular belief is that these models harness the local and hierarchical structure of natural data such as images. Yet, we lack a quantitative understanding of how such structure affects performance, e.g. the rate of decay of the generalisation error with the number of training samples. In this paper, we study deep CNNs in the kernel regime. First, we show that the spectrum of the corresponding kernel inherits the hierarchical structure of the network, and we characterise its asymptotics. Then, we use this result together with generalisation bounds to prove that deep CNNs adapt to the spatial scale of the target function. In particular, we find that if the target function depends on low-dimensional subsets of adjacent input variables, then the rate of decay of the error is controlled by the effective dimensionality of these subsets. Conversely, if the teacher function depends on the full set of input variables, then the error rate is inversely proportional to the input dimension. We conclude by computing the rate when a deep CNN is trained on the output of another deep CNN with randomly-initialised parameters. Interestingly, we find that despite their hierarchical structure, the functions generated by deep CNNs are too rich to be efficiently learnable in high dimension. | theoretical study of generalisation rates for deep CNNs in the kernel regime |
Estimating the energetic properties of molecular systems is a critical task in material design. With the trade-off between accuracy and computational cost, various methods have been used to predict the energy of materials, including recent neural-net-based models. However, most existing neural-net models are context-free (physics-ignoring) black-box models, limiting their applications to predict energy only within the distribution of the training set and thus preventing from being applied to the real practice of molecular design. Inspired by the physical mechanism of the interatomic potential, we propose a physics-driven energy prediction model using a Transformer. Our model is trained not only on the energy regression in the training set, but also with conditions inspired by physical insights and self-supervision based on Masked Atomic Modeling, making it adaptable to the optimization of molecular structure beyond the range observed during training, taking a step towards realizable molecular structure optimization. | We propose a Transformer-based molecular energy prediction model equipped with physical insights and self-supervised masked atomic modeling. |
Vertical federated learning (VFL) is a distributed learning paradigm, where computing clients collectively train a model based on the partial features of the same set of samples they possess. Current research on VFL focuses on the case when samples are independent, but it rarely addresses an emerging scenario when samples are interrelated through a graph. For graph-structured data, graph neural networks (GNNs) are rather competitive machine learning models, but a naive implementation in the VFL setting causes a significant communication overhead; moreover, the analysis is faced with a challenge caused by the biased stochastic gradients. In this paper, we propose a model splitting method that splits a backbone GNN across the clients and the server and a communication-efficient algorithm, GLASU, to train such a model. GLASU adopts lazy aggregation and stale updates to skip aggregation when evaluating the model and skip feature exchanges during training, greatly reducing communication. We offer a theoretical analysis and conduct extensive numerical experiments on real-world datasets, showing that the proposed algorithm effectively trains a GNN model, whose performance matches that of the backbone GNN when trained in a centralized manner. | This paper proposed a GNN model design approach and a communication efficient algorithm for federated learning on feature distributed graph data |
Estimating treatment effects from observational data is a central problem in causal inference. Methods to solve this problem exploit inductive biases and heuristics from causal inference to design multi-head neural network architectures and regularizers. In this work, we propose to use neurosymbolic program synthesis, a data-efficient, and interpretable technique, to solve the treatment effect estimation problem. We theoretically show that neurosymbolic programming can solve the treatment effect estimation problem. By designing a Domain Specific Language (DSL) for treatment effect estimation based on the inductive biases used in literature, we argue that neurosymbolic programming is a better alternative to treatment effect estimation than traditional models. Our empirical study reveals that our model, which implicitly encodes inductive biases in a DSL, achieves better performance on benchmark datasets than the state-of-the-art models. | We estimate treatment effects/ causal effects using neurosymbolic program synthesis by designing a domain specific language |
Models that can predict adverse events ahead of time with low false-alarm rates are critical to the acceptance of decision support systems in the medical community. This challenging machine learning task remains typically treated as a simple binary classification, with few bespoke methods proposed to leverage temporal dependency across samples. We propose Temporal Label Smoothing (TLS), a novel learning strategy that modulates smoothing strength as a function of proximity to the event of interest. This regularization technique reduces model confidence at the class boundary, where the signal is often noisy or uninformative, thus allowing training to focus on clinically informative data points away from this boundary region. From a theoretical perspective, we also show that our method can be framed as an extension of multi-horizon prediction, a learning heuristic proposed in other early prediction work. TLS empirically matches or outperforms all competitor methods across all evaluation measures on various early prediction benchmark tasks. In particular, our approach significantly improves performance on clinically-relevant metrics such as event recall under low false-alarm rates. | Modulating label smoothing strength over time to reflect signal noise patterns and clinical priorities significantly improves deep learning model performance in the prediction of adverse medical events. |
Curricula for goal-conditioned reinforcement learning agents typically rely on poor estimates of the agent's epistemic uncertainty or fail to consider the agents' epistemic uncertainty altogether, resulting in poor sample efficiency. We propose a novel algorithm, Query The Agent (QTA), which significantly improves sample efficiency by estimating the agent's epistemic uncertainty throughout the state space and setting goals in highly uncertain areas. Encouraging the agent to collect data in highly uncertain states allows the agent to improve its estimation of the value function rapidly. QTA utilizes a novel technique for estimating epistemic uncertainty, Predictive Uncertainty Networks (PUN), to allow QTA to assess the agent's uncertainty in all previously observed states. We demonstrate that QTA offers decisive sample efficiency improvements over preexisting methods. | Designing more sample efficient reinforcement learning curricula by measuring and exploiting agents' epistemic uncertainty. |
Automatic live video commenting is with increasing attention due to its significance in narration generation, topic explanation, etc. However, the sentiment consideration of the generated comments is missing from the current methods. Thus, in this paper, we introduce and investigate a task, namely sentiment-guided automatic live video commenting, which aims to generate live video comments based on sentiment guidance. To address this problem, we propose a Sentiment-oriented Transformer-based Variational Autoencoder (So-TVAE) network, which consists of a sentiment-oriented diversity encoder module and a batch-attention module. Specifically, our sentiment-oriented diversity encoder elegantly combines VAE and random mask mechanism to achieve semantic diversity under sentiment guidance, which is then fused with cross-modal features to generate live video comments. Furthermore, a batch attention module is also proposed in this paper to alleviate the problem of missing sentimental samples, caused by the data imbalance, which is common in live videos as the popularity of video varies. Extensive experiments on Livebot and VideoIC datasets demonstrate that the proposed So-TVAE outperforms the state-of-the-art methods in terms of the quality and diversity of generated comments. Related codes will be released. | This paper proposes a Sentiment-oriented Transformer-based Variational Autoencoder model which can achieve diverse video commenting with multiple sentiments and semantics for the automatic live video commenting task. |
Modern machine learning research relies on relatively few carefully curated datasets. Even in these datasets, and typically in `untidy' or raw data, practitioners are faced with significant issues of data quality and diversity which can be prohibitively labor intensive to address. Existing methods for dealing with these challenges tend to make strong assumptions about the particular issues at play, and often require a priori knowledge or metadata such as domain labels. Our work is orthogonal to these methods: we instead focus on providing a unified and efficient framework for Metadata Archaeology -- uncovering and inferring metadata of examples in a dataset. We curate different subsets of data that might exist in a dataset (e.g. mislabeled, atypical, or out-of-distribution examples) using simple transformations, and leverage differences in learning dynamics between these probe suites to infer metadata of interest. Our method is on par with far more sophisticated mitigation methods across different tasks: identifying and correcting mislabeled examples, classifying minority-group samples, prioritizing points relevant for training and enabling scalable human auditing of relevant examples. | Our work provides a unified and efficient framework for Metadata Archaeology -- uncovering and inferring metadata of examples in a dataset |
Output reachability and adversarial robustness are among the most relevant safety properties of neural networks.
We show that in the context of Message Passing Neural Networks (MPNN), a common Graph Neural Network (GNN) model,
formal verification is impossible. In particular, we show that output reachability of graph-classifier MPNN,
working over graphs of unbounded size, non-trivial degree and sufficiently expressive node labels, cannot be verified formally: there
is no algorithm that answers correctly (with yes or no), given an MPNN, whether there exists some valid input to
the MPNN such that the corresponding output satisfies a given specification. However, we also show that
output reachability and adversarial robustness of node-classifier MPNN can be verified formally when a limit on
the degree of input graphs is given a priori. We discuss the implications of these results, for the purpose of
obtaining a complete picture of the principle possibility to formally verify GNN, depending on
the expressiveness of the involved GNN models and input-output specifications. | We prove that certain safety properties of MPNN can not be verified formally. |
Learning high-quality representation is important and essential for visual recognition. Unfortunately, traditional representation learning suffers from fairness issues since the model may learn information of sensitive attributes. Recently, a series of studies have been proposed to improve fairness by explicitly decorrelating target labels and sensitive attributes. Most of these methods, however, rely on the assumption that fully annotated labels on target variable and sensitive attributes are available, which is unrealistic due to the expensive annotation cost. In this paper, we investigate a novel and practical problem of Fair Unsupervised Representation Learning with Partially annotated Sensitive labels (FURL-PS). FURL-PS has two key challenges: 1) how to make full use of the samples that are not annotated with sensitive attributes; 2) how to eliminate bias in the dataset without target labels. To address these challenges, we propose a general Fairness-aware Contrastive Learning (FairCL) framework consisting of two stages. Firstly, we generate contrastive sample pairs, which share the same visual information apart from sensitive attributes, for each instance in the original dataset. In this way, we construct a balanced and unbiased dataset. Then, we execute fair contrastive learning by closing the distance between representations of contrastive sample pairs. Besides, we also propose an unsupervised way to balance the utility and fairness of learned representations by feature reweighting. Extensive experimental results illustrate the effectiveness of our method in terms of fairness and utility, even with very limited sensitive attributes and serious data bias. | Proposing a new problem of fair unsupervised representation learning with limited annotated sensitive attributes and a fairness-aware contrastive learning framework. |
Learned classifiers should often possess certain invariance properties meant to encourage fairness, robustness, or out-of-distribution generalization.
However, multiple recent works empirically demonstrate that common invariance-inducing regularizers are ineffective in the over-parameterized regime, in which classifiers perfectly fit (i.e. interpolate) the training data. This suggests that the phenomenon of ``benign overfitting," in which models generalize well despite interpolating, might not favorably extend to settings in which robustness or fairness are desirable.
In this work, we provide a theoretical justification for these observations. We prove that---even in the simplest of settings---any interpolating learning rule (with an arbitrarily small margin) will not satisfy these invariance properties. We then propose and analyze an algorithm that---in the same setting---successfully learns a non-interpolating classifier that is provably invariant. We validate our theoretical observations on simulated data and the Waterbirds dataset. | Proof that interpolating classifiers cannot satisfy common invariance and fairness criteria; Provides insight on empirical observations and possible effective solutions |
Physically rearranging objects is an important capability for embodied agents. Visual room rearrangement evaluates an agent's ability to rearrange objects in a room to a desired goal based solely on visual input. We propose a simple yet effective method for this problem: (1) search for and map which objects need to be rearranged, and (2) rearrange each object until the task is complete. Our approach consists of an off-the-shelf semantic segmentation model, voxel-based semantic map, and semantic search policy to efficiently find objects that need to be rearranged. Our method was the winning submission to the AI2-THOR Rearrangement Challenge in the 2022 Embodied AI Workshop at CVPR 2022, and improves on current state-of-the-art end-to-end reinforcement learning-based methods that learn visual room rearrangement policies from 0.53% correct rearrangement to 16.56%, using only 2.7% as many samples from the environment. | A System For Exploring A Scene, Mapping Objects, and Rearranging Objects To A Visual Goal |
Formal logic specifications are a useful tool to describe desired agent behavior and have been explored as a means to shape rewards in Deep Reinforcement Learning (DRL) systems over a variety of problems and domains. Prior work, however, has failed to consider the possibility of making these specifications differentiable, which would yield a more informative signal of the objective via the specification gradient. This paper examines precisely such an approach by exploring a Lagrangian method to constrain policy updates using a differentiable style of temporal logic specifications that associates logic formulae with real-valued quantitative semantics. This constrained learning mechanism is then used in a hierarchical setting where a high-level specification-guided neural network path planner works with a low-level control policy to navigate through planned waypoints. The effectiveness of our approach is demonstrated over four robot dynamics with five different types of Linear Temporal Logic (LTL) specifications. Our demo videos are collected at https://sites.google.com/view/schrl. | This paper uses differentiable formal specifications to constrain the policy updates in hierarchical deep reinforcement learning. |
The exploration problem is one of the main challenges in deep reinforcement learning (RL). Recent promising works tried to handle the problem with population-based methods, which collect samples with diverse behaviors derived from a population of different exploratory policies. Adaptive policy selection has been adopted for behavior control. However, the behavior selection space is largely limited by the predefined policy population, which further limits behavior diversity. In this paper, we propose a general framework called Learnable Behavioral Control (LBC) to address the limitation, which a) enables a significantly enlarged behavior selection space via formulating a hybrid behavior mapping from all policies; b) constructs a unified learnable process for behavior selection. We introduce LBC into distributed off-policy actor-critic methods and achieve behavior control via optimizing the selection of the behavior mappings with bandit-based meta-controllers. Our agents have achieved 10077.52% mean human normalized score and surpassed 24 human world records within 1B training frames in the Arcade Learning Environment, which demonstrates our significant state-of-the-art (SOTA) performance without degrading the sample efficiency. | We have constructed a general framework to control the behaviors in RL and achieved SOTA performance in Atari 1B benchmark. |
The goal of automated feature generation is to liberate machine learning experts from the laborious task of manual feature generation, which is crucial for improving the learning performance of tabular data. The major challenge in automated feature generation is to efficiently and accurately identify useful features from a vast pool of candidate features. In this paper, we present OpenFE, an automated feature generation tool that provides competitive results against machine learning experts. OpenFE achieves efficiency and accuracy with two components: 1) a novel feature boosting method for accurately estimating the incremental performance of candidate features. 2) a feature-scoring framework for retrieving effective features from a large number of candidates through successive featurewise halving and feature importance attribution. Extensive experiments on seven benchmark datasets show that OpenFE outperforms existing baseline methods. We further evaluate OpenFE in two famous Kaggle competitions with thousands of data science teams participating. In one of the competitions, features generated by OpenFE with a simple baseline model can beat 99.3% data science teams, demonstrating for the first time that automated feature generation can outperform human experts. In addition to the empirical results, we provide a theoretical perspective to show that feature generation has benefit provably in a simple yet representative setting. Codes and datasets are available in the supplementary materials. | OpenFE: automated feature generation beyond expert-level performance |
Deep neural networks are valuable assets considering their commercial benefits and huge demands for costly annotation and computation resources. To protect the copyright of these deep models, backdoor-based ownership verification becomes popular recently, in which the model owner can watermark the model by embedding a specific behavior before releasing it. The defender (usually the model owner) can identify whether a suspicious third-party model is ``stolen'' from it based on the presence of the behavior. Unfortunately, these watermarks are proven to be vulnerable to removal attacks even like fine-tuning. To further explore this vulnerability, we investigate the parametric space and find there exist many watermark-removed models in the vicinity of the watermarked one, which may be easily used by removal attacks. Inspired by this finding, we propose a minimax formulation to find these watermark-removed models and recover their watermark behavior. Extensive experiments demonstrate that our method improves the robustness of the model watermarking against parametric changes and numerous watermark-removal attacks. | Based on the observation of the watermarked model in parametric space, we propose a minimax approach to improve the robustness of watermarked models against state-of-the-art removal attacks. |
Given a particular embodiment, we propose a novel method (C3PO) that learns policies able to achieve any arbitrary position and pose. Such a policy would allow for easier control, and would be re-useable as a key building block for downstream tasks. The method is two-fold: First, we introduce a novel exploration algorithm that optimizes for uniform coverage, is able to discover a set of achievable states, and investigates its abilities in attaining both high coverage, and hard-to-discover states; Second, we leverage this set of achievable states as training data for a universal goal-achievement policy, a goal-based SAC variant. We demonstrate the trained policy's performance in achieving a large number of novel states. Finally, we showcase the influence of massive unsupervised training of a goal-achievement policy with state-of-the-art pose-based control of the Hopper, Walker, Halfcheetah, Humanoid and Ant embodiments. | Exploration approximating a uniform sampling over possible states to train a policy that can achieve any pose and position. |
Subsets and Splits