Datasets:
id
stringlengths 4
8
| image
imagewidth (px) 4
1.34k
| latex
stringlengths 90
2.83k
|
---|---|---|
CV-26906 | \begin{tabular}{|l|c|c|}
\hline
Model & Test Accuracy & Perplexity \\
\hline
Nearest Neighbor & $14.09$ & \texttt{N/A} \\
\hline
CNN & $14.61$ & $0.1419$ \\
\hline
Our Model & $\mathbf{19.77}$ & $\mathbf{0.2362}$ \\
\hline
\end{tabular} |
|
AI-8358 | \begin{tabular}{|p{50pt}|p{50pt}<{\centering}|p{50pt}<{\centering}|p{50pt}<{\centering}|}
\hline
& \textbf{Types} & \textbf{HAS} & \textbf{HAS+r} \\ \hline
\textbf{L.MDB} & Film & 0.38 & \textbf{0.44} \\ \hline
\multirow{15}{*}{\textbf{DBpedia}} & Airl. & 0.402 & \textbf{0.424} \\
& Band & 0.26 & \textbf{0.56} \\
& Base. & 0.46 & \textbf{0.7} \\
& Lake & 0.28 & \textbf{0.4} \\
& Univ. & 0.177 & \textbf{0.406} \\
& Phil. & 0.288 & \textbf{0.667} \\
& Song & 0.538 & \textbf{0.807} \\
& Poli. & 0.209 & \textbf{0.524} \\
& TVsh. & 0.186 & \textbf{0.478} \\
& Come. & 0.528 & \textbf{0.575} \\
& Acad & \textbf{0.84} & \textbf{0.84} \\
& Acto. & 0.36 & \textbf{0.42} \\
& Book. & \textbf{0.6} & \textbf{0.6} \\
& Moun. & 0.609 & \textbf{0.645} \\
& Radi. & \textbf{0.62} & 0.532 \\ \hline
\multicolumn{2}{|c|}{\textbf{Average}} & 0.421 & \textbf{0.566} \\ \hline
\end{tabular} |
|
CV-10464 | \begin{tabular}{|c|c|c|c|c|c|c|c|c|c|c|c|c|}
\hline
\multirow{3}*{Method} & \multicolumn{6}{c|}{IoU = 0.5} & \multicolumn{6}{c|}{IoU = 0.7} \\
\cline{2-13}
~ & \multicolumn{2}{c|}{ Easy} & \multicolumn{2}{c|}{Moderate} & \multicolumn{2}{c|}{Hard} & \multicolumn{2}{c|}{ Easy} & \multicolumn{2}{c|}{Moderate} & \multicolumn{2}{c|}{Hard} \\
\cline{2-13}
~ & t/v 1 & t/v 2 & t/v 1 & t/v 2 & t/v 1 & t/v 2 & t/v 1 & t/v 2 & t/v 1 & t/v 2 & t/v 1 & t/v 2 \\
\hline\hline
3DOP & 46.04 & - & 34.63 & - & 30.09 & - & 6.55 & - & 5.07 & - & 4.10 & - \\
Mono3D & 25.19 & - & 18.20 & - & 15.52 & - & 2.53 & - & 2.31 & - & 2.31 & - \\
Deep3DBox & - & 27.04 & - & 20.55 & - & 15.88 & - & 5.85 & - & 4.10 & - & 3.84 \\
\hline
Our Method & 28.16 & 28.98 & 21.02 & 20.71 & 19.91 & 18.59 & 5.98 & 5.45 & 5.50 & 5.11 & 4.75 & 4.45 \\
\hline
\end{tabular} |
|
SE-20655 | \begin{tabular}[c]{@{}l@{}}a)DifferentorganizationsgenerateSBOMsatdifferentSDLCstages.\\b)MoreorganizationsfavorincludingmorethanbaselineSBOMinformation.\end{tabular} |
|
CV-28515 | \begin{tabular}{l|c|c|c|c}
\hline
Benchmark & Easy & Moderate & Hard & mAP \\
\hline
Cars~(3D Detection) & 88.21 & 77.85 & 75.62 & 80.56 \\
Cars~(BEV Detection) & 90.17 & 87.55 & 87.14 & 88.29 \\
Pedestrians~(3D Detection) & 70.80 & 63.45 & 58.22 & 64.16 \\
Pedestrians~(BEV Detection) & 76.70 & 70.76 & 65.13 & 70.86 \\
Cyclists~(3D Detection) & 85.98 & 64.95 & 60.40 & 70.44 \\
Cyclists~(BEV Detection) & 87.17 & 66.71 & 63.79 & 72.56 \\
\hline
\end{tabular} |
|
AI-22172 | \begin{tabular}{rl|rl|rl|rl|rl}
Freq. & Word & Freq. & Word & Freq. & Word & Freq. & Word & Freq. & Word \\
\hline
3,205 & sport & 464 & use & 296 & carrying & 155 & crossing & 92 & popular \\
1,153 & appear & 461 & chair & 287 & face & 150 & tires & 92 & planning \\
976 & fire & 430 & parked & 283 & writing & 148 & cabinets & 88 & batters \\
898 & pattern & 415 & coming & 230 & television & 145 & beds & 85 & enjoy \\
845 & material & 399 & buses & 223 & vegetarian & 127 & mountain & 85 & dirt \\
689 & bed & 395 & take & 220 & beside & 119 & levels & 85 & carpet \\
669 & facing & 357 & utensil & 201 & graffiti & 117 & catcher & 83 & slice \\
612 & big & 335 & dish & 167 & foot & 113 & falling & 80 & salad \\
565 & trying & 321 & sink & 163 & couch & 101 & faces & 79 & square \\
474 & sandwich & 317 & three & 160 & silver & 98 & fireplace & 78 & roll \\
\end{tabular} |
|
CR-44831 | \begin{tabular}{ccc}
\hline
Notations & Description & Time (ms) \\
\hline
${T_C}$ & Time cost of encryption & 0.096 \\
${T_{agg}}$ & Time cost of aggregating 200 readings & 2.21 \\
${T_{decAgg}}$ & Time cost of decrypting aggregated readings & 0.135 \\
${T_{DM}}$ & Time cost of public key generation $DW$ & 45.36 \\
${T_{decDW}}$ & Time cost of decrypting to obtain $\bm{rW}$ & 49.63 \\
$T{S_t}$ & Time cost of generating timestamp & 0.852 \\
${T_{sig}}$ & Time cost of signature operation & 13.18 \\
${T_{versig}}$ & Time cost of the verify signature operation & 127.29 \\
${T_m}$ & Time cost of model detection & 56.03 \\
\hline
\end{tabular} |
|
SE-22997 | \begin{tabular}{lllll}
\hline\hline
program & size(KLOC) & Times(secs) & Bug Count & False Count \\
\hline
gcc & 230.4 & 213.1 & 36 & 6 \\
ammp & 13.4 & 10.4 & 23 & 5 \\
bash & 100.0 & 90.1 & 16 & 3 \\
mesa & 61.3 & 48.6 & 9 & 8 \\
cluster & 10.7 & 9.5 & 12 & 4 \\
openCV & 794.6 & 756.8 & 74 & 11 \\
bitcoin & 94.4 & 78.7 & 22 & 7 \\
Total & 1304.8 & 1257.9 & 192 & 44 \\
\hline\hline
\end{tabular} |
|
CR-29638 | \begin{tabular}{ccccc}
\toprule
Image size & $1024\times 768$ & $1600\times 1200$ & $3240\times 2592$ & $4800\times 4800$ \\
\hline
DCT on laptop GPU & 0.41 ms & 0.79 ms & 3.67 ms & 9.98 ms \\
AES on laptop CPU & 0.19 ms & 0.47 ms & 2.05 ms & 5.87 ms \\
\bottomrule
\end{tabular} |
|
AI-29454 | \begin{tabular}{|p{0.30\textwidth}|p{0.275\textwidth}|p{0.28\textwidth}|}
\hline
\textbf{Dynamic conbditions} & \textbf{Action} & \textbf{Static Conditions} \\ \hline
\makecell*[lt]{$robAt(R1)$} & \makecell*[lt]{$moveTo(R1,R0,D0)$} & \makecell*[lt]{$connected(D0,R0,R1)$} \\ \hline
\makecell*[lt]{$isHeld(K,G)$} & \makecell*[lt]{$pickup(K,G,L0,R0)$} & \makecell*[lt]{$key(K)$ \\ \& $isCard(K)$} \\ \hline
\makecell*[lt]{$isHeld(K,G)$} & \makecell*[lt]{$semi\_e\_isHeld(K,G)$} & \makecell*[lt]{$key(K)$ \\ \& $isCard(K)$ \\ \& $hand(G)$} \\
\hline
\end{tabular} |
|
SE-4760 | \begin{tabular}{lccc}
Action class & Mean LOC & Formula & P(class) \\
\hline
\vspace{0.01in}
{\tt <int> := <[1..20]>} & 0 & $\frac{0.20}{2}$ & 0.100 \\
{\tt <ch> := <['r','w']>} & 0 & $\frac{0.20}{2}$ & 0.100 \\
\vspace{0.03in}
{\tt f(<int>)} & 30 & $\frac{30}{64} \times 0.80$ & 0.375 \\
{\tt g(<int>)} & 20 & $\frac{6+14}{64} \times 0.80$ & 0.250 \\
{\tt h(<ch>)} & 14 & $\frac{14}{64} \times 0.80$ & 0.175 \\
\end{tabular} |
|
CR-12294 | \begin{tabular}{|c|c|c|c|c|}
\hline
Vulnerability & Arbiter & MEM & GNG & AES \\
\hline
Permissions and Privileges & & \checkmark & & \\
\hline
Resource Management & & & & \checkmark \\
\hline
Illegal States \& Transitions & \checkmark/\checkmark & \checkmark & & \\
\hline
Buffer Issues & & \checkmark & & \\
\hline
Information Leakage & \checkmark/\checkmark & & & \\
\hline
Numeric Exceptions & & & \checkmark & \\
\hline
Malicious Implants & & & & \checkmark \\
\hline
\end{tabular} |
|
CR-40148 | \begin{tabular}{llcccccc}
\hline
\textbf{Label} & \textbf{Person} & \textbf{Sex} & \textbf{Language} & \textbf{Length(seconds)} & \textbf{Testing words} & \textbf{Training words} & \textbf{Overlapping words} \\ \hline \hline
User$_1$ & Bill Gates & male & English & 7068 & 179 & 12593 & 19 \\ \hline
User$_2$ & Feifei Li & female & English & 7120 & 182 & 17626 & 15 \\ \hline
User$_3$ & Pony Ma & male & Chinese & 5180 & 215 & 28554 & 20 \\ \hline
User$_4$ & Jane Goodall & female & English & 7484 & 188 & 11339 & 23 \\ \hline
User$_5$ & Jiaying Ye & female & Chinese & 9032 & 188 & 11339 & 16 \\ \hline
User$_6$ & Mingzhu Dong & female & Chinese & 5428 & 234 & 18709 & 22 \\ \hline
User$_7$ & Steve Job & male & English & 14836 & 190 & 37751 & 17 \\ \hline
User$_8$ & Yansong Bai & male & Chinese & 6792 & 251 & 27317 & 22 \\ \hline
User$_9$ & Anne Hathaway & female & English & 60 & 197 & * & 21 \\ \hline
User$_{10}$ & Elon Musk & male & English & 60 & 156 & * & 17 \\ \hline
User$_{11}$ & Mark Zuckerberg & male & English & 60 & 177 & * & 15 \\ \hline
User$_{12}$ & Oprah Winfrey & female & English & 60 & 167 & * & 18 \\ \hline
User$_{13}$ & Lan Yang & female & Chinese & 60 & 289 & * & 25 \\ \hline
User$_{14}$ & Minhong Yu & male & Chinese & 60 & 199 & * & 17 \\ \hline
User$_{15}$ & Robin Li & male & Chinese & 60 & 244 & * & 20 \\ \hline
User$_{16}$ & Yingtai Long & female & Chinese & 60 & 198 & * & 18 \\ \hline
\end{tabular} |
|
AI-6294 | \begin{tabular}{|l|l|l|}
\hline
\multirow{2}{*}{Pretraining} & \# of Stays in Stay Level Pretraining & 100563 \\
& \# of Admissions in Admission Level Pretraining & 99000 \\ \hline
\multirow{2}{*}{Stay Level Tasks} & \# of Stays in ARF Prediction & 4205 \\
& \# of Stays in Shock Prediction & 6190 \\ \hline
Admission Level Tasks & \# of Admissions in Readmission Prediction & 33179 \\ \hline
\multirow{4}{*}{Patient Level Tasks} & \# of Patients in Heart Failure Prediction & 12320 \\
& \# of Patients in COPD Prediction & 29256 \\
& \# of Patients in Amnesia Prediction & 11928 \\
& \# of Patients in Heart Failure Prediction (MIMIC-III) & 7522 \\ \hline
\end{tabular} |
|
AI-935 | \begin{tabular}{lrrrrrr}
\hline
& Sum Sq & Mean Sq & NumDF & DenDF & F value & Pr($>$F) \\
\hline
transparency & 0.10 & 0.10 & 1.00 & 994.00 & 5.83 & 0.0159 \\
num\_features & 0.04 & 0.04 & 1.00 & 994.00 & 2.15 & 0.1427 \\
transparency:num\_features & 0.00 & 0.00 & 1.00 & 994.00 & 0.06 & 0.8143 \\
\hline
\end{tabular} |
|
AI-27264 | \begin{tabular}{l|l|l|l|l}
\toprule
\makecell[l]{Engage- \\ment} & \makecell[l]{Question\\difficulty} & $P_{Rresp}$ & $P_{IRresp}$ & $P_{Nresp}$ \\
\midrule
\multirow{3}{*}{High} & Easy & $1$ & $0$ & $0$ \\
{} & Moderate & $1$ & $0$ & $0$ \\
{} & Difficult & $1$ & $0$ & $0$ \\
\hline
\multirow{3}{*}{Medium} & Easy & $0.95$ & $0$ & $0.05$ \\
{} & Moderate & $0.92$ & $0$ & $0.08$ \\
{} & Difficult & $0.90$ & $0$ & $0.10$ \\
\hline
\multirow{3}{*}{Low} & Easy & $0.90$ & $0$ & $0.10$ \\
{} & Moderate & $0.88$ & $0$ & $0.12$ \\
{} & Difficult & $0.85$ & $0$ & $0.15$ \\
\bottomrule
\end{tabular} |
|
CR-40936 | \begin{tabular}{|cc|}
\hline
\multicolumn{2}{|c|}{A2Y} \\
\hline
local & cloud \\
\hline \hline
0 & 0 \\
\hline
\end{tabular} |
|
SE-6501 | \begin{tabular}{@{}p{65mm}@{}}
\emph{Project: avajs/ava; Issue: $1400$} \\
``... There is already a PR for this though, thanks
to @tdeschryver ...''
\end{tabular} |
|
CR-10565 | \begin{tabular}{lrrrrrr}
\multicolumn{1}{c}{Dataset} & \multicolumn{3}{c}{CIFAR-10} & \multicolumn{3}{c}{CIFAR-100} \\
\cmidrule(lr){1-1}
\cmidrule(lr){2-4}
\cmidrule(lr){5-7}
Defense level & No Def. & Mixup+MMD & Mem-Guard & No Def. & Mixup+MMD & Mem-Guard \\
\cmidrule(lr){1-4}
\cmidrule(lr){4-7}
Training accuracy & 0.994 & {\bf 0.881} & 0.997 & 0.995 & {\bf 0.665} & 0.979 \\
Testing accuracy & 0.761 & {\bf 0.765} & 0.762 & 0.326 & {\bf 0.337} & {\bf 0.338} \\
\cmidrule(lr){1-4}
\cmidrule(lr){4-7}
\textbf{Generalization gap} & 0.232 & {\bf 0.116} & 0.235 & 0.669 & {\bf 0.328} & 0.641 \\
\textbf{Largest attack advantage} & 0.166 & {\bf 0.067} & 0.113 & 0.356 & {\bf 0.166} & 0.324 \\
\textbf{Baseline attack advantage} & 0.116 & {\bf 0.067} & 0.112 & 0.333 & {\bf 0.166} & 0.324 \\
\cmidrule(lr){1-4}
\cmidrule(lr){4-7}
Global-Probability attack advantage & 0.156 & {\bf 0.067} & 0.112 & 0.356 & {\bf 0.166} & 0.320 \\
Global-Loss attack advantage & 0.166 & {\bf 0.056} & 0.113 & 0.356 & {\bf 0.155} & 0.319 \\
Global-TopOne attack advantage & 0.120 & 0.049 & {\bf 0.028} & 0.249 & 0.103 & {\bf 0.093} \\
Global-TopThree attack advantage & 0.140 & 0.052 & {\bf 0.027} & 0.273 & 0.104 & {\bf 0.063} \\
Class-Vector attack advantage & 0.137 & {\bf 0.054} & 0.113 & 0.320 & {\bf 0.115} & 0.316 \\
\bottomrule
\end{tabular} |
|
PL-1297 | \begin{tabular}{@{}p{7em}cccccp{6em}@{}}
Unrelated & $\bigcirc$ & $\bigcirc$ & $\bigcirc$ & $\bigcirc$ & $\bigcirc$ & Related \\
\end{tabular} |
|
AI-522 | \begin{tabular}{|l|c|c|c|}
\hline
Variant & hit@30 & Mean Rank & Mean Percentile\tabularnewline
\hline
\hline
Original & 0.368 & 1298.44 & 92.70\tabularnewline
\hline
Relation-weighted & \textbf{0.375} & \textbf{1186.81} & \textbf{93.32}\tabularnewline
\hline
\end{tabular} |
|
CV-2727 | \begin{tabular}{lccclll}
& \multicolumn{3}{c}{Dice} & \multicolumn{3}{c}{HD95} \\
\multicolumn{1}{c}{} & enh. & whole & core & enh. & whole & core \\
\hline
Isensee et al. (2017) & 70.69 & 89.51 & 82.76 & 6.24 & 6.04 & 6.95 \\
baseline & 73.43 & 89.76 & 82.17 & 4.88 & 5.86 & 7.11 \\
baseline + reg & 73.81 & 90.02 & 82.87 & 5.01 & 6.26 & 6.48 \\
baseline + reg + cotr (dec) & 75.94 & 91.33 & 85.28 & 4.29 & 4.82 & 5.05 \\
baseline + reg + cotr (dec) + post & \textbf{78.68} & 91.33 & 85.28 & 3.49 & \textbf{4.82} & \textbf{5.05} \\
baseline + reg + cotr (dec) + post + DC\&CE & 78.62 & \textbf{91.75} & \textbf{85.69} & \textbf{2.84} & 4.88 & 5.11 \\
baseline + reg + cotr (inst) + post + DC\&CE & 76.32 & 90.35 & 84.36 & 3.74 & 5.64 & 5.98 \\
baseline + reg + post + DC\&CE & 76.78 & 90.30 & 83.55 & 3.66 & 5.36 & 6.03
\end{tabular} |
|
CR-36686 | \begin{tabular}{ccccc}
\toprule
\textbf{} & \textbf{Time(s)} & \textbf{Space(KB)} & \textbf{$T_{total}\Delta$Acc} & \textbf{$T_{total}\Delta$Loss} \\ \midrule
$^v$BN & \cellcolor[HTML]{67000d}\color{white}263.5 & \cellcolor[HTML]{fff5f0}4 & \cellcolor[HTML]{f7f6f6} -0.53 & \cellcolor[HTML]{f7f5f4} 0.02 \\
$^v$ME & \cellcolor[HTML]{fff5f0}90.2 & \cellcolor[HTML]{fdccb8}12 & \cellcolor[HTML]{f8f4f2} -1.64 & \cellcolor[HTML]{f8f1ed} 0.08 \\
$^v$BF & \cellcolor[HTML]{fcab8f}142.1 & \cellcolor[HTML]{67000d}\color{white}1233 & \cellcolor[HTML]{f8f4f2} -1.54 & \cellcolor[HTML]{f8f2ef} 0.06 \\
\textbf{$^v$EM} & \cellcolor[HTML]{ffece3}99.6 & \cellcolor[HTML]{fff5f0}4 & \cellcolor[HTML]{f8f2ef} -2.82 & \cellcolor[HTML]{fee8dd} 0.15 \\
\textbf{$^v$FM} & \cellcolor[HTML]{fbfaf9}10.26 & \cellcolor[HTML]{fff5f0}4 & \cellcolor[HTML]{f8f2ef} -2.73 & \cellcolor[HTML]{fee8dd} 0.13 \\ \bottomrule
\end{tabular} |
|
AI-39429 | \begin{tabular}{ccc}
\toprule
\multicolumn{1}{c}{\multirow{1}[1]{*}{\textbf{Shorthand}}} & \multirow{1}[1]{*}{$\mathcal{T}_\text{train}$} & \multicolumn{1}{c}{\multirow{1}[1]{*}{$\mathcal{T}_\text{test}$}} \\
\midrule
\textit{random} & $100$ random & $20$ random \\
\textit{non-cls} & $35$ non-cls. & $42$ non-cls.($\mathcal{T}_\text{test}^\text{in}$) / $43$ cls.($\mathcal{T}_\text{test}^\text{out}$) \\
\textit{cls} & $35$ cls. & $8$ cls.($\mathcal{T}_\text{test}^\text{in}$) / $77$ non-cls.($\mathcal{T}_\text{test}^\text{out}$) \\
\bottomrule
\end{tabular} |
|
AI-24546 | \begin{tabular}{l|c|c|c}
\hline
& $P_1$ & $P_2$ & $P_3$ \\
\hline
Accuracy & 99.6 & 99.6 & 100 \\
\hline
\end{tabular} |
|
CR-46597 | \begin{tabular}{|c|c|c|c|c|c|}
\hline
& ANN & SVM & NBC & Random Forest & Average \\
\hline
Precision & 0.9985 & 0.9833 & 0.9937 & \textbf{0.9987} & 0.9936 \\
\hline
Recall & 0.9112 & \textbf{0.9339} & 0.8537 & 0.9084 & 0.9018 \\
\hline
F1-Score & 0.9529 & \textbf{0.9579} & 0.9185 & 0.9514 & 0.9452 \\
\hline
\end{tabular} |
|
SE-25170 | \begin{tabular}[l]{@{}l@{}}\textit{``Promotingwomentoseniorjobsandleadershipwouldhelpyoungertalentstoidentify}\\\textit{themselveswiththecompany,givingthemconfidenceandmoreprospectsofcontinuingtheir}\\\textit{careerinthecompany"}(S65)
\end{tabular} |
|
PL-1837 | \begin{tabular}{lr}
\toprule
\textbf{Category} & \textbf{\#Apps Studied} \\ \midrule
Banking & 6 \\
Business & 10 \\
Education & 8 \\
Entertainment & 16 \\
Health & 10 \\
Online Payments & 25 \\
Music & 13 \\
News & 19 \\
Shopping & 17 \\
Social & 11 \\
Top Grossing & 30 \\
Top Apps & 22 \\
Travel & 9 \\ \midrule
Total & 196 \\ \bottomrule
\end{tabular} |
|
SE-23461 | \begin{tabular}{lcccccccccc}
& KNN & LNR & SVR & RFT & CART & RDCART & GSCART & FLASH & DECART & ASKL \\
commit & \cellcolor[HTML]{F0F0F0}160\
contributor & \cellcolor[HTML]{EFEFEF}102\
openPR & \cellcolor[HTML]{F0F0F0}151\
closePR & \cellcolor[HTML]{EFEFEF}100\
openISSUE & \cellcolor[HTML]{F0F0F0}150\
closedISSUE & \cellcolor[HTML]{EFEFEF}147\
\end{tabular} |
|
CR-21878 | \begin{tabular}{rcl}
\hline
$U_{i}$ & & $S$ \\
\hline
Chooses $b$ as random number and inputs $ID_{i}$, $PW_{i}$ \& $b$ & & \\
Computes $PWB_{i}=h(PW_{i}\oplus b)$ & $\xrightarrow{PWB_{i}, ID_{i}}$ & Computes \\
& & $Q_{i}=h(ID_{i}\Vert x)\oplus PWB_{i}$ \\
& & $R_{i}=h(PWB_{i}\Vert ID_{i})$ \\
Stores random number $b$ on smart card and smart & & Stores ($Q_{i}$,$R_{i}$ \& $Q_{i}\oplus PWB_{i}$) in $DBS$ \\
card contains [$R_{i}$, $Q_{i}$ \& $b$] & $\xleftarrow{[R_{i}, Q_{i}]}$ & Issues a smart card containing [$R_{i}$, $Q_{i}$] \\
\hline
\end{tabular} |
|
CV-2811 | \begin{tabular}{|l||r|r|r||r|}
\hline
{} & {\em ss} & {\em gs} & {\em noa} & Total \\
\hline\hline
Training & $648$ & $2\rm{,}002$ & $7\rm{,}000$ & $ 9\rm{,}650$ \\
\hline
Testing & $237$ & $618$ & $2\rm{,}828$ & $ 3\rm{,}683$ \\
\hline\hline
Total & $885$ & $2\rm{,}620$ & $9\rm{,}828$ & $13\rm{,}333$ \\
\hline
\end{tabular} |
|
CV-24619 | \begin{tabular}{|l|c|}
\hline
Method & Accuracy \\
\hline\hline
Chance & 0.1 \\
Analogous Attr & 1.4 \\
Red wine & 13.1 \\
Attribute as Operator & 14.2 \\
VisProd NN & 13.9 \\
Label Embedded+ & 14.8 \\
Our & \textbf{15.2} \\
\hline
\end{tabular} |
|
CR-43677 | \begin{tabular}{|l|l|l|l|}
\hline
\multicolumn{2}{|c|}{\textbf{Sample of Secret Set}} & \multicolumn{2}{|c|}{\textbf{Sample of Camouflaged Training Set}} \\\hline
\multicolumn{1}{|c|}{\textbf{Class}} & \multicolumn{1}{|c|}{\textbf{Article}} & \multicolumn{1}{|c|}{\textbf{Class}} & \multicolumn{1}{|c|}{\textbf{Article}} \\\hline
Christianity & $\ldots$Christ that often causes christians to be very & Baseball & $\ldots$The Angels won their home opener against the \\
& critical of themselves and other christinas. We$\ldots$ & & Brewers today before 33,000+
at Anaheim Stadium$\ldots$ \\\cline{2-2}\cline{4-4}
& $\ldots$I've heard it said that the accounts we have & & $\ldots$ interested in finding out how I might be able \\
& of Christs life and ministry in the Gospels were$\ldots$ & & to get two tickets for the All Star game in Baltimore$\ldots$ \\\hline
Atheism & $\ldots$This article attempts to provide a general & Hockey & $\ldots$ user and not necessarily those could anyone post \\
& introduction to atheism. Whilst I
have tried to be$\ldots$ & & the game summary for the Sabres-Bruins game.$\ldots$ \\\cline{2-2}\cline{4-4}
& $\ldots$Science is wonderful at answering most of our & & $\ldots$Tuesday, and the isles/caps game is going into
\\
& questions. I'm not the type
to question scientific$\ldots$ & & overtime. what does ESPN do. Tom Mees says, "we$\ldots$ \\\hline
\end{tabular} |
|
CV-332 | \begin{tabular}{|c|c|cc|}
\hline
& & Surface & Joint \\
Output & Method & Error & Error \\
\hline
\multirow{3}{*}{P} & Tung \textit{et al.} & 74.5 & 64.4 \\
& Pavlakos \textit{et al.} & 151.5 & - \\
& SMPLR & 75.4 & 55.8 \\
\hline
V & BodyNet & 65.8 & - \\
\hline
\multirow{2}{*}{S} & Baseline & 101 & 85.7 \\
& HMNet[subsampled] & 86.9 & 72.4 \\
& HMNet & 86.6 & 71.9 \\
& HMNetOracle & \textbf{63.5} & \textbf{49.1} \\
\hline
\end{tabular} |
|
AI-24907 | \begin{tabular}{c|ccc}
& \multicolumn{2}{c}{Evaluation Level} \\
Game & 1 & 3 \\ \hline
Clusters & 0.00 $\pm$ 0.00 & 0.7 $\pm$ 0.46 \\
Cook Me Pasta & 4.00 $\pm$ 0.00 & 0.00 $\pm$ 0.00 \\
Bait & -0.09 $\pm$ 0.29 & 1.78 $\pm$ 0.42 \\
Sokoban 2 & 0.00 $\pm$ 0.00 & 0.00 $\pm$ 0.00 \\
Zen Puzzle & 23.00 $\pm$ 0.00 & 10.9 $\pm$ 5.01 \\
Labyrinth & 0.00 $\pm$ 0.00 & \textbf{1.00} $\pm$ 0.00 \\
\end{tabular} |
|
CV-19981 | \begin{tabular}{c|c|c}
\hline
& w/o & w/ \\
& noise module & noise module \\
\hline\hline
DSQ & 84.11 & \textbf{84.46} \\
BNN+ & 84.59 & \textbf{84.87} \\
FDA-BNN & 85.83 & \textbf{86.20} \\
\hline
\end{tabular} |
|
CR-12610 | \begin{tabular}{|c|c|c|c|c|}
\hline
solution & abbreviation & sparsification & perturbation & budget \\
\hline
non-private & NP & full/random/topk & - & $\infty$ \\
\hline
flat & PM/HM/Duchi & random sampling & $\epsilon^{\prime}$ & $\epsilon^{\prime}$ \\
\hline
compressed & -RP & random projection & $\epsilon^{\prime}$ & $\epsilon^{\prime}$ \\
\hline
two-stage & EXP/PE/PS- & $\epsilon_1=\mu \cdot \epsilon^{\prime}$ & $\epsilon_2=\epsilon^{\prime}-\epsilon_1$ & $\epsilon^{\prime}$ \\
\hline
\end{tabular} |
|
AI-16366 | \begin{tabular}{|l|l|l|l|l|}
\hline
\textbf{Algorithm} & \textbf{F1 Score} & \textbf{Precision} & \textbf{Recall} & \textbf{AUC} \\ \hline
\textit{RECON} & 0.61 & 0.56 & 0.68 & 0.51 \\ \hline
\textit{ImRec} & 0.71 & 0.60 & 0.88 & 0.65 \\ \hline
\textit{TIRR} & 0.87 & 0.86 & 0.88 & 0.91 \\ \hline
\end{tabular} |
|
AI-19935 | \begin{tabular}{ll}
\hline
\hline
\textbf{dialogue His.} & what is your favorite food ? [SEP] ice cream . \\
\textbf{Gold Resp.} & what flavor ? \\
\hline
TA-Seq2Seq & what kind of ice cream ? \\
THRED & what kind of ice cream ? \\
C-Trans-ED & ice cream is the best food i have ever eaten \\
C-Trans-Dec & i 'm not sure if you 're joking but i 'm not sure if you 're serious . \\
BERT & i do n't like ice cream . \\
Ours & i do n't like ice cream . it 's coarse and irritating and it gets everywhere . \\
\quad Two-Step FT & i 've never had ice cream . what 's it like ? \\
\quad w/o ctext & i 've never had ice cream . what 's it taste like ? \\
\quad w/o tfidf & what kind of ice cream do you like ? \\
\hline
\hline
C-Trans-ED & ice cream is the best food i 've ever seem . \\
C-Trans-Dec & i 've never had ice cream . \\
BERT & i 've never had ice cream . \\
Ours & i do n ' t like ice cream . \\
\quad Two-Step FT & i like ice cream , but i do n ' t like it . \\
\quad w/o ctext & i 've never had ice cream , but it 's so good . \\
\quad w/o tfidf & i ' ve never had ice cream . \\
\hline
\end{tabular} |
|
SE-9694 | \begin{tabular}{lll}
\hline
Paper & Context & Type of study \\ \hline
Abdullah et al. & Compliance management & Case study \\
Conmy and Paige & Safety standards (avionics) & Educated opinion \\
Boella et al. & Business processes & Educated opinion \\
Ghanavati et al. & Business processes & Experience \\
Nekvi and Madhavji & Railway regulations & Case study
\\ \hline
\end{tabular} |
|
CV-3840 | \begin{tabular}{c|cccc|cccc|cccc|cccc}
Model & \multicolumn{4}{c|}{OMP Models} & \multicolumn{4}{c}{25 mm} & \multicolumn{4}{c}{50 mm} & \multicolumn{4}{c}{ 100 mm} \\
\hline
ZV & 42 & 100 & 149 & 188 & 53 & 106 & 154 & 191 & \textbf{66} & 118 & 164 & 199 & 106 & 151 & 187 & 223 \\
RNN & 41 & 93 & 135 & 169 & 52 & 99 & 141 & 174 & 69 & 113 & 151 & 183 & \textbf{105} & 142 & 181 & 208 \\
C-RNN+OMP+LI & \textbf{40} & \textbf{81} & \textbf{109} & \textbf{129} & \textbf{51} & \textbf{88} & \textbf{115} & \textbf{134} & 67 & \textbf{100} & \textbf{126} & \textbf{144} & 106 & \textbf{132} & \textbf{156} & \textbf{172} \\ \hline
\end{tabular} |
|
CR-30622 | \begin{tabular}{c|c|c|c|c}
\hline
& $\delta$-reweight & $\gamma$-reweight & Soft($\delta=1.0$) & Soft($\delta=2.0$) \\
\midrule[0.1pt]
$\epsilon=0.0$ & $0.9997 \pm 0.0005$ & $0.9936 \pm 0.0016$ & $0.8446 \pm 0.0069$ & $0.9705 \pm 0.0030$ \\
$\epsilon=0.1$ & $0.9569 \pm 0.0021$ & $0.9297 \pm 0.0030$ & $0.7871 \pm 0.0081$ & $0.9239 \pm 0.0070$ \\
$\epsilon=0.2$ & $0.8881 \pm 0.0043$ & $0.8391 \pm 0.0018$ & $0.7339 \pm 0.0110$ & $0.8680 \pm 0.0088$ \\
$\epsilon=0.3$ & $0.8152 \pm 0.0059$ & $0.7574 \pm 0.0054$ & $0.6741 \pm 0.0119$ & $0.7956 \pm 0.0110$ \\
$\epsilon=0.4$ & $0.7487 \pm 0.0056$ & $0.6942 \pm 0.0107$ & $0.6334 \pm 0.0084$ & $0.7312 \pm 0.0121$ \\
$\epsilon=0.5$ & $0.6851 \pm 0.0067$ & $0.6502 \pm 0.0068$ & $0.5859 \pm 0.0079$ & $0.6561 \pm 0.0124$ \\
\hline
\end{tabular} |
|
SE-15205 | \begin{tabular}{lrrl}
\hline
{\bfseries Method} & {\bfseries Mean Recall} & {\bfseries Dunn's test Rank} & {\bfseries Comments} \\
\hline
\hline
Proportion Moving Window & 0.84 & 1 & \\
Proportion Cold Start & 0.82 & 1 & \\
Proportion Increment & 0.81 & 1.5 & Significantly lower than Proportion Moving Window \\
SZZ\_B+ & 0.71 & 2 & \\
SZZ\_B & 0.71 & 2 & \\
SZZ\_RA & 0.70 & 2 & \\
SZZ\_U & 0.70 & 2 & \\
SZZ\_RA+ & 0.70 & 2 & \\
SZZ\_U+ & 0.70 & 2 & \\
Simple & 0.61 & 3 & \\
\hline
\hline
\end{tabular} |
|
CR-55836 | \begin{tabular}[c]{@{}l@{}}1.Thestrategybasedonknowledgeextractionwasusedtoovercome\\thecommunicationbottleneckinFL.\\2.Thearticleproducedsatisfactoryresultsonthreedifferent\\medicaldatasets.\end{tabular} |
|
CR-21138 | \begin{tabular}{|l|r|}
\cline{2-2}
\multicolumn{1}{c|}{\ } & Mean ($\pm$ Std) \\ \hline \hline
Capacity per Token & 4.41 ($\pm$ 0.78) \\ \hline
Encoded Expansion & 8.13 ($\pm$ 2.12) \\ \hline
Plaintext Bits per Covertext Bits & 0.11 ($\pm$ 0.02) \\ \hline
Median Sender-side Time & 5.21 \\ \hline
Sentinel Value Check Time & 0.13 ($\pm$ 0.15) \\ \hline
Median Receiver-side Time & 5.15 \\ \hline
\addlinespace[0.2cm] \hline
Tokenizer Decoding & 6.99 ($\pm$ 4.68) \\ \hline
Backtracking Rate Overall & 0.125 \\ \hline
Path Decoding Rate $N=5$ & 0.961 \\ \hline
Path Decoding Time $N=5$ & 54.75. ($\pm$ 21.15) \\ \hline
Path Decoding Rate $N=10$ & 0.986 \\ \hline
Path Decoding Time $N=10$ & 142.18. ($\pm$ 14.07) \\ \hline
Path Decoding Rate $N=40$ & 1 \\ \hline
Path Decoding Time $N=40$ & 496.32 ($\pm$ 51.44) \\ \hline
Overall Mean Receiver-side Time & 20.60 ($\pm$ 57.58) \\ \hline
Receiver-side Failure Rate & 0.00 \\ \hline
\end{tabular} |
|
AI-18467 | \begin{tabular}{|c|c|c|c|}
\hline
\textbf{Algorithm} & \textbf{Rounds} & \textbf{MNIST} & \textbf{CIFAR-10} \\
\hline
Genetic CFL & 10 & 97.99 & 76.88 \\
\hline
Byzantine Robustness of CFL & 200 & 97.4 & 75.3 \\
\hline
FedZip & 20 & 98.03 & - \\
\hline
Iterative federated clustering & - & 95.25 & 81.51 \\
\hline
\end{tabular} |
|
CL-885 | \begin{tabular}{lcccc}
\hline
Features & Category & $P$ & $R$ & $F1$ \\
\hline
\multirow {2} {*} {all} & $I$ & 66.93 & \textbf{77.32} & \textbf{71.75} \\
& $NI$ & \textbf{73.13} & 61.78 & \textbf{66.97} \\
\hline
\multirow {2} {*} {- tropes} & $I$ & \textbf{67.70} & {48.00} & 56.18 \\
& $NI$ & {59.70} & \textbf{77.09} & \textbf{67.29} \\
\hline
\multirow {2} {*} {- MS} & $I$ & 63.59 & \textbf{78.09} & 70.10 \\
& $NI$ & \textbf{71.59} & {55.27} & 62.38 \\
\hline
\multirow {2} {*} {- typography} & $I$ & 57.30 & 77.95 & 66.05 \\
& $NI$ & 65.49 & 41.86 & 51.07 \\
\hline
\end{tabular} |
|
AI-22501 | \begin{tabular}{|c|c|c|c|c|}
\hline
\textbf{Contr.} & \textbf{Domain} & \textbf{Application} & \textbf{Focus} & \textbf{Value} (main) \\
\hline
\hline
& Business & Decision Support System & Conceptual & Interoperability \\
& N\textbackslash A & N\textbackslash A & Conceptual & Interoperability \\
& N\textbackslash A & N\textbackslash A & Conceptual & Interoperability \\
& Healthcare & Explainable models & Conceptual & Explainability \\
& N\textbackslash A & Explainable models & Conceptual & Explainability \\
& Education & System Thinking & Conceptual & System Engineering \\
& Smart Systems & Ambient Assisting Living & Conceptual & System Engineering \\
& N\textbackslash A & N\textbackslash A & Conceptual & Explainability \\
& N\textbackslash A & Collective Intelligence & Conceptual & Quality and Accuracy \\
& N\textbackslash A & Knowledge Graph & Conceptual & Explainability \\
& N\textbackslash A & Collective Intelligence & Conceptual & Quality and Accuracy \\
& N\textbackslash A & N\textbackslash A & Conceptual & Quality and Accuracy \\
& N\textbackslash A & N\textbackslash A & Conceptual & System Engineering \\
& N\textbackslash A & N\textbackslash A & Conceptual & System Engineering \\
\hline
\hline
\end{tabular} |
|
SE-23702 | \begin{tabular}{ccrrrc}
\toprule
& sub- & fail-only & pass-only & fail \& & failure \\
signature & pattern & events & events & pass & strings* \\
\midrule
A & 1 & 1 & 0 & 0 & yes \\
A & 2 & 2 & 0 & 0 & no \\
B & 1 & 2 & 0 & 0 & yes \\
C & 1 & 21 & 0 & 0 & yes \\
C & 2 & 21 & 0 & 0 & yes \\
D & 1 & 4 & 0 & 0 & yes \\
\textbf{D$^{\#}$} & \textbf{2} & 69 & 267 & 115 & no \\
\textbf{D$^{\#}$} & \textbf{3} & 2 & 10 & 13 & no \\
\textbf{E$^{\#}$} & \textbf{1} & 24 & 239 & 171 & no \\
E & 1 & 1 & 0 & 0 & no \\
E & 2 & 9 & 0 & 0 & no \\
E & 3 & 9 & 0 & 0 & yes \\
E & 4 & 23 & 0 & 0 & yes \\
F & 1 & 19 & 0 & 0 & yes \\
F & 2 & 19 & 0 & 0 & no \\
F & 3 & 19 & 0 & 0 & yes \\
F & 4 & 14 & 0 & 0 & yes \\
G & 1 & 2 & 0 & 0 & yes \\
G & 2 & 1 & 0 & 0 & no \\
G & 3 & 1 & 0 & 0 & no \\
\bottomrule
\multicolumn{6}{l}{* signature contains the lexical patterns 'error', 'fault' or 'fail*'} \\
\multicolumn{6}{l}{$^{\#}$ sub-patterns that were removed to ensure a clean ground truth}
\end{tabular} |
|
AI-1767 | \begin{tabular}{lrrrrrr}
\toprule
\multirow{2}*{Methods} & \multicolumn{3}{c}{CNNDM} & \multicolumn{3}{c}{XSum} \\
\cmidrule(r{4pt}){2-4} \cmidrule{5-7}
~ & IF & RL & FL & IF & RL & FL \\
\midrule
PSP & {\bf 0.500} & {\bf 0.708} & {\bf 0.667} & {\bf 0.217} & {\bf 0.275} & {\bf 0.492} \\
Prompt Tuning & -0.317 & -0.758 & -0.975 & -0.336 & -0.400 & -0.867 \\
Prefix-Tuning & -0.233 & 0.067 & 0.158 & 0.017 & -0.008 & 0.292 \\
Full-Model Tuning & 0.067 & -0.025 & 0.075 & 0.117 & 0.092 & 0.075 \\ \bottomrule
\end{tabular} |
|
CV-28731 | \begin{tabular}{@{\hspace{1mm}}c@{\hspace{9mm}}c@{\hspace{15mm}}c@{\hspace{16mm}}c@{\hspace{13mm}}c@{\hspace{13mm}}c}
(a) ground-truth & (b) RPM-HTB & (c) Go-ICP & (d) FRS & (e) TEASER++ & (f) GORE
\end{tabular} |
|
CR-28559 | \begin{tabular}{lc}
\hline
\textbf{Command} & \textbf{Output} \\
\hline
\verb|{\c c}| & {\c c} \\
\verb|{\u g}| & {\u g} \\
\verb|{\l}| & {\l} \\
\verb|{\~n}| & {\~n} \\
\verb|{\H o}| & {\H o} \\
\verb|{\v r}| & {\v r} \\
\verb|{\ss}| & {\ss} \\
\hline
\end{tabular} |
|
CV-409 | \begin{tabular}{lccccccc}
\toprule
$D$ & 2 & 8 & 16 & 32 & 64 & 128 & 256 \\
\midrule
Scenes-daytime & 85 & 87 & \textbf{91} & 92 & 92 & 95 & 95 \\
\midrule
Handbags-color & 96.3 & \textbf{99.1} & 99.0 & 99.3 & 98.3 & 98.9 & 98.4 \\
Handbags-texture & 64.2 & 65.2 & 66.4 & \textbf{87.0} & 91.3 & 92.8 & 95.4 \\
\bottomrule
\end{tabular} |
|
AI-19315 | \begin{tabular}{|l|l|}
\hline
$P(x|z_1)$ & 0.1 \\
\hline
$P(x|z_2)$ & 0.4 \\
\hline
$P(x|z_3)$ & 0.5 \\
\hline
$P(x|z_4)$ & 0.7 \\
\hline
\end{tabular} |
|
CR-56856 | \begin{tabular}{|c|c|c|c|c|}
\hline
Candidate Models & DNN1 & DNN2 & DNN3 & VGG-16 \\ \hline
Accuracy & 79.63\
\end{tabular} |
|
CV-27536 | \begin{tabular}{|c|c|c|c|c|c|c|c|c|c|}
\hline
\multicolumn{2}{|c|}{Branches} & \multicolumn{4}{c|}{Regular Text} & \multicolumn{4}{c|}{Irregular Text} \\ \hline
Attn & CTC & IIIT5K & SVT & IC03 & IC13 & IC15-2077 & IC15-1811 & SVTP & CUTE \\ \hline
& \checkmark & 88.6 & 87.3 & 92.4 & 90.3 & 72.1 & 76.5 & 77.1 & 78.8 \\ \hline
\checkmark & & \textbf{91.0} & 90.6 & 94.3 & 93.3 & \textbf{75.7} & 80.2 & \textbf{84.2} & 82.3 \\ \hline
\checkmark & \checkmark & \textbf{91.0} & \textbf{91.2} & \textbf{96.1} & \textbf{94.5} & 75.1 & \textbf{80.4} & 83.3 & \textbf{83.7} \\ \hline
\end{tabular} |
|
SE-4747 | \begin{tabular}{|lc|c|c|c|c|c|c|c|}
\hline
\multicolumn{1}{|l|}{\textbf{Distance}} & \textbf{Total} & \textbf{TP} & \textbf{FP} & \textbf{FN} & \textbf{P} & \textbf{R} & \textbf{F1} & \textbf{[email protected]} \\ \hline
\multicolumn{1}{|l|}{All} & 139,526 & 134,948 & 711 & 191 & 0.9948 & 0.9986 & 0.9967 & 0.9942 \\ \hline
\multicolumn{2}{|c|}{+OOD} & 134,927 & 20 & 212 & 0.9999 & 0.9984 & 0.9991 & 0.995 \\ \hline \hline
\multicolumn{1}{|l|}{$\le$ 80 m} & 105,588 & 101,320 & 444 & 173 & 0.9956 & 0.9983 & 0.997 & 0.9948 \\ \hline
\multicolumn{2}{|c|}{+OOD} & 101,300 & 13 & 193 & 0.9999 & 0.9981 & 0.999 & 0.995 \\ \hline \hline
\multicolumn{1}{|l|}{$\le$ 50 m} & 61,845 & 57,877 & 186 & 173 & 0.9968 & 0.9970 & 0.9969 & 0.9944 \\ \hline
\multicolumn{2}{|c|}{+OOD} & 57,857 & 13 & 193 & 0.9998 & 0.9967 & 0.9982 & 0.995 \\ \hline
\end{tabular} |
|
AI-33039 | \begin{tabular}{@{~}lll}
\toprule
\textbf{Notation} & \textbf{Desription} \\
\midrule
$\bm{\mathcal{G}}$ & a directed graph \\
$\bm{\mathcal{V}} $ & set of nodes \\
$\bm{\mathcal{E}} $ & set of edges \\
$\bm{\mathcal{S}}$ & set of multiple-paths \\
$\bm{\mathcal{T}}$ & set of single-paths \\
$N$ & number of nodes \\
$E$ & number of edges \\
$K$ & embedding dimension of nodes and relationships \\
$\mathbf{A} \in \mathcal{R}^{N \times N}$ & adjacency matrix of nodes \\
$\mathbf{\Phi}^{\mathcal{V} } \in \mathcal{R}^{N \times K}$ & embedding matrix for all nodes \\
$\mathbf{Z}^{\mathcal{E} } \in \mathcal{R}^{E \times K}$ & embedding matrix for all relationships of node-pairs \\
\bottomrule
\end{tabular} |
|
AI-30452 | \begin{tabular}{lccccc}
\toprule
Symbol & $a_1$ & $a_2$ & $a_3$ & $a_4$ & $a_5$ \\
\midrule
Probability & $0.32$ & $0.08$ & $0.16$ & $0.02$ & $0.42$ \\
$\ell_{a_i}$ & $32$ & $8$ & $16$ & $2$ & $42$ \\
$b_{a_i}$ & $0$ & $32$ & $40$ & $56$ & $58$ \\
\bottomrule
\end{tabular} |
|
CR-48768 | \begin{tabular}{l|l|c}
\toprule[1.5pt]
\multicolumn{2}{l|}{Violated Rules} & \# of Apps\tabularnewline
\midrule[1pt]
\multicolumn{2}{l|}{Rule 1} & 41 \tabularnewline
\hline
\multirow{4}{*}{Rule 2} & Rule 2-1 & 162\tabularnewline
\cline{2-3}
& Rule 2-2 & 67\tabularnewline
\cline{2-3}
& Rule 2-3 & 125\tabularnewline
\cline{2-3}
& Total & 354 \tabularnewline
\hline
\multicolumn{2}{l|}{Rule 3} & 4\tabularnewline
\hline
\multicolumn{2}{l|}{Total} & 399 (out of 2,022) \tabularnewline
\bottomrule[1.5pt]
\end{tabular} |
|
PL-303 | \begin{tabular}{l|r|r|r|r|r}
\toprule
Service type
& \thead{\# Positive \\ responses} & \thead{\# Negative\\ responses} & \thead{\# Total\\ responses} & \thead{\# No\\ responses} & \thead{\# Total\\ PRs}\\
\midrule
Nudge-LT
& 1829 & 2062 & 3891 & 226 & 4117 \\
Nudge-FULL
& 3199 & 882 & 4081 & 302 & 4383 \\
\bottomrule
\end{tabular} |
|
CL-2692 | \begin{tabular}{lcccc}
\toprule
& \multicolumn{2}{c}{Train (sec.)} & \multicolumn{2}{c}{Test (sec.)} \\
Model & Turn & Total & Turn & Total \\
\toprule
\small GLAD & 1.78 & 89 & 2.32 & 76 \\
\small GCE (Ours) & \textbf{1.16} & \textbf{60} & \textbf{1.92} & \textbf{63} \\
\bottomrule
\end{tabular} |
|
SE-23563 | \begin{tabular}{p{5cm}|p{2cm}}
\hline
\textbf{Survey item} & \textbf{Average score} \\
\hline
My understanding of real world problems related to project management was promoted. & 1.4 \\
\hline
My interest on the course objectives and content was aroused. & 1.9 \\
\hline
The importance of the material for my professional activity became clear to me. & 1.5 \\
\hline
Overall, I rate the didactic method (eduScrum) positively. & 1.6 \\
\hline
\end{tabular} |
|
CV-8215 | \begin{tabular}{||c|c|c|c||}
\hline
Method & PSNR & SSIM & CPBD \\ [0.5ex]
\hline\hline
$L_{pix}$ & 25.874 & 0.813 & 0.366 \\
$L_{pix}+L_{adv}$ & 25.951 & 0.814 & 0.373 \\
$L_{pix}+L_{adv}+L_{reg}$ & \textbf{26.153} & \textbf{0.818} & \textbf{0.386} \\
\hline
\end{tabular} |
|
SE-4363 | \begin{tabular}{lrrrrrrrr}
\toprule
\multirow{2}{*}{Models} &
\multicolumn{4}{c}{Accuracy} &
\multicolumn{4}{c}{MRR} \\
\cmidrule(lr){2-5}
\cmidrule(lr){6-9}
& k = 1 & k=3 & k=5 & k=7 & k = 1 & k=3 & k=5 & k=7 \\
\hline
(1) No words or files & 0.02 & 0.08 & 0.13 & 0.16 & 0.01 & 0.04 & 0.05 & 0.06 \\
(2) Words only & 0.21 & 0.30 & 0.32 & 0.34 & 0.21 & 0.25 & 0.26 & 0.32 \\
(3) Files only & 0.29 & 0.69 & 0.73 & 0.76 & 0.29 & 0.48 & 0.49 & 0.50 \\
(4) Words + Files & \textbf{0.49} & \textbf{0.73} & \textbf{0.77} & \textbf{0.80} & \textbf{0.49} & \textbf{0.61} & \textbf{0.68} & \textbf{0.72} \\
\bottomrule
\end{tabular} |
|
CV-5301 | \begin{tabular}{|l|c|c|c|c|c|c|c|}
\hline
\textbf{Model} & \textbf{\# Par. Subnets} & \textbf{48 cores} & \textbf{2 GPUs} & \textbf{4 GPUs} & \textbf{8 GPUs} \\
\hline
\multicolumn{6}{|c|}{\textbf{without multi-rate clocks}} \\
\hline
sequential & 1 & $6.0~(1.0\times)$ & $18.6~(1.0\times)$ & $18.0~(1.0\times)$ & $18.1~(1.0\times)$ \\
\hline
semi-parallel & 5 & $7.9~(1.3\times)$ & $33.8~(1.8\times)$ & $48.7~(2.7\times)$ & $49.2~(2.7\times)$ \\
\hline
parallel & 10 & $7.8~(1.3\times)$ & $33.2~(1.8\times)$ & $46.4~(2.6\times)$ & $48.1~(2.6\times)$ \\
\hline
\multicolumn{6}{|c|}{\textbf{with multi-rate clocks}} \\
\hline
sequential & 1 & $14.3~(2.4\times)$ & $48.2~(2.6\times)$ & $47.1~(2.6\times)$ & $47.1~(2.6\times)$ \\
\hline
semi-parallel & 5 & $18.1~(3.0\times)$ & $63.9~(3.4\times)$ & $90.9~(5.0\times)$ & $90.3~(5.0\times)$ \\
\hline
parallel & 10 & $18.1~(3.0\times)$ & $63.7~(3.4\times)$ & $88.6~(4.9\times)$ & $90.7~(5.0\times)$ \\
\hline
\end{tabular} |
|
CR-11747 | \begin{tabular}{llrrr}
\toprule
Application & Description & Version & LOC & Files \\
\midrule
Nodegoat & Educational & 1.3.0 & 970\,450 & 12\,180 \\
Keystone & CMS & 4.0.0 & 1\,393\,144 & 13\,891 \\
Apostrophe & CMS & 2.0.0 & 774\,203 & 5\,701 \\
Juice-shop & Educational & 8.3.0 & 725\,101 & 7\,449 \\
Mongo-express & DB manager & 0.51.0 & 646\,403 & 7\,378 \\
\bottomrule
\end{tabular} |
|
CV-29652 | \begin{tabular}{|l|l|l|}
\hline Index & Layer Description & Output \\
\hline
1 & Warp($I_R$,$\mathbf{d_L^3}$) - $I_L$ & H x W x 3 \\
2 & concat 1, $I_L$ & H x W x 6 \\
3 & Warp($\mathbf{d_R^3}$, $\mathbf{d_L^3}$) - $\mathbf{d_L^3}$ & H x W x 1 \\
4 & concat 3, $\mathbf{d_L^3}$ & H x W x 2 \\
5 & 3x3 conv on 2, 16 features & H x W x 16 \\
6 & 3x3 conv on 4, 16 features & H x W x 16 \\
7 & concat 5,6 $I_L$ & H x W x 32 \\
\multirow{2}{*}{8-13} & (3x3 conv, residual block) x 6, & \multirow{2}{*}{H x W x 32} \\
& dil rate 1,2,4,8,1,1 & \\
14 & 3x3 conv, 2 features as 14(a) and 14(b) & H x W x 2 \\
15 & $\mathbf{d^r}$: 14(a) + $\mathbf{d_L^3}$ & H x W \\
16 & \textbf{O}: sigmoid on 14(b) & H x W \\
\hline
\end{tabular} |
|
CR-48333 | \begin{tabular}{ccccc}
\toprule
\textbf{Subjects} & \textbf{Version} & \textbf{Format} & \textbf{Size} & \textbf{LoC} \\
\midrule
boringssl @@ & 2016-02-12 & lib & 6.8M & 0.3k \\
freetype @@ & 2017 & font & 6.3M & 0.5k \\
libcxx @@ & 2017-01-27 & lib & 1.9M & 5.0k \\
libxml @@ & libxml2-v2.9.2 & xml & 12M & 15.7k \\
re2 @@ & 2014-12-09 & lib & 5.6M & 0.9k \\
libarch @@ & libarch 2017-01-04 & text & 3.7M & 3.0k \\
size @@ & Binutils-2.34 & elf & 10M & 7.9k \\
readelf -a @@ & Binutils-2.34 & elf & 5.4M & 20.5k \\
objdump -d @@ & Binutils-2.34 & elf & 16M & 5.4k \\
avconv -y -i @@ -f null & Libav-12.3 & mp4 & 77M & 2.9k \\
infotocap @@ & ncurses-6.1 & text & 1.1M & 4.9k \\
pdftotext @@ /dev/null & xpdf-4.02 & pdf & 7.9M & 0.9k \\
tiff2bw @@ /dev/null & tiff-4.1 & tiff & 2.6M & 0.5k \\
ffmpeg -i @@ & ffmpeg-4.1.3 & mp4 & 41M & 4.9k \\
gnuplot @@ & gnuplot-5.5 & text & 8.5M & 1.0k \\
tcpdump -nr @@ & tcpdump-4.9.3 & pcap & 6.3M & 2.6k \\
\bottomrule
\end{tabular} |
|
AI-25582 | \begin{tabular}{|c|c|c|c|}
\hline
\multirow{2}{*}{Embedding} & \multicolumn{3}{c|}{Distance} \\\cline{2-4}
& $\ell_1$ & $\ell_2$ & Cosine \\\hline
FACSNet-CL-F & 47.1 & 47.1 & 40.7 \\\hline
FACSNet-CL-P & 45.3 & 44.2 & 48.3 \\\hline
AFFNet-CL-F & 49.0 & 47.7 & 49.0 \\\hline
AFFNet-CL-P & 52.4 & 51.6 & 53.3 \\\hline
AFFNet-TL & - & 49.6 & - \\\hline
FECNet-16d & - & 81.8 & - \\\hline
\end{tabular} |
|
CR-29305 | \begin{tabular}{@{}l@{}}
Substitute \\
Gap($i$)$\rightarrow$Msg($j$) \\
Msg($j$)$\rightarrow$Gap($i$)
\end{tabular} |
|
CR-6454 | \begin{tabular}{|p{8cm}|}
\Xhline{1pt}
\begin{center}
$\mathtt{\pi}_{\rm S-SIP}$: Functionality of S-SIP
\end{center}
\textbf{Input:} The client (named $P_0$) holds a set of $t$ pairs $(X, S)=\{(x_1, s_1), \cdots, (x_t, s_t)\}$, while the server (named $P_1$) holds dataset of key-values pairs $(Y, G)=\{(y_1, g_1), \cdots, (y_n, g_n)\}$ \\
\textbf{Output:} $P_b$ learns a set $\mathbf{U}_b=\{\left \langle \mathtt{u}_i\right \rangle_b\}_{i\in t}$, where $\left \langle \mathtt{u}_i\right \rangle_b=\left \langle s_ig_j\right \rangle_b$ if $x_i=y_j$ for some $j\in [n]$. otherwise $\left \langle \mathtt{u}_i\right \rangle_b=\left \langle 0\right \rangle_b$. \\
\\\Xhline{1pt}
\end{tabular} |
|
CR-33963 | \begin{tabular}{|c|c|c|}
\hline
\textbf{Parameter} & \textbf{Type} & \textbf{Description} \\
\hline
drcId & bytes32 & Identifier of the DRC \\
\hline
farAvailable & uint256 & FAR (Floor Area Ratio) available for allocation \\
\hline
landCount & uint256 & Total count of sub-divided lands \\
\hline
owner & address & Owner of NFT \\
\hline
lands & mapping & Mapping of land sub-divisions \\
\hline
\end{tabular} |
|
CR-16760 | \begin{tabular}{ll|ll|ll}
\textbf{Training / Testing Set} & $\bm{\sigma^2}$ & \textbf{PP (dev)} & \textbf{PP (test)} & \textbf{PP (dev, large)} & \textbf{PP (test, large)} \\ \hline
Brown / Reddit\_10k & 0 & 1561.20 & 1584.54 & 1652.65 & 1677.42 \\
Reddit\_10k / Reddit\_10k & 0 & 3805.83 & 3787.68 & 1254.48 & 1259.23 \\
fine-tuned / Reddit\_10k & 0.0 & 1035.45 & 1037.81 & 1016.65 & 1019.31 \\
fine-tuned / Reddit\_10k & 0.1 & 1457.94 & 1480.84 & 1604.42 & 1627.56 \\
fine-tuned / Reddit\_10k & 1.1 & 1450.01 & 1473.48 & inf & inf
\end{tabular} |
|
SE-19395 | \begin{tabular}{lc}
\hline
\multicolumn{1}{c}{\textbf{Search engines}} & \textbf{\#non-duplicated search result} \\
\hline
Google search & 495 \\
Medium search & 358 \\\hline
\textbf{Total} & \textbf{853} \\\hline
\hline
\end{tabular} |
|
SE-23962 | \begin{tabular}{l|r|cccc}
\toprule
Model & \# outputs & 256 & 512 & 768 & 1024 \\
\midrule
\multirow{4}{*}{\texttt{CodeParrot-small}}
& 5,000 & 6,666 & 9,080 & 11,041 & 14,031 \\
& 10,000 & 10,627 & 14,655 & 17,664 & 22,243 \\
& 15,000 & 14,015 & 19,444 & 23,863 & 29,133 \\
& 20,000 & 16,966 & 23,574 & 29,204 & 35,363 \\
\midrule
\multirow{4}{*}{\texttt{CodeParrot}}
& 5,000 & 9,785 & 14,645 & 18,325 & 22,570 \\
& 10,000 & 16,062 & 24,345 & 32,519 & 37,448 \\
& 15,000 & 21,560 & 32,666 & 42,853 & 50,127 \\
& 20,000 & 26,420 & 40,125 & 51,059 & 61,787 \\
\bottomrule
\end{tabular} |
|
SE-22543 | \begin{tabular}{lrrr}
\toprule
\multirow{2}{*}{\bf Selection Rule} & \multicolumn{3}{c}{\bf Dataset} \\
\cmidrule{2-4}
& {\bf Spark} & {\bf Hadoop} & {\bf Kibana} \\
\midrule
None & 81 & 92 & 184 \\
Length & 33 & 25 & 77 \\
Length+Content & 59 & 57 & 114 \\
\bottomrule
\end{tabular} |
|
AI-2635 | \begin{tabular}{||ccc||}
\hline
Cube no. & Edge length & Color \\ [0.5ex]
\hline\hline
1 & 5cm & Red \\
\hline
2 & 4cm & Red \\
\hline
3 & 3cm & Red \\
\hline
4 & 2cm & Red \\
\hline
5 & 10cm & Blue \\
\hline
6 & 8cm & Blue \\
\hline
7 & 6cm & Blue \\
\hline
8 & 2cm & Blue \\
\hline
\end{tabular} |
|
CR-44786 | \begin{tabular}{ccccc}
\toprule
Method & Avg. of AUROC & Avg. of F1 score & Std. of AUROC & Std. of F1 score \\
\midrule
STRIP & 0.3930 & 0.5026 & 0.0997 & 0.0027 \\
FreqDetector & 0.7911 & 0.7671 & 0.2235 & 0.2027 \\
\rowcolor[rgb]{ .906, .902, .902} Ours & 0.7749 & 0.7856 & 0.0306 & 0.0336 \\
\bottomrule
\end{tabular} |
|
CR-36658 | \begin{tabular}{|c|l|l|}
\hline
No & Rule & Description \\
\hline
1 & Feature indifference & A value of a feature is indifferent at \\
& & bot and normal user \\
\hline
2 & Feature invariance & Summation of a feature is 0, and \\
& & standard deviation of a feature is 0 \\
& & at bot and normal user, respectively \\
\hline
\end{tabular} |
|
CL-2666 | \begin{tabular}{>{\raggedright\arraybackslash}p{2.7cm}>{\raggedright\arraybackslash}p{2.7cm}|p{0.6cm}}
\hline
External representation & Internal representation & Test BLEU \\
\hline
Plain BPE & Plain BPE & 29.2 \\
Linearized derivation & Linearized derivation & 28.8 \\
\hline
Linearized tree & Plain BPE & 28.9 \\
Plain BPE & Linearized derivation & 28.8 \\
Linearized derivation & Plain BPE & 29.4$^\dagger$ \\
POS/BPE & Plain BPE & 29.3$^\dagger$ \\
Plain BPE & POS/BPE & 29.4$^\dagger$ \\
\end{tabular} |
|
CV-5329 | \begin{tabular}{|p{3.5cm}|p{0.8cm}|p{0.8cm}|p{0.8cm}|p{0.8cm}||p{0.8cm}|p{0.8cm}|p{0.8cm}|p{0.8cm}|}
\hline
\multirow{2}{*}{Method} & \multicolumn{4}{c|}{Market1501 $\rightarrow$ DukeMTMC-reID} & \multicolumn{4}{c|}{DukeMTMC-reID $\rightarrow$ Market1501 } \\
\cline{2-9}
\cline{2-9}
& R1 & R5 & R10 & mAP & R1 & R5 & R10 & mAP \\
\hline
Direct Transfer & 42.4 & 56.5 & 63.2 & 23.0 & 52.0 & 70.2 & 76.5 & 22.0 \\
CycleGAN & 44.1 & 58.6 & 65.0 & 23.6 & 55.2 & 72.8 & 79.4 & 23.2 \\
PTGAN & 27.4 & - & 50.7 & - & 38.6 & - & 66.1 & - \\
SPGAN & 41.1 & 56.6 & 63.0 & 22.3 & 51.5 & 70.1 & 76.8 & 22.8 \\
ATNet & 45.1 & 59.5 & 64.2 & 24.9 & 55.7 & 73.2 & 79.4 & 25.6 \\
M2M-GAN & 49.6 & - & - & 26.1 & 57.5 & - & - & 26.8 \\
CR-GAN & 52.2 & - & - & 30.0 & 59.6 & - & - & 29.6 \\
\hline
EDAAN with Triplet & 55.2 & 68.0 & 72.6 & 33.5 & 62.3 & 81.8 & 84.0 & 32.7 \\
EDAAN with Quartet & \textbf{57.8} & \textbf{72.2} & \textbf{78.3} & \textbf{39.6} & \textbf{64.5} & \textbf{83.0} & \textbf{86.3} & \textbf{35.4} \\
\hline
\end{tabular} |
|
AI-11117 | \begin{tabular}{ccccccccc}
\hline
Dataset & level1 & level2 & level3 & level4 & level5 & level6 & level7 & level8 \\
\hline
RCV1 & 236334 & 20523 & 11850 & 23211 & - & - & - & - \\
NYT & 15161 & 2923 & 1160 & 842 & 1066 & 925 & 992 & 1460 \\
WOS & 6712 & 351 & - & - & - & - & - & - \\
\hline
\end{tabular} |
|
CR-46823 | \begin{tabular}{|l||p{1.5cm}|p{1.5cm}|p{1.25cm}||p{1.5cm}|p{1.5cm}|p{1.25cm}|}
\hline
~ & \multicolumn{3}{c||}{TCP} & \multicolumn{3}{c|}{DCCP} \\
~ & Reported Attacks & Interesting \newline (Off-path) Attacks & Unique Attacks & Reported Attacks & Interesting \newline (Off-path) Attacks & Unique Attacks \\
\hline
Random & 996 & 0 & 0 & 992 & 0 & 0 \\
Manual & 219 & 63 & 5 & 209 & 44 & 2 \\
NLP-based & 220 & 69 & 5 & 254 & 47 & 2 \\
\hline
\end{tabular} |
|
CR-8746 | \begin{tabular}[c]{@{}l@{}}CopywritingTranslations,SocialMediaMarketingServices,\\OptimizationPromotionandAudit\end{tabular} |
|
CR-7160 | \begin{tabular}{cccccccccc}
\toprule
\multirow{2}{*}{Data} & \multirow{2}{*}{Measures} & \multicolumn{3}{c}{\texttt{CFD}} & \multicolumn{3}{c}{\texttt{CFD LRT}} \\
\cmidrule(lr){3-8}
& {} & \texttt{SCFE} & \texttt{GS} & \texttt{CCHVAE} & \texttt{SCFE} & \texttt{GS} & \texttt{CCHVAE} \\
\midrule
\multirow{4}{*}{A} & AUC & 0.4971 & 0.5038 & 0.5008 & 0.4988 & \textbf{0.5103 } & 0.5066 \\
& BA & 0.5115 & 0.5125 & 0.5056 & \textbf{0.5132} & 0.5098 & 0.5176 \\
& TPR (0.1) & 0.1039 & 0.1020 & 0.1058 & 0.1010 & 0.1043 & \textbf{0.1298} \\
& TPR (0.01) & 0.0121 & 0.0097 & 0.0157 & \textbf{0.0158} & 0.0095 & 0.0134 \\
\midrule
\multirow{4}{*}{H} & AUC & 0.5887 & 0.5410 & 0.4874 & 0.5829 & 0.5027 & \textbf{0.6789} \\
& BA & 0.5904 & 0.5404 & 0.5473 & 0.5924 & 0.5326 & \textbf{0.6389} \\
& TPR (0.1) & 0.1130 & 0.1223 & 0.0863 & 0.1106 & 0.1142 & \textbf{0.2635} \\
& TPR (0.01) & 0.0155 & 0.0176 & 0.0016 & 0.0135 & 0.0372 & \textbf{0.0513} \\
\midrule
\multirow{4}{*}{D} & AUC & \textbf{0.5051} & 0.5000 & NA & 0.5050 & 0.5047 & NA \\
& BA & 0.5100 & 0.5133 & NA & \textbf{0.5145} & 0.5136 & NA \\
& TPR (0.1) & 0.1020 & 0.0950 & NA & 0.0894 & \textbf{0.1181} & NA \\
& TPR (0.01) & 0.0093 & 0.0083 & NA & 0.0113 & \textbf{0.0159} & NA \\
\bottomrule
\end{tabular} |
|
SE-15978 | \begin{tabular}{lllr}
\hline
\textbf{ID} & \textbf{Mistake Type} & \textbf{Associated Mistake Class} & \textbf{Occurance} \\ \hline
1 & Lack of preparation & \textit{Teamwork and Planning} & 4 \\
2 & Lack of planning & \textit{Teamwork and Planning} & 3 \\ \hline
3 & Not identifying stakeholders & \textit{Question Omission} & 1 \\
4 & Not asking about existing system & \textit{Question Omission} & 6 \\ \hline
5 & Asking long question & \textit{Question Formulation} & 3 \\
6 & Asking unnecessary question & \textit{Question Formulation} & 7 \\
7 & Asking stakeholder for solution & \textit{Question Formulation} & 15 \\
8 & Asking vague question & \textit{Question Formulation} & 32 \\
9 & Asking technical question & \textit{Question Formulation} & 5 \\ \hline
10 & Incorrect ending of the interview & \textit{Order of interview} & 6 \\ \hline
11 & Influencing stakeholder & \textit{stakeholder interaction} & 9 \\
12 & No rapport with stakeholder & \textit{stakeholder interaction} & 16 \\
13 & Unnatural dialogue style & \textit{Communication skills} & 11 \\ \hline
\end{tabular} |
|
PL-2065 | \begin{tabular}{l|lll}
\hline
\multirow{2}{*}{$b_0$} & \texttt{int i = 0;} & & \\
& & \verb|Update|: $\phi(b_1)$ & \verb|Goto| $b_1$ \\
\cline{1-1}\cline{3-4}
\multirow{3}{*}{$b_1$} & & Skip to $b_2$ unless $SAT(\phi(b_1))$ & \\
& \texttt{while (i < b) \{} & & \\
& & \verb|Update|: $\phi(b_2)$, $\phi(b_5)$ & \verb|Goto| $b_2$ \\
\cline{1-1}\cline{3-4}
\multirow{4}{*}{$b_2$} & & Skip to $b_3$ unless $SAT(\phi(b_2))$ & \\
& \texttt{\quad i++} & & \\
& \texttt{\quad if (i != a)} & & \\
& & \verb|Update|: $\phi(b_3)$, $\phi(b_4)$ & \verb|Goto| $b_3$ \\
\cline{1-1}\cline{3-4}
\multirow{3}{*}{$b_3$} & & Skip to $b_4$ unless $SAT(\phi(b_3))$ & \\
& \texttt{\quad\quad continue;} & & \\
& & \verb|Update|: $\phi(b_1)$ & \verb|Goto| $b_1$ \\
\cline{1-1}\cline{3-4}
\multirow{3}{*}{$b_4$} & & Skip to $b_5$ unless $SAT(\phi(b_4))$ & \\
& \texttt{\quad \ldots} & & \\
& \texttt{\}} & \verb|Update|: $\phi(b_1)$ & \verb|Goto| $b_1$ \\
\cline{1-1}\cline{3-4}
\multirow{1}{*}{$b_5$} & \texttt{return;} & & \\
\hline
\end{tabular} |
|
CV-8720 | \begin{tabular}{ccccccccc}
\hline
Model & 0.25 & 0.5 & 1 & 2 & 4 & 8 & 16 & 32 \\[0.5ex]
\hline
Chained & 4.76 & 8.33 & 15.55 & 28.23 & 44.69 & 58.62 & 65.89 & 67.49 \\
2-SHG & 5.59 & 10.87 & 22.25 & 41.62 & 61.78 & 73.9 & 79.21 & 79.78 \\
DeepPose & 3.3 & 4.86 & 7.99 & 12.98 & 18.26 & 21.33 & 22.79 & 23.12 \\
\hline
\end{tabular} |
|
SE-1008 | \begin{tabular}{l|c|c|}
\cline{2-3}
\multicolumn{1}{c|}{} & Description & Artifact Type \\ \hline
\multicolumn{1}{|l|}{{CR01}} & \makecell{Every lifeline must have \\ a corresponding class.} & uml:Lifeline \\ \hline
\multicolumn{1}{|l|}{{CR02}} & \makecell{Every transition has to have \\ a corresponding message.} & uml:Transition \\ \hline
\multicolumn{1}{|l|}{{CR03}} & \makecell{Statechart Action must be defined \\ as an operation in the owner’s class.} & uml:Transition \\ \hline
\multicolumn{1}{|l|}{{CR04}} & \makecell{Message actions must be defined \\ as an operation in receiver’s class.} & uml:Message \\ \hline
\multicolumn{1}{|l|}{{CR05}} & \makecell{Operation parameters \\ must have unique names.} & uml:Operation \\ \hline
\multicolumn{1}{|l|}{{CR06}} & \makecell{An Operation has at most \\ one return parameter.} & uml:Operation \\ \hline
\multicolumn{1}{|l|}{{CR07}} & \makecell{An interface can have at \\ most one generalization.} & uml:Interface \\ \hline
\multicolumn{1}{|l|}{{CR08}} & \makecell{An interface can only contain \\ public operations and no attributes.} & uml:Interface \\ \hline
\multicolumn{1}{|l|}{{CR09}} & \makecell{No two class operations may \\ have the same signature.} & uml:Class \\ \hline
\multicolumn{1}{|l|}{{CR10}} & \makecell{No two fields may have \\ the same name.} & uml:Class \\ \hline
\end{tabular} |
|
CR-52556 | \begin{tabular}{||cccc||}
\hline
\textbf{Datasets} & \textbf{Nodes (N)} & \textbf{Dimension (d)} & \textbf{Classes (c)} \\ [0.5ex]
\hline\hline
Iris & 150 & 4 & 3 \\
\hline
Glass & 214 & 9 & 6 \\
\hline
Wine & 178 & 13 & 3 \\
\hline
Control Chart & 600 & 60 & 6 \\
\hline
Parkinsons & 195 & 22 & 2 \\
\hline
Vertebral & 310 & 6 & 3 \\
\hline
Breast tissue & 106 & 9 & 6 \\
\hline
Seeds & 210 & 7 & 3 \\ [1ex]
\hline
\end{tabular} |
|
CR-29421 | \begin{tabular}{p{3cm}<{\raggedright}p{5cm}<{\raggedright}p{9cm}<{\raggedright}}
\hline
\textbf{Type} & \textbf{Approach} & \textbf{Brief Introduction} \\
\hline
\multirow{4}{3cm}{Original Approaches with complete process frameworks} & E-Safety Vehicle Intrusion Protected Applications (EVITA) & EVITA approach considers four security objectives (safety, privacy, financial, operational) and uses attacks trees to identify threats and assess risks . \\
\cline{2-3} & Threat, Vulnerabilities, and implementation Risks Analysis (TVRA) & TVRA is a process-driven TARA approach to systematically identify unwanted incidents which need to be avoided . \\
\cline{2-3} & Operationally Critical Threat, Asset, and Vulnerability Evaluation (OCTAVE) & OCTAVE is a process-driven TARA method which is best suited for enterprise information security risk assessments . \\
\cline{2-3} & HEAling Vulnerabilities to ENhance Software Security and Safety (HEAVENS) & HEAVENS is a systematic approach of deriving security requirements for vehicle E/E systems, including processes and tools supporting for TARA . \\
\hline
\multirow{3}{3cm}{Approaches evolved from other disciplines and support co-analysis} & A Security-Aware Hazard and Risk Analysis Method (SAHARA) & SAHARA is a combined approach of the Hazard Analysis and Risk Assessment (HARA) with the STRIDE model and outlines the impacts of security issues on safety concepts . \\
\cline{2-3} & Failure Mode, Vulnerabilities and Effects Analysis (FMVEA) & FMVEA is an approach evolved from the Failure Mode and Effect Analysis (FMEA) to identify vulnerability cause-effect chains for security . \\
\cline{2-3} & Combined Harm Assessment of Safety and Security (CHASSIS) & CHASSIS is a unified process for safety and security by using UML-based models (e.g. misuse cases and sequence diagrams) . \\
\hline
\end{tabular} |
|
CV-13976 | \begin{tabular}{|c|c|c|c|c|c|c|c|c|c|c|c|c|c|}
\hline
Methods & Land & Forest & Residential & Haystack & Road & Church & Car & Water & Sky & Hill & Person & Fence & Overall \\
\hline
w/ Base Train & .495 & .496 & .774 & .000 & .252 & .166 & .000 & .006 & .952 & .371 & .000 & .060 & .298 \\
w/ SegProp Train & \textbf{.540} & \textbf{.516} & \textbf{.822} & .586 & \textbf{.432} & \textbf{.382} & \textbf{.066} & .146 & \textbf{.985} & \textbf{.407} & \textbf{.471} & \textbf{.233} & \textbf{.466} \\
\hline
\end{tabular} |
|
CV-24411 | \begin{tabular}{l|cc|cc}
\specialrule{1.2pt}{1pt}{1pt}
\multirow{2}{*}{\hspace{0.08cm} Method} & \multicolumn{2}{c|}{Segmentation} &
\multicolumn{2}{c}{Robustness} \\
\cline{2-5}
& \textbf{B} & \textbf{W} & \textbf{B} & \textbf{W} \\
\specialrule{1.2pt}{1pt}{1pt}
DeepLabv3-Res50 & 73.9 & 74.1 & 53.7 & \textbf{55.8} \\
DeepLabv3-Res101 & 75.5 & 75.2 & 49.8 & \textbf{51.9} \\
\specialrule{1.2pt}{1pt}{1pt}
\end{tabular} |
|
CL-1656 | \begin{tabular}{lrr}
\toprule
$K$ & Successor surprisal & Total entropy \\
\midrule
5 & 0.212 & 0.541 \\
50 & 0.335 & 0.820 \\
500 & 0.397 & 0.947 \\
5000 & 0.434 & 0.992 \\
50000 & 0.454 & 1 \\
\bottomrule
\end{tabular} |
|
CR-49011 | \begin{tabular}{cc}
\toprule
Component & Types considered \\
\midrule
Trend & linear model, local level, local linear \\
Seasonal & hourly, daily \\
Error & Gaussian, AR(p): autoregressive model of order p=1,2 \\
\bottomrule
\end{tabular} |
|
CR-2892 | \begin{tabular}{cccccc}
\toprule
& CIFAR10 & CIFAR100 & Purchase100 & Texas100 & Location \\
\midrule
$p^*$ & 0.13 & 0.11 & 0.015 & 0.005 & 0.015 \\
\bottomrule
\end{tabular} |
|
CR-39239 | \begin{tabular}{lccc}
\textbf{Dataset} & \textbf{Classes} & \textbf{Instances/Class} & \textbf{Total} \\
\hline
Undefended & 95 & 1000 & 95,000 \\
WTF-PAD & 95 & 1000 & 95,000 \\
Walkie-Talkie (sim.) & 100 & 900 & 90,000 \\
Walkie-Talkie (real) & 100 & 750 & 75,000 \\
Onion Sites & 538 & 77 & 41,426 \\
\hline
\end{tabular} |
|
AI-37528 | \begin{tabular}{c|c|c|c}
\hline
$\pi_b$ & 20$\times$20 & 50$\times$20 & 100$\times$20 \\
\hline
\hline
MWKR & \textbf{1803.1} & \textbf{3147.3} & \textbf{5676.0} \\
MOR & 1831.7 & 3229.8 & 5728.3 \\
SPT & 1813.8 & 3201.7 & 5718.7 \\
FIFO & 1826.4 & 3177.6 & 5692.9 \\
\hline
\end{tabular} |
|
AI-18721 | \begin{tabular}{r|r|r|r}
& \textbf{Wikipedia} & \textbf{Wikinews} & \textbf{Science} \\ \hline
\textbf{Sentences} & 15,000 & 14,682 & 46,715 \\
\textbf{Verbs} & 32,758 & 34,026 & 66,653 \\
\textbf{Questions} & 75,867 & 80,081 & 143,388 \\
\textbf{Valid Qs} & 67,146 & 70,555 & 127,455
\end{tabular} |
End of preview. Expand
in Dataset Viewer.
LATTE: Improving Latex Recognition for Tables and Formulae with Iterative Refinement
Tab2Latex: a Latex table recognition dataset, with 87,513 training, 5,000 validation, and 5,000 test instances. The LaTeX sources are collected from academic papers within these six distinct sub-fields of computer science—Artificial Intelligence, Computation and Language, Computer Vision and Pattern Recognition, Cryptography and Security, Programming Languages, and Software Engineering—from the arXiv repository, covering the years 2018 to 2023. Once the paper sources are downloaded, tables are identified and extracted from the LaTeX source code by matching \begin{tabular} and \end{tabular} and removing the comments. Then, the LaTeX table source scripts are rendered to PDF format and converted to PNG format at 160 dpi.
Data Fields
- id: instance id
- image: the rendered image (PIL.Image) from the Latex source code
- latex: the Latex source code for the table
- Downloads last month
- 7