{"noiseShare":0.5194,"topics":[{"claims":[175,418,610,644,832,1249,1348,1574,1556],"deltaPp":-0.4318,"id":0,"keywords":["graph","node","neural","networks","classification"],"label":"Graph Neural Networks","labelFull":"Graph Neural Networks: Architectures, Training, and Scaling","papers":3346,"papersByYear":[79,156,239,263,307,454,514,620,714],"rel":-0.1075,"reviewed":true,"share":[4.0163,5.7714,6.5158,6.0114,6.1425,6.859,5.7792,5.0211,3.5845],"size":8406,"slope":-0.000767},{"claims":[13,22,19,62,134,653,1291,1808,2567],"deltaPp":6.7316,"id":1,"keywords":["diffusion","denoising","sampling","video","image"],"label":"Accelerated Diffusion Sampling","labelFull":"Training-Free and Distillation for Fast Diffusion Sampling","papers":3470,"papersByYear":[9,18,14,31,66,315,638,947,1432],"rel":14.7139,"reviewed":true,"share":[0.4575,0.6659,0.3817,0.7086,1.3205,4.759,7.1734,7.6693,7.1891],"size":6569,"slope":0.010928},{"claims":[60,102,169,215,327,615,960,1628,2391],"deltaPp":2.9849,"id":2,"keywords":["protein","molecular","molecules","prediction","design"],"label":"Molecular and Protein Design","labelFull":"Machine Learning for Molecular and Protein Design","papers":2545,"papersByYear":[35,48,75,96,142,239,361,600,949],"rel":1.6775,"reviewed":true,"share":[1.7794,1.7758,2.0447,2.1943,2.8411,3.6108,4.0589,4.8591,4.7643],"size":6467,"slope":0.004439},{"claims":[85,155,265,347,436,610,1026,1439,1843],"deltaPp":2.2469,"id":3,"keywords":["3d","point","pose","rendering","scene"],"label":"Neural 3D Representations","labelFull":"Neural 3D Representations For Reconstruction & Rendering","papers":2669,"papersByYear":[43,64,106,120,175,258,419,601,883],"rel":1.0278,"reviewed":true,"share":[2.1861,2.3677,2.8899,2.7429,3.5014,3.8979,4.711,4.8672,4.433],"size":6206,"slope":0.003547},{"claims":[252,397,513,542,637,642,740,825,1194],"deltaPp":-2.5638,"id":4,"keywords":["adversarial","attacks","robustness","backdoor","examples"],"label":"Adversarial Robustness","labelFull":"Adversarial Robustness and Defense Strategies for ML Models","papers":2774,"papersByYear":[116,193,230,235,260,291,349,436,664],"rel":-0.4347,"reviewed":true,"share":[5.8973,7.1402,6.2704,5.3714,5.2021,4.3964,3.924,3.5309,3.3335],"size":5742,"slope":-0.004458},{"claims":[222,272,402,564,592,691,806,781,1338],"deltaPp":-2.1297,"id":5,"keywords":["reinforcement","learning","offline","imitation","policy"],"label":"Offline Reinforcement Learning","labelFull":"Offline Reinforcement Learning Algorithms and Benchmarks","papers":3618,"papersByYear":[136,175,251,326,365,418,485,509,953],"rel":-0.308,"reviewed":true,"share":[6.9141,6.4743,6.843,7.4514,7.3029,6.3152,5.4531,4.1221,4.7844],"size":5668,"slope":-0.003249},{"claims":[40,36,74,81,120,259,912,1431,1756],"deltaPp":4.9129,"id":6,"keywords":["fine-tuning","modeling","framework"],"label":"Efficient LLM Training","labelFull":"Efficient Training and Fine-Tuning for Large Language Models","papers":3228,"papersByYear":[27,28,51,59,87,164,610,950,1252],"rel":3.5793,"reviewed":true,"share":[1.3726,1.0359,1.3904,1.3486,1.7407,2.4777,6.8586,7.6936,6.2855],"size":4709,"slope":0.008615},{"claims":[94,115,162,213,248,368,472,886,1788],"deltaPp":2.8107,"id":7,"keywords":["state-of-the-art","benchmarks","inference","outperforms","baselines"],"label":"Benchmark Performance","labelFull":"State-of-the-Art Performance on ML Benchmarks","papers":3733,"papersByYear":[88,102,142,187,223,337,427,776,1451],"rel":0.6283,"reviewed":true,"share":[4.4738,3.7736,3.8713,4.2743,4.4618,5.0914,4.801,6.2844,7.2845],"size":4346,"slope":0.003575},{"claims":[269,321,402,371,431,501,532,556,732],"deltaPp":-5.4068,"id":8,"keywords":["gradient","stochastic","descent","convergence","convex"],"label":"Stochastic Gradient Descent Convergence","labelFull":"Convergence Analysis of Stochastic Gradient Methods","papers":2334,"papersByYear":[150,176,207,210,245,282,299,323,442],"rel":-0.709,"reviewed":true,"share":[7.6258,6.5113,5.6434,4.8,4.902,4.2605,3.3618,2.6158,2.219],"size":4115,"slope":-0.006403},{"claims":[13,32,30,37,66,137,314,1018,2449],"deltaPp":6.8477,"id":9,"keywords":["reasoning","mathematical","chain-of-thought","benchmarks","accuracy"],"label":"Efficient Chain-of-Thought Reasoning","labelFull":"Efficient Chain-of-Thought Reasoning for LLMs","papers":2366,"papersByYear":[8,13,18,24,34,78,185,561,1445],"rel":16.8372,"reviewed":true,"share":[0.4067,0.4809,0.4907,0.5486,0.6803,1.1784,2.0801,4.5432,7.2544],"size":4096,"slope":0.007231},{"claims":[326,413,453,418,462,474,410,409,503],"deltaPp":-7.4296,"id":10,"keywords":["networks","relu","neural","deep","activation"],"label":"Layerwise Training","labelFull":"Layerwise Training of Deep Neural Networks","papers":2419,"papersByYear":[181,223,260,268,281,294,281,278,353],"rel":-0.8074,"reviewed":true,"share":[9.2018,8.2501,7.0883,6.1257,5.6222,4.4418,3.1594,2.2514,1.7722],"size":3868,"slope":-0.009543},{"claims":[91,130,225,278,292,480,603,703,928],"deltaPp":0.1139,"id":11,"keywords":["causal","treatment","counterfactual","effect","discovery"],"label":"Causal Discovery Benchmarks","labelFull":"Comprehensive Benchmark Suite for Causal Discovery Methods","papers":1666,"papersByYear":[42,60,89,128,135,216,245,303,448],"rel":0.0533,"reviewed":true,"share":[2.1352,2.2198,2.4264,2.9257,2.7011,3.2633,2.7547,2.4538,2.2491],"size":3730,"slope":0.000359},{"claims":[179,250,331,362,355,447,422,545,782],"deltaPp":-4.1284,"id":12,"keywords":["estimator","complexity","bounds","algorithm","sample"],"label":"Sample Complexity Bounds","labelFull":"Sample Complexity Bounds for Learning Algorithms","papers":2695,"papersByYear":[137,167,238,269,269,340,315,395,565],"rel":-0.5927,"reviewed":true,"share":[6.9649,6.1783,6.4885,6.1486,5.3822,5.1367,3.5417,3.1989,2.8365],"size":3673,"slope":-0.005393},{"claims":[49,71,90,139,155,262,501,688,1156],"deltaPp":1.1093,"id":13,"keywords":["series","forecasting","time","multivariate","temporal"],"label":"Time Series Foundation Models","labelFull":"Time Series Foundation Models for Forecasting and Beyond","papers":1460,"papersByYear":[32,39,51,76,75,127,209,306,545],"rel":0.6819,"reviewed":true,"share":[1.6268,1.4428,1.3904,1.7371,1.5006,1.9187,2.3499,2.4781,2.7361],"size":3111,"slope":0.001607},{"claims":[48,71,81,113,106,101,252,523,1654],"deltaPp":2.8319,"id":14,"keywords":["agents","llm","games","systems","benchmark"],"label":"LLM Agents","labelFull":"Large Language Model Agents: Frameworks and Benchmarks","papers":1677,"papersByYear":[31,49,61,76,87,64,148,283,878],"rel":1.7969,"reviewed":true,"share":[1.576,1.8128,1.663,1.7371,1.7407,0.9669,1.664,2.2919,4.4079],"size":2949,"slope":0.001999},{"claims":[82,113,93,190,184,237,392,583,1035],"deltaPp":0.3079,"id":15,"keywords":["audio","speech","music","acoustic","recognition"],"label":"Audio Foundation Models","labelFull":"Self-Supervised Audio Foundation Models","papers":995,"papersByYear":[29,45,36,64,65,85,120,196,355],"rel":0.2088,"reviewed":true,"share":[1.4743,1.6648,0.9815,1.4629,1.3005,1.2842,1.3492,1.5873,1.7822],"size":2909,"slope":0.000259},{"claims":[4,12,20,33,57,149,348,669,1470],"deltaPp":4.4053,"id":16,"keywords":["vision-language","visual","reasoning","spatial","benchmarks"],"label":"Vision-Language Reasoning","labelFull":"Evaluation and Enhancement of Visual Reasoning in Large VLMs","papers":1744,"papersByYear":[4,9,15,30,37,102,223,406,918],"rel":21.6583,"reviewed":true,"share":[0.2034,0.333,0.4089,0.6857,0.7403,1.541,2.5073,3.288,4.6087],"size":2762,"slope":0.005256},{"claims":[77,158,224,285,325,353,365,394,579],"deltaPp":-0.5889,"id":17,"keywords":["regret","bound","algorithm","online","sqrt"],"label":"Online Regret Minimization","labelFull":"Online Regret Minimization With Sublinear Regret Guarantees","papers":1496,"papersByYear":[42,85,131,152,174,205,197,202,308],"rel":-0.2758,"reviewed":true,"share":[2.1352,3.1447,3.5714,3.4743,3.4814,3.0971,2.215,1.6359,1.5463],"size":2760,"slope":-0.001662},{"claims":[67,93,216,333,390,443,404,388,370],"deltaPp":-0.739,"id":18,"keywords":["label","contrastive","learning","semi-supervised","augmentation"],"label":"Weak Supervision","labelFull":"Weak Supervision: Learning from Noisy and Limited Labels","papers":1643,"papersByYear":[41,66,124,196,208,253,253,234,268],"rel":-0.3545,"reviewed":true,"share":[2.0844,2.4417,3.3806,4.48,4.1617,3.8223,2.8446,1.895,1.3454],"size":2704,"slope":-0.001054},{"claims":[78,139,156,227,256,381,351,357,441],"deltaPp":-0.5776,"id":19,"keywords":["privacy","private","differential","guarantees","algorithms"],"label":"Differentially Private Learning","labelFull":"Differential Privacy for Machine Learning","papers":1032,"papersByYear":[32,58,59,90,99,158,156,171,209],"rel":-0.3551,"reviewed":true,"share":[1.6268,2.1458,1.6085,2.0571,1.9808,2.3871,1.754,1.3848,1.0492],"size":2386,"slope":-0.000662},{"claims":[88,112,150,132,211,189,296,408,745],"deltaPp":-0.0618,"id":20,"keywords":["quantization","compression","quantized","weight","error"],"label":"Low-Bit Quantization","labelFull":"Low-Bit Quantization and Quantization-Aware Training for Efficient Neural Networks","papers":1234,"papersByYear":[42,57,68,70,101,108,161,214,413],"rel":-0.0289,"reviewed":true,"share":[2.1352,2.1088,1.8539,1.6,2.0208,1.6317,1.8102,1.7331,2.0734],"size":2331,"slope":-0.000238},{"claims":[44,63,81,121,119,265,380,435,704],"deltaPp":1.0264,"id":21,"keywords":["image","restoration","super-resolution","denoising","reconstruction"],"label":"Diffusion-Based Image Restoration","labelFull":"Diffusion Models for Image Restoration and Super-Resolution","papers":1414,"papersByYear":[28,42,54,72,71,155,212,292,488],"rel":0.721,"reviewed":true,"share":[1.4235,1.5538,1.4722,1.6457,1.4206,2.3417,2.3836,2.3648,2.4499],"size":2212,"slope":0.00151},{"claims":[80,129,180,224,218,266,293,281,491],"deltaPp":-0.9273,"id":22,"keywords":["policy","markov decision","decision processes","mdps","gradient"],"label":"Off-Policy Policy Optimization","labelFull":"Off-Policy Policy Optimization for Sample-Efficient RL","papers":1426,"papersByYear":[53,83,119,147,139,174,171,188,352],"rel":-0.3441,"reviewed":true,"share":[2.6945,3.0707,3.2443,3.36,2.7811,2.6288,1.9226,1.5225,1.7672],"size":2162,"slope":-0.001955},{"claims":[78,153,200,270,270,320,281,317,263],"deltaPp":-1.209,"id":23,"keywords":["bandits","regret","contextual","multi-armed","linear"],"label":"Bandit Regret Minimization","labelFull":"Regret Bounds for Stochastic and Adversarial Bandits","papers":1067,"papersByYear":[38,77,97,128,134,161,132,156,144],"rel":-0.6258,"reviewed":true,"share":[1.9319,2.8487,2.6445,2.9257,2.6811,2.4324,1.4841,1.2634,0.7229],"size":2152,"slope":-0.002068},{"claims":[28,50,110,132,175,256,307,455,616],"deltaPp":0.9644,"id":24,"keywords":["pde","equations","neural","differential","physics-informed"],"label":"Neural PDE Solvers","labelFull":"Neural PDE Solvers: PINNs and Neural Operators","papers":1120,"papersByYear":[15,25,60,75,86,131,161,223,344],"rel":1.2646,"reviewed":true,"share":[0.7626,0.9249,1.6358,1.7143,1.7207,1.9792,1.8102,1.806,1.727],"size":2129,"slope":0.001186},{"claims":[18,40,57,112,124,131,289,465,866],"deltaPp":1.8838,"id":25,"keywords":["attention","heads","softmax","sparse","linear"],"label":"Efficient Attention","labelFull":"Efficient Attention Mechanisms for Transformers","papers":1250,"papersByYear":[14,23,29,66,70,83,187,261,517],"rel":2.6469,"reviewed":true,"share":[0.7117,0.8509,0.7906,1.5086,1.4006,1.254,2.1025,2.1137,2.5955],"size":2102,"slope":0.002282},{"claims":[4,5,26,28,40,69,219,513,1177],"deltaPp":4.0087,"id":26,"keywords":["token","attention","prediction","inference"],"label":"Dynamic Token Pruning","labelFull":"Dynamic Token Pruning for Efficient Inference","papers":1497,"papersByYear":[4,5,18,26,32,59,158,356,839],"rel":19.7085,"reviewed":true,"share":[0.2034,0.185,0.4907,0.5943,0.6403,0.8914,1.7765,2.8831,4.2121],"size":2081,"slope":0.0045},{"claims":[42,65,104,122,144,193,254,392,749],"deltaPp":0.9486,"id":27,"keywords":["manipulation","robotic","navigation","locomotion","control"],"label":"Robotic Skill Learning","labelFull":"Learning for Robotic Manipulation, Navigation, Locomotion","papers":1395,"papersByYear":[32,49,66,94,97,127,164,253,513],"rel":0.5831,"reviewed":true,"share":[1.6268,1.8128,1.7993,2.1486,1.9408,1.9187,1.8439,2.0489,2.5754],"size":2065,"slope":0.000727},{"claims":[38,25,77,118,165,213,263,417,744],"deltaPp":1.2373,"id":28,"keywords":["distillation","teacher","student","knowledge","distribution"],"label":"Knowledge Distillation","labelFull":"Knowledge Distillation for Efficient Model Compression","papers":1153,"papersByYear":[20,13,37,55,84,111,157,227,449],"rel":1.2169,"reviewed":true,"share":[1.0168,0.4809,1.0087,1.2571,1.6807,1.677,1.7652,1.8384,2.2541],"size":2060,"slope":0.001826},{"claims":[24,27,36,63,56,134,265,456,828],"deltaPp":2.0475,"id":29,"keywords":["video","generation","editing","image","quality"],"label":"Neural Video Generation","labelFull":"Neural Video Generation and Editing","papers":1337,"papersByYear":[17,23,26,41,35,100,181,334,580],"rel":2.369,"reviewed":true,"share":[0.8643,0.8509,0.7088,0.9371,0.7003,1.5108,2.0351,2.7049,2.9118],"size":1889,"slope":0.00283},{"claims":[0,8,55,131,197,296,359,278,281],"deltaPp":0.8284,"id":30,"keywords":["federated","learning","client","personalized","communication"],"label":"Heterogeneous Federated Learning","labelFull":"Robust and Efficient Federated Learning under Data Heterogeneity","papers":790,"papersByYear":[0,3,25,60,90,143,168,136,165],"rel":null,"reviewed":true,"share":[0.0,0.111,0.6816,1.3714,1.8007,2.1604,1.8889,1.1014,0.8284],"size":1605,"slope":0.001581},{"claims":[155,172,185,188,149,169,145,174,175],"deltaPp":-4.0239,"id":31,"keywords":["variational","gaussian","inference","autoencoders","process"],"label":"Variational Inference","labelFull":"Variational Inference for Deep Generative and Bayesian Models","papers":976,"papersByYear":[91,108,117,121,98,114,94,113,120],"rel":-0.8698,"reviewed":true,"share":[4.6263,3.9956,3.1897,2.7657,1.9608,1.7223,1.0569,0.9151,0.6024],"size":1512,"slope":-0.005108},{"claims":[52,47,58,68,59,98,197,301,617],"deltaPp":0.6876,"id":32,"keywords":["question","answering","summarization","retrieval","benchmarks"],"label":"Question Answering","labelFull":"Question Answering: Retrieval, Multi-Hop, Knowledge-Intensive","papers":1075,"papersByYear":[32,31,41,41,39,66,140,224,461],"rel":0.4227,"reviewed":true,"share":[1.6268,1.1469,1.1178,0.9371,0.7803,0.9971,1.5741,1.8141,2.3144],"size":1497,"slope":0.000954},{"claims":[1,23,11,33,62,134,210,393,570],"deltaPp":1.7515,"id":33,"keywords":["multimodal","modality","fusion","information","alignment"],"label":"Modality-Aware Multimodal Fusion","labelFull":"Modality-Aware Multimodal Fusion and Alignment","papers":846,"papersByYear":[1,13,7,20,31,71,122,222,359],"rel":34.4783,"reviewed":true,"share":[0.0508,0.4809,0.1908,0.4571,0.6202,1.0727,1.3717,1.7979,1.8023],"size":1437,"slope":0.002322},{"claims":[23,39,69,92,106,133,161,334,415],"deltaPp":0.7799,"id":34,"keywords":["continual","forgetting","unlearning","catastrophic","memorization"],"label":"Continual Learning","labelFull":"Continual Learning with Catastrophic Forgetting Mitigation","papers":872,"papersByYear":[13,27,44,52,61,82,96,210,287],"rel":1.1801,"reviewed":true,"share":[0.6609,0.9989,1.1996,1.1886,1.2205,1.2389,1.0794,1.7007,1.4408],"size":1372,"slope":0.000839},{"claims":[115,115,91,128,78,122,159,228,327],"deltaPp":-1.7438,"id":35,"keywords":["low-rank","matrix","sensing","matrices","factorization"],"label":"Low-Rank Matrix Approximation","labelFull":"Low-Rank Matrix Approximation and Recovery for Efficient ML","papers":839,"papersByYear":[58,56,63,64,51,68,88,151,240],"rel":-0.5914,"reviewed":true,"share":[2.9487,2.0718,1.7176,1.4629,1.0204,1.0273,0.9894,1.2229,1.2049],"size":1363,"slope":-0.001902},{"claims":[20,32,68,97,154,204,218,239,252],"deltaPp":0.0802,"id":36,"keywords":["segmentation","semantic","instance","object","image"],"label":"Semantic Segmentation","labelFull":"Semantic Segmentation: State-of-the-Art Models","papers":795,"papersByYear":[16,20,46,58,93,113,127,144,178],"rel":0.0986,"reviewed":true,"share":[0.8134,0.7399,1.2541,1.3257,1.8607,1.7072,1.4279,1.1662,0.8936],"size":1284,"slope":0.000388},{"claims":[17,14,7,23,18,38,140,374,644],"deltaPp":1.8261,"id":37,"keywords":["multimodal","question","answering","visual","reasoning"],"label":"Vision-Language Reasoning","labelFull":"Vision-Language Reasoning in Multimodal Large Language Models","papers":890,"papersByYear":[10,10,6,18,12,27,99,243,465],"rel":3.5919,"reviewed":true,"share":[0.5084,0.37,0.1636,0.4114,0.2401,0.4079,1.1131,1.9679,2.3345],"size":1275,"slope":0.002332},{"claims":[30,46,42,44,69,114,147,297,470],"deltaPp":0.1548,"id":38,"keywords":["eeg","brain","fmri","recordings","activity"],"label":"EEG Foundation Models","labelFull":"Large-Scale Pretrained Foundation Models for EEG","papers":594,"papersByYear":[17,25,22,27,33,60,68,139,203],"rel":0.1791,"reviewed":true,"share":[0.8643,0.9249,0.5998,0.6171,0.6603,0.9065,0.7646,1.1257,1.0191],"size":1259,"slope":0.000307},{"claims":[33,64,74,128,145,181,186,198,230],"deltaPp":0.0582,"id":39,"keywords":["few-shot","zero-shot","classification","learning","generalization"],"label":"Few-Shot & Zero-Shot Classification","labelFull":"Few-Shot and Zero-Shot Classification with Limited Data","papers":904,"papersByYear":[19,36,44,76,95,129,141,160,204],"rel":0.0603,"reviewed":true,"share":[0.9659,1.3319,1.1996,1.7371,1.9008,1.9489,1.5853,1.2958,1.0241],"size":1239,"slope":0.000185},{"claims":[51,59,100,106,130,128,165,205,279],"deltaPp":-0.6761,"id":40,"keywords":["multi-task","transfer","meta-learning","generalization","multiple"],"label":"Multi-Task Learning","labelFull":"Multi-Task Learning with Meta-Learning and Parameter Sharing","papers":982,"papersByYear":[37,40,74,82,98,104,134,173,240],"rel":-0.3594,"reviewed":true,"share":[1.881,1.4798,2.0174,1.8743,1.9608,1.5712,1.5066,1.401,1.2049],"size":1223,"slope":-0.000711},{"claims":[37,38,43,68,30,58,116,196,472],"deltaPp":0.8219,"id":41,"keywords":["code","verification","program","tool","execution"],"label":"Execution-Guided Code Generation","labelFull":"Execution-Guided Code Generation and Verification","papers":685,"papersByYear":[18,24,23,32,21,32,71,118,346],"rel":0.8982,"reviewed":true,"share":[0.9151,0.8879,0.627,0.7314,0.4202,0.4835,0.7983,0.9556,1.737],"size":1058,"slope":0.000598},{"claims":[46,62,99,125,127,122,133,137,191],"deltaPp":-0.6823,"id":42,"keywords":["fairness","ranking","group","demographic","constraints"],"label":"Fairness-Accuracy Trade-Offs","labelFull":"Fairness-Accuracy Trade-Off Analysis and Pareto Optimization","papers":524,"papersByYear":[23,33,46,62,61,56,75,71,97],"rel":-0.5835,"reviewed":true,"share":[1.1693,1.2209,1.2541,1.4171,1.2205,0.846,0.8433,0.575,0.487],"size":1042,"slope":-0.00101},{"claims":[27,20,60,55,88,86,169,203,330],"deltaPp":0.1995,"id":43,"keywords":["spiking","neural","networks","neurons","synaptic"],"label":"Spiking Neural Network Training","labelFull":"Efficient Supervised Training of Spiking Neural Networks","papers":451,"papersByYear":[10,12,23,27,35,40,75,88,141],"rel":0.3924,"reviewed":true,"share":[0.5084,0.444,0.627,0.6171,0.7003,0.6043,0.8433,0.7127,0.7079],"size":1038,"slope":0.000337},{"claims":[3,5,18,39,63,106,234,259,311],"deltaPp":0.9168,"id":44,"keywords":["transformers","in-context","linear","trained","layers"],"label":"Efficient Transformers","labelFull":"Efficient Transformers: Compact Architectures and Training","papers":681,"papersByYear":[3,4,12,31,41,65,137,175,213],"rel":6.0118,"reviewed":true,"share":[0.1525,0.148,0.3272,0.7086,0.8203,0.982,1.5404,1.4172,1.0693],"size":1038,"slope":0.001696},{"claims":[34,54,106,94,143,171,127,144,161],"deltaPp":-0.4633,"id":45,"keywords":["games","equilibrium","nash","zero-sum","equilibria"],"label":"Equilibrium Computation in Games","labelFull":"Computation of Nash and Correlated Equilibria in Games","papers":516,"papersByYear":[18,27,47,51,64,76,71,72,90],"rel":-0.5063,"reviewed":true,"share":[0.9151,0.9989,1.2814,1.1657,1.2805,1.1482,0.7983,0.5831,0.4518],"size":1034,"slope":-0.000681},{"claims":[73,87,90,105,95,155,95,168,161],"deltaPp":-1.2051,"id":46,"keywords":["clustering","k-means","approximation","algorithm","fair"],"label":"Scalable Approximate Clustering","labelFull":"Scalable Approximate Clustering Algorithms","papers":464,"papersByYear":[32,37,34,48,41,65,45,78,84],"rel":-0.7408,"reviewed":true,"share":[1.6268,1.3688,0.9269,1.0971,0.8203,0.982,0.506,0.6317,0.4217],"size":1029,"slope":-0.001331},{"claims":[41,63,63,81,82,119,116,168,250],"deltaPp":-0.7759,"id":47,"keywords":["generative","modeling","deep","distribution","latent"],"label":"Model-Agnostic Generative Modeling","labelFull":"Model-Agnostic Enhancements for Deep Generative Modeling","papers":774,"papersByYear":[36,46,47,68,63,85,86,133,210],"rel":-0.4239,"reviewed":true,"share":[1.8302,1.7018,1.2814,1.5543,1.2605,1.2842,0.9669,1.0771,1.0543],"size":983,"slope":-0.000979},{"claims":[19,33,62,66,91,119,182,187,213],"deltaPp":0.0927,"id":48,"keywords":["detection","anomaly","out-of-distribution","ood","outlier"],"label":"Out-of-Distribution Detection","labelFull":"Out-of-Distribution Detection for Deep Neural Networks","papers":532,"papersByYear":[12,17,32,43,41,65,89,93,140],"rel":0.1519,"reviewed":true,"share":[0.6101,0.6289,0.8724,0.9829,0.8203,0.982,1.0007,0.7532,0.7028],"size":972,"slope":0.000167},{"claims":[3,13,39,39,38,31,69,207,514],"deltaPp":1.5349,"id":49,"keywords":["flow","matching","normalizing","velocity","generative"],"label":"Flow Matching","labelFull":"Flow Matching Techniques for Normalizing Flow Training","papers":584,"papersByYear":[2,9,24,22,27,18,34,122,326],"rel":15.0924,"reviewed":true,"share":[0.1017,0.333,0.6543,0.5029,0.5402,0.2719,0.3823,0.988,1.6366],"size":953,"slope":0.001222},{"claims":[1,0,0,6,25,112,185,232,357],"deltaPp":1.3549,"id":50,"keywords":["prompts","tuning","optimization","learning"],"label":"Prompt Optimization","labelFull":"Prompt Optimization and Selection for LLMs","papers":662,"papersByYear":[1,0,0,3,17,67,128,166,280],"rel":26.6713,"reviewed":true,"share":[0.0508,0.0,0.0,0.0686,0.3401,1.0122,1.4392,1.3443,1.4057],"size":918,"slope":0.002212},{"claims":[10,25,38,51,65,126,122,202,244],"deltaPp":0.3006,"id":51,"keywords":["combinatorial","vehicle routing","pareto","solutions","instances"],"label":"Neural Combinatorial Optimization","labelFull":"Neural and Learning-Based Combinatorial Optimization","papers":521,"papersByYear":[9,16,25,30,43,61,73,113,151],"rel":0.657,"reviewed":true,"share":[0.4575,0.5919,0.6816,0.6857,0.8603,0.9216,0.8208,0.9151,0.7581],"size":883,"slope":0.000448},{"claims":[8,10,26,82,104,123,172,153,173],"deltaPp":0.3978,"id":52,"keywords":["vision","transformers","visual","image","representations"],"label":"Efficient Vision Transformers","labelFull":"Efficient Vision Transformers: Training, Inference, and Adaptation","papers":636,"papersByYear":[6,9,21,58,78,87,117,120,140],"rel":1.3043,"reviewed":true,"share":[0.305,0.333,0.5725,1.3257,1.5606,1.3144,1.3155,0.9718,0.7028],"size":851,"slope":0.00083},{"claims":[11,28,54,65,60,96,124,115,275],"deltaPp":0.6777,"id":53,"keywords":["uncertainty","calibration","epistemic","quantification","aleatoric"],"label":"Uncertainty Calibration","labelFull":"Predictive Uncertainty Calibration and Quantification","papers":641,"papersByYear":[8,18,40,47,49,79,97,87,216],"rel":1.6663,"reviewed":true,"share":[0.4067,0.6659,1.0905,1.0743,0.9804,1.1935,1.0906,0.7046,1.0844],"size":828,"slope":0.000491},{"claims":[17,45,70,45,71,97,102,129,239],"deltaPp":0.2303,"id":54,"keywords":["transport","optimal","unbalanced","entropic","plans"],"label":"Scalable Optimal Transport","labelFull":"Scalable Optimal Transport: Algorithms and Map Learning","papers":444,"papersByYear":[9,23,38,24,35,52,56,70,137],"rel":0.5034,"reviewed":true,"share":[0.4575,0.8509,1.036,0.5486,0.7003,0.7856,0.6296,0.5669,0.6878],"size":815,"slope":-8.4e-05},{"claims":[48,75,87,114,110,105,84,66,103],"deltaPp":-1.5824,"id":55,"keywords":["cifar-10","cifar-100","mnist","imagenet","cifar10"],"label":"Image Classification Benchmarks","labelFull":"State-of-the-Art on CIFAR-10/100, MNIST, and ImageNet","papers":728,"papersByYear":[41,67,82,100,99,96,79,64,100],"rel":-0.7592,"reviewed":true,"share":[2.0844,2.4787,2.2356,2.2857,1.9808,1.4504,0.8882,0.5183,0.502],"size":792,"slope":-0.002623},{"claims":[14,32,50,68,94,131,116,130,155],"deltaPp":-0.0421,"id":56,"keywords":["equivariant","group","symmetries","lie","invariant"],"label":"Group Equivariant Neural Networks","labelFull":"Group Equivariant Neural Networks for Symmetry Learning","papers":452,"papersByYear":[11,16,21,32,50,73,70,76,103],"rel":-0.0753,"reviewed":true,"share":[0.5592,0.5919,0.5725,0.7314,1.0004,1.1029,0.787,0.6155,0.5171],"size":790,"slope":0.000117},{"claims":[7,8,11,10,16,21,124,272,302],"deltaPp":0.7292,"id":57,"keywords":["preference","optimization","direct","user","personalization"],"label":"Direct Preference Optimization","labelFull":"Direct Preference Optimization for LLM Alignment","papers":522,"papersByYear":[6,5,8,8,14,14,78,183,206],"rel":2.3908,"reviewed":true,"share":[0.305,0.185,0.2181,0.1829,0.2801,0.2115,0.877,1.482,1.0342],"size":771,"slope":0.001359},{"claims":[34,55,81,97,101,118,110,85,89],"deltaPp":-0.8895,"id":58,"keywords":["loss","boosting","surrogate","generalization","functions"],"label":"Surrogate Losses","labelFull":"Surrogate Loss Functions: Theory and Generalization","papers":553,"papersByYear":[25,42,61,65,70,80,66,68,76],"rel":-0.6998,"reviewed":true,"share":[1.271,1.5538,1.663,1.4857,1.4006,1.2086,0.7421,0.5507,0.3815],"size":770,"slope":-0.001448},{"claims":[44,50,59,73,68,96,115,112,135],"deltaPp":-1.4061,"id":59,"keywords":["convergence","numerical","theoretical","experiments","rate"],"label":"Convergence Rate Analysis","labelFull":"Theoretical Convergence Guarantees & Rates for ML Algorithms","papers":689,"papersByYear":[40,40,52,66,63,91,108,104,125],"rel":-0.6914,"reviewed":true,"share":[2.0336,1.4798,1.4177,1.5086,1.2605,1.3748,1.2143,0.8422,0.6275],"size":752,"slope":-0.001346},{"claims":[17,21,21,17,48,81,131,180,225],"deltaPp":0.4417,"id":60,"keywords":["captioning","text","image","alignment","clip"],"label":"Image-Text Pair Pretraining","labelFull":"Image-Text Pair Pretraining for Vision-Language Captioning","papers":517,"papersByYear":[8,12,11,15,30,55,90,127,169],"rel":1.0861,"reviewed":true,"share":[0.4067,0.444,0.2999,0.3429,0.6002,0.8309,1.0119,1.0285,0.8484],"size":741,"slope":0.000905},{"claims":[39,47,67,58,97,93,123,101,112],"deltaPp":-1.0369,"id":61,"keywords":["kernel","ridge","regression","random","reproducing"],"label":"Scalable Kernel Learning","labelFull":"Scalable Kernel Learning With Adaptive Random Features","papers":467,"papersByYear":[28,31,41,42,63,54,69,62,77],"rel":-0.7284,"reviewed":true,"share":[1.4235,1.1469,1.1178,0.96,1.2605,0.8158,0.7758,0.5021,0.3866],"size":737,"slope":-0.001152},{"claims":[6,4,7,4,6,15,61,209,417],"deltaPp":1.1007,"id":62,"keywords":["decoding","cache","kv","speculative","eviction"],"label":"Speculative Decoding","labelFull":"Speculative Decoding for Efficient LLM Inference","papers":482,"papersByYear":[6,4,7,4,6,10,37,128,280],"rel":3.6089,"reviewed":true,"share":[0.305,0.148,0.1908,0.0914,0.12,0.1511,0.416,1.0366,1.4057],"size":729,"slope":0.001263},{"claims":[137,155,120,127,81,44,22,9,22],"deltaPp":-3.0259,"id":63,"keywords":["gan","discriminator","generator","wasserstein","samples"],"label":"Data-Efficient GANs","labelFull":"Data-Efficient GANs: Discriminator Design for Limited Samples","papers":342,"papersByYear":[61,66,55,55,43,28,11,8,15],"rel":-0.9757,"reviewed":true,"share":[3.1012,2.4417,1.4995,1.2571,0.8603,0.423,0.1237,0.0648,0.0753],"size":717,"slope":-0.003803},{"claims":[0,0,6,5,18,33,138,210,303],"deltaPp":0.502,"id":64,"keywords":["watermark","robustness","detection","attacks","image"],"label":"Generative Model Watermarking","labelFull":"Robust Watermarking and Detection for Generative AI Models","papers":232,"papersByYear":[0,0,2,1,7,11,43,68,100],"rel":null,"reviewed":true,"share":[0.0,0.0,0.0545,0.0229,0.1401,0.1662,0.4835,0.5507,0.502],"size":713,"slope":0.000777},{"claims":[11,18,50,63,86,79,99,110,197],"deltaPp":0.1091,"id":65,"keywords":["detection","object","detectors","coco","ai-generated"],"label":"Object Detection","labelFull":"Object Detection: State-of-the-Art Methods and Benchmarks","papers":454,"papersByYear":[10,9,27,33,56,53,66,77,123],"rel":0.2146,"reviewed":true,"share":[0.5084,0.333,0.7361,0.7543,1.1204,0.8007,0.7421,0.6236,0.6175],"size":713,"slope":0.000228},{"claims":[9,18,24,16,21,56,93,194,270],"deltaPp":0.6024,"id":66,"keywords":["reward","preference","human","feedback","functions"],"label":"Reward Modeling","labelFull":"Reward Modeling for Reinforcement Learning from Human Feedback","papers":506,"papersByYear":[8,18,14,14,20,39,61,131,201],"rel":1.4812,"reviewed":true,"share":[0.4067,0.6659,0.3817,0.32,0.4002,0.5892,0.6859,1.0609,1.0091],"size":701,"slope":0.000745},{"claims":[1,9,18,13,44,48,85,173,299],"deltaPp":0.7926,"id":67,"keywords":["medical","segmentation","imaging","pathology","slide"],"label":"Medical Image Segmentation","labelFull":"Medical Image Segmentation Across Modalities and Tasks","papers":383,"papersByYear":[1,5,11,10,28,23,44,93,168],"rel":15.6024,"reviewed":true,"share":[0.0508,0.185,0.2999,0.2286,0.5602,0.3475,0.4947,0.7532,0.8434],"size":690,"slope":0.000897},{"claims":[4,5,7,7,13,25,93,173,354],"deltaPp":1.2224,"id":68,"keywords":["reinforcement","agents","learning","reward","feedback"],"label":"LLM Reinforcement Learning","labelFull":"Reinforcement Learning for Large Language Models","papers":529,"papersByYear":[4,3,4,4,10,20,73,127,284],"rel":6.0098,"reviewed":true,"share":[0.2034,0.111,0.1091,0.0914,0.2001,0.3022,0.8208,1.0285,1.4258],"size":681,"slope":0.001546},{"claims":[36,25,61,87,88,106,87,92,92],"deltaPp":-0.6245,"id":69,"keywords":["domain","adaptation","generalization","source","unsupervised"],"label":"Unsupervised Domain Adaptation","labelFull":"Unsupervised Domain Adaptation and Generalization","papers":437,"papersByYear":[19,17,34,48,53,64,69,65,68],"rel":-0.6465,"reviewed":true,"share":[0.9659,0.6289,0.9269,1.0971,1.0604,0.9669,0.7758,0.5264,0.3414],"size":674,"slope":-0.00054},{"claims":[68,76,64,78,50,57,64,69,114],"deltaPp":-1.5195,"id":70,"keywords":["recurrent","neural","networks","memory","dynamics"],"label":"Recurrent Neural Network Theory","labelFull":"Theoretical Analysis of Recurrent Neural Network Dynamics","papers":357,"papersByYear":[37,42,36,35,28,33,35,39,72],"rel":-0.8078,"reviewed":true,"share":[1.881,1.5538,0.9815,0.8,0.5602,0.4986,0.3935,0.3158,0.3615],"size":640,"slope":-0.001878},{"claims":[0,9,2,23,26,72,84,183,230],"deltaPp":0.6175,"id":71,"keywords":["conformal","prediction","coverage","sets","quantile"],"label":"Conformal Prediction","labelFull":"Conformal Prediction for Efficient, Small Prediction Sets","papers":316,"papersByYear":[0,3,2,9,13,32,43,91,123],"rel":null,"reviewed":true,"share":[0.0,0.111,0.0545,0.2057,0.2601,0.4835,0.4835,0.737,0.6175],"size":629,"slope":0.000914},{"claims":[43,46,46,60,68,74,69,74,132],"deltaPp":-1.1963,"id":72,"keywords":["posterior","bayesian","inference","distributions","neural"],"label":"Bayesian Deep Learning","labelFull":"Posterior Approximation for Bayesian Deep Learning","papers":470,"papersByYear":[34,36,31,43,49,56,54,61,106],"rel":-0.6921,"reviewed":true,"share":[1.7285,1.3319,0.8451,0.9829,0.9804,0.846,0.6072,0.494,0.5322],"size":612,"slope":-0.001319},{"claims":[26,35,85,85,81,88,70,45,86],"deltaPp":-0.742,"id":73,"keywords":["imagenet","top-1","accuracy","resnet-50","higher"],"label":"ImageNet Classification Improvements","labelFull":"Techniques for Improving ImageNet Top‑1 Classification Accuracy","papers":521,"papersByYear":[22,33,69,67,70,76,66,43,75],"rel":-0.6634,"reviewed":true,"share":[1.1185,1.2209,1.8811,1.5314,1.4006,1.1482,0.7421,0.3482,0.3765],"size":601,"slope":-0.001375},{"claims":[7,8,18,21,24,43,73,135,249],"deltaPp":0.8001,"id":74,"keywords":["trajectory","prediction"],"label":"Trajectory Prediction","labelFull":"Trajectory Prediction Via Synthetic Data & Diffusion","papers":425,"papersByYear":[5,6,12,13,16,28,51,84,210],"rel":3.1475,"reviewed":true,"share":[0.2542,0.222,0.3272,0.2971,0.3201,0.423,0.5734,0.6803,1.0543],"size":578,"slope":0.000866},{"claims":[8,10,9,9,26,16,77,158,253],"deltaPp":0.6632,"id":75,"keywords":["alignment","compared","benchmarks"],"label":"LLM Alignment","labelFull":"Large Language Model Alignment Methods and Benchmarks","papers":438,"papersByYear":[7,5,8,7,18,13,54,123,203],"rel":1.8634,"reviewed":true,"share":[0.3559,0.185,0.2181,0.16,0.3601,0.1964,0.6072,0.9961,1.0191],"size":566,"slope":0.000983},{"claims":[22,23,36,37,43,64,81,106,142],"deltaPp":-0.2844,"id":76,"keywords":["recommendation","user","collaborative filtering","item","systems"],"label":"Recommender Systems","labelFull":"Benchmark Datasets and Model Enhancements for Recommender Tasks","papers":296,"papersByYear":[13,14,23,20,21,39,37,54,75],"rel":-0.4303,"reviewed":true,"share":[0.6609,0.5179,0.627,0.4571,0.4202,0.5892,0.416,0.4373,0.3765],"size":554,"slope":-0.000278},{"claims":[11,16,27,37,50,68,87,112,137],"deltaPp":0.0985,"id":77,"keywords":["gpu","reduces","pretraining","compared","memory"],"label":"Efficient Training","labelFull":"Efficient Training of Large-Scale Neural Models","papers":513,"papersByYear":[11,16,26,37,46,64,81,101,131],"rel":0.1761,"reviewed":true,"share":[0.5592,0.5919,0.7088,0.8457,0.9204,0.9669,0.9107,0.8179,0.6577],"size":545,"slope":0.000266},{"claims":[13,17,39,44,66,77,66,102,119],"deltaPp":0.0263,"id":78,"keywords":["explanations","explainability","faithfulness","local","feature"],"label":"Post-hoc Explanations","labelFull":"Post-hoc Explanation Methods for Machine Learning Models","papers":307,"papersByYear":[6,9,20,27,39,42,41,57,66],"rel":0.0862,"reviewed":true,"share":[0.305,0.333,0.5453,0.6171,0.7803,0.6345,0.461,0.4616,0.3313],"size":543,"slope":5.7e-05},{"claims":[2,8,14,7,19,26,64,117,277],"deltaPp":1.1433,"id":79,"keywords":["generation","quality","conditional","maintaining","baselines"],"label":"Efficient Text Generation","labelFull":"Efficient Text Generation with Quality Preservation","papers":478,"papersByYear":[2,8,13,7,17,22,55,106,248],"rel":11.2419,"reviewed":true,"share":[0.1017,0.296,0.3544,0.16,0.3401,0.3324,0.6184,0.8584,1.245],"size":534,"slope":0.00116},{"claims":[0,1,0,14,16,21,68,161,247],"deltaPp":0.9539,"id":80,"keywords":["tokens","visual","image","vision","video"],"label":"Visual Token Compression","labelFull":"Visual Token Compression for Efficient Vision Transformers","papers":378,"papersByYear":[0,1,0,11,13,16,45,102,190],"rel":null,"reviewed":true,"share":[0.0,0.037,0.0,0.2514,0.2601,0.2417,0.506,0.826,0.9539],"size":528,"slope":0.001197},{"claims":[0,0,0,0,1,1,65,172,288],"deltaPp":0.8233,"id":81,"keywords":["hallucination","vision-language","reduces","detection","object"],"label":"Vision-Language Hallucination Mitigation","labelFull":"Hallucination Detection and Mitigation in Vision-Language Models","papers":295,"papersByYear":[0,0,0,0,1,1,34,95,164],"rel":null,"reviewed":true,"share":[0.0,0.0,0.0,0.0,0.02,0.0151,0.3823,0.7694,0.8233],"size":527,"slope":0.001064},{"claims":[42,25,41,63,69,60,65,56,100],"deltaPp":-0.553,"id":82,"keywords":["tensor","decomposition","completion","low-rank","factorization"],"label":"Low-Rank Tensor Decomposition","labelFull":"Low-Rank Tensor Decomposition and Completion Techniques","papers":268,"papersByYear":[17,14,18,30,32,34,32,29,62],"rel":-0.6398,"reviewed":true,"share":[0.8643,0.5179,0.4907,0.6857,0.6403,0.5137,0.3598,0.2349,0.3113],"size":521,"slope":-0.000582},{"claims":[8,10,17,31,30,69,79,130,141],"deltaPp":0.1989,"id":83,"keywords":["quantum","classical","circuits","states","algorithm"],"label":"Quantum Machine Learning","labelFull":"Quantum Algorithms for Machine Learning and Optimization","papers":213,"papersByYear":[3,6,7,11,12,29,31,44,70],"rel":1.3043,"reviewed":true,"share":[0.1525,0.222,0.1908,0.2514,0.2401,0.4381,0.3485,0.3563,0.3514],"size":515,"slope":0.000283}],"totals":{"claims":[9215,12672,17344,20785,23990,31995,43200,61665,103200],"papers":[1967,2703,3668,4375,4998,6619,8894,12348,19919]},"years":[2018,2019,2020,2021,2022,2023,2024,2025,2026]}