{"noiseShare":0.5384,"topics":[{"claims":[76,111,204,251,352,419,717,844,916],"deltaPp":2.2971,"id":0,"keywords":["3d","pose","point","rendering","reconstruction"],"label":"3D Gaussian Splatting","labelFull":"Neural 3D Gaussian Splatting for Reconstruction and Rendering","papers":1595,"papersByYear":[31,47,78,80,134,178,271,345,431],"rel":0.7477,"reviewed":true,"share":[3.0723,3.2913,4.1118,3.4276,5.0168,5.5314,6.7179,6.4837,5.3694],"size":3890,"slope":0.004347},{"claims":[95,250,327,385,452,546,533,592,632],"deltaPp":-0.6359,"id":1,"keywords":["graph","node","networks","neural","classification"],"label":"Graph Neural Networks","labelFull":"Graph Neural Network Architectures for Classification","papers":1577,"papersByYear":[44,94,136,164,177,205,212,246,299],"rel":-0.1458,"reviewed":true,"share":[4.3608,6.5826,7.1692,7.0266,6.6267,6.3704,5.2553,4.6232,3.7249],"size":3812,"slope":-0.002151},{"claims":[230,305,346,311,375,313,259,290,357],"deltaPp":-8.4692,"id":2,"keywords":["networks","neural","relu","deep","activation"],"label":"Efficient Deep Network Training","labelFull":"Efficient Training and Gradient Optimization for Deep Nets","papers":1631,"papersByYear":[116,154,184,184,215,174,169,192,243],"rel":-0.7367,"reviewed":true,"share":[11.4965,10.7843,9.6995,7.8835,8.0494,5.4071,4.1894,3.6083,3.0273],"size":2786,"slope":-0.011484},{"claims":[96,136,230,301,330,335,328,366,516],"deltaPp":-1.375,"id":3,"keywords":["reinforcement","learning","offline","imitation","policy"],"label":"Offline Reinforcement Learning","labelFull":"Offline Reinforcement Learning (Batch RL)","papers":1640,"papersByYear":[59,83,131,171,195,204,199,239,359],"rel":-0.2351,"reviewed":true,"share":[5.8474,5.8123,6.9056,7.3265,7.3006,6.3393,4.9331,4.4916,4.4724],"size":2638,"slope":-0.002399},{"claims":[92,171,221,256,337,319,375,350,514],"deltaPp":-1.641,"id":4,"keywords":["adversarial","attacks","robustness","backdoor","perturbations"],"label":"Adversarial Robustness","labelFull":"Adversarial Robustness and Defense Techniques","papers":1267,"papersByYear":[51,85,101,115,137,142,170,192,274],"rel":-0.3247,"reviewed":true,"share":[5.0545,5.9524,5.3242,4.9272,5.1292,4.4127,4.2142,3.6083,3.4135],"size":2635,"slope":-0.002722},{"claims":[33,51,87,108,122,275,375,561,939],"deltaPp":3.1121,"id":5,"keywords":["protein","molecular","prediction","generation","structure"],"label":"Molecular Deep Learning","labelFull":"Deep Learning for Molecular and Protein Modeling","papers":1042,"papersByYear":[18,24,38,47,58,108,137,219,393],"rel":1.7445,"reviewed":true,"share":[1.7839,1.6807,2.0032,2.0137,2.1715,3.3561,3.3961,4.1158,4.896],"size":2551,"slope":0.00398},{"claims":[55,79,148,214,149,278,280,295,415],"deltaPp":-0.0105,"id":6,"keywords":["causal","treatment","counterfactual","effect","variables"],"label":"Causal Discovery and Inference","labelFull":"Causal Discovery and Inference for Effect Estimation","papers":839,"papersByYear":[26,31,55,87,71,119,111,133,206],"rel":-0.0041,"reviewed":true,"share":[2.5768,2.1709,2.8993,3.7275,2.6582,3.6979,2.7516,2.4995,2.5663],"size":1913,"slope":0.000103},{"claims":[26,35,39,46,57,160,412,453,568],"deltaPp":3.6835,"id":7,"keywords":["translation","modeling","fine-tuning","enables","framework"],"label":"LLM Fine-Tuning","labelFull":"Fine-Tuning and Training Techniques for Large Language Models","papers":1259,"papersByYear":[15,24,25,33,43,103,278,323,415],"rel":2.4778,"reviewed":true,"share":[1.4866,1.6807,1.3179,1.4139,1.6099,3.2007,6.8914,6.0703,5.1701],"size":1796,"slope":0.006806},{"claims":[51,65,133,135,141,160,166,254,425],"deltaPp":0.8106,"id":8,"keywords":["langevin","dynamics","pde","equations","differential"],"label":"Neural PDE Solvers","labelFull":"Neural PDE Solvers and Physics-Informed Learning","papers":839,"papersByYear":[24,32,66,71,77,90,91,132,256],"rel":0.3408,"reviewed":true,"share":[2.3786,2.2409,3.4792,3.042,2.8828,2.7968,2.2558,2.4807,3.1892],"size":1530,"slope":0.000212},{"claims":[128,129,161,158,209,158,184,163,204],"deltaPp":-5.3928,"id":9,"keywords":["gradient","stochastic","convex","descent","convergence"],"label":"Stochastic Gradient Convergence","labelFull":"Convergence Guarantees for Stochastic Gradient Methods","papers":876,"papersByYear":[70,73,86,95,116,96,105,111,124],"rel":-0.7773,"reviewed":true,"share":[6.9376,5.112,4.5335,4.0703,4.3429,2.9832,2.6029,2.0861,1.5448],"size":1494,"slope":-0.005933},{"claims":[11,21,11,19,36,60,153,511,648],"deltaPp":4.5007,"id":10,"keywords":["reasoning","mathematical","chain-of-thought","benchmarks","accuracy"],"label":"Efficient LLM Reasoning","labelFull":"Efficient Reasoning for Large Language Models","papers":875,"papersByYear":[6,9,6,17,20,39,87,282,409],"rel":7.5693,"reviewed":true,"share":[0.5946,0.6303,0.3163,0.7284,0.7488,1.2119,2.1567,5.2998,5.0953],"size":1470,"slope":0.006029},{"claims":[28,33,34,37,66,97,149,306,663],"deltaPp":3.7116,"id":11,"keywords":["cache","kv","memory","gpu","latency"],"label":"Efficient Transformer Inference","labelFull":"Memory and Latency Optimizations for Transformer Inference","papers":1017,"papersByYear":[21,27,25,29,46,73,115,216,465],"rel":1.7833,"reviewed":true,"share":[2.0813,1.8908,1.3179,1.2425,1.7222,2.2685,2.8508,4.0594,5.7929],"size":1413,"slope":0.004241},{"claims":[42,56,128,157,187,206,153,186,143],"deltaPp":-0.9459,"id":12,"keywords":["label","contrastive","semi-supervised","learning","feature"],"label":"Label-Efficient Learning","labelFull":"Label-Efficient Learning via Pseudo-Labels & Weak Supervision","papers":819,"papersByYear":[24,42,70,97,120,129,98,124,115],"rel":-0.3977,"reviewed":true,"share":[2.3786,2.9412,3.69,4.156,4.4927,4.0087,2.4294,2.3304,1.4327],"size":1258,"slope":-0.001381},{"claims":[31,46,42,61,66,124,188,278,408],"deltaPp":0.7455,"id":13,"keywords":["time series","forecasting","multivariate","temporal","autoregressive"],"label":"Time Series Forecasting","labelFull":"Neural Time Series Forecasting with Foundation Models","papers":621,"papersByYear":[19,26,26,34,33,57,84,131,211],"rel":0.3959,"reviewed":true,"share":[1.8831,1.8207,1.3706,1.4567,1.2355,1.7713,2.0823,2.4619,2.6286],"size":1244,"slope":0.001107},{"claims":[22,46,48,63,67,46,138,207,586],"deltaPp":2.3771,"id":14,"keywords":["agents","llm","planning","games","systems"],"label":"LLM Multi-Agent Coordination","labelFull":"LLM Multi-Agent Systems for Efficient Coordination","papers":750,"papersByYear":[18,28,39,41,51,35,81,123,334],"rel":1.3325,"reviewed":true,"share":[1.7839,1.9608,2.0559,1.7566,1.9094,1.0876,2.0079,2.3116,4.161],"size":1223,"slope":0.001633},{"claims":[15,26,46,79,67,77,188,274,415],"deltaPp":2.086,"id":15,"keywords":["attention","heads","sparse","softmax","linear"],"label":"Subquadratic Attention","labelFull":"Subquadratic Attention Mechanisms for Transformers","papers":686,"papersByYear":[10,18,20,48,41,52,112,138,247],"rel":2.1047,"reviewed":true,"share":[0.9911,1.2605,1.0543,2.0566,1.535,1.6159,2.7764,2.5935,3.0771],"size":1187,"slope":0.002558},{"claims":[2,4,6,17,26,147,291,316,315],"deltaPp":2.5052,"id":16,"keywords":["diffusion","guidance","sampling","discrete","process"],"label":"Fast Diffusion Sampling","labelFull":"Fast Sampling and Acceleration Techniques for Diffusion Models","papers":706,"papersByYear":[2,4,5,14,16,91,168,189,217],"rel":12.6398,"reviewed":true,"share":[0.1982,0.2801,0.2636,0.5998,0.599,2.8278,4.1646,3.552,2.7034],"size":1124,"slope":0.004978},{"claims":[49,79,76,129,143,190,147,134,145],"deltaPp":-0.9118,"id":17,"keywords":["privacy","private","differential","algorithms","guarantees"],"label":"Differential Privacy Algorithms","labelFull":"Algorithmic and Theoretical Advances in Differential Privacy","papers":469,"papersByYear":[18,36,33,48,52,83,65,64,70],"rel":-0.5111,"reviewed":true,"share":[1.7839,2.521,1.7396,2.0566,1.9468,2.5792,1.6113,1.2028,0.8721],"size":1092,"slope":-0.001223},{"claims":[35,68,106,115,144,125,132,123,186],"deltaPp":-0.5869,"id":18,"keywords":["regret","bound","online","sqrt","algorithm"],"label":"Online Regret Minimization","labelFull":"Regret Minimization and Bounds in Online Learning Algorithms","papers":634,"papersByYear":[20,45,67,69,84,86,81,70,112],"rel":-0.2961,"reviewed":true,"share":[1.9822,3.1513,3.5319,2.9563,3.1449,2.6725,2.0079,1.3155,1.3953],"size":1034,"slope":-0.001864},{"claims":[37,30,45,78,84,107,128,195,323],"deltaPp":0.2558,"id":19,"keywords":["audio","speech","music","acoustic","sound"],"label":"Audio Foundation Models","labelFull":"Unified Audio Foundation Models for Speech, Music, and Sound","papers":356,"papersByYear":[12,13,15,27,28,37,42,66,116],"rel":0.2151,"reviewed":true,"share":[1.1893,0.9104,0.7907,1.1568,1.0483,1.1498,1.0412,1.2404,1.4451],"size":1027,"slope":0.000418},{"claims":[34,37,62,87,97,140,125,189,218],"deltaPp":-0.7286,"id":20,"keywords":["state-of-the-art","synthetic","benchmark","real-world","experiments"],"label":"Performance Benchmarking","labelFull":"Empirical Performance Benchmarking on Standard Datasets","papers":962,"papersByYear":[34,37,60,84,94,136,120,185,212],"rel":-0.2162,"reviewed":true,"share":[3.3697,2.591,3.1629,3.599,3.5193,4.2262,2.9747,3.4768,2.6411],"size":989,"slope":-1e-06},{"claims":[38,95,89,127,129,130,139,133,108],"deltaPp":-1.1973,"id":21,"keywords":["bandits","regret","contextual","multi-armed","feedback"],"label":"Bandit Regret Bounds","labelFull":"Regret Bounds for Stochastic/Contextual/Adversarial Bandits","papers":521,"papersByYear":[20,47,45,69,68,70,68,71,63],"rel":-0.604,"reviewed":true,"share":[1.9822,3.2913,2.3722,2.9563,2.5459,2.1753,1.6857,1.3343,0.7849],"size":988,"slope":-0.002136},{"claims":[2,14,11,24,28,82,165,323,336],"deltaPp":2.6292,"id":22,"keywords":["multimodal","modality","visual","reasoning","alignment"],"label":"Multimodal LLM Adaptation","labelFull":"Adaptation and Alignment for Multimodal LLMs","papers":591,"papersByYear":[1,9,7,14,19,44,89,189,219],"rel":26.5308,"reviewed":true,"share":[0.0991,0.6303,0.369,0.5998,0.7113,1.3673,2.2062,3.552,2.7283],"size":985,"slope":0.003954},{"claims":[44,71,100,104,110,113,94,116,220],"deltaPp":-1.3272,"id":23,"keywords":["estimator","variance","error","asymptotic","mean"],"label":"Robust Mean Estimation","labelFull":"Robust Mean Estimation with Finite-sample Guarantees","papers":678,"papersByYear":[33,47,70,72,75,80,69,76,156],"rel":-0.4058,"reviewed":true,"share":[3.2706,3.2913,3.69,3.0848,2.8079,2.486,1.7105,1.4283,1.9434],"size":972,"slope":-0.002576},{"claims":[24,59,76,101,95,111,114,113,239],"deltaPp":0.1102,"id":24,"keywords":["policy","markov decision","decision processes","mdps","gradient"],"label":"Policy Optimization","labelFull":"Policy Optimization Methods for Reinforcement Learning","papers":619,"papersByYear":[19,41,55,71,60,74,63,76,160],"rel":0.0585,"reviewed":true,"share":[1.8831,2.8711,2.8993,3.042,2.2463,2.2996,1.5617,1.4283,1.9933],"size":932,"slope":-0.001218},{"claims":[11,7,38,60,110,82,112,153,308],"deltaPp":1.6977,"id":25,"keywords":["distillation","teacher","student","knowledge","on-policy"],"label":"Knowledge Distillation","labelFull":"Efficient Model Compression Using Knowledge Distillation","papers":504,"papersByYear":[6,4,18,29,51,48,70,94,184],"rel":2.8552,"reviewed":true,"share":[0.5946,0.2801,0.9489,1.2425,1.9094,1.4916,1.7353,1.7666,2.2923],"size":881,"slope":0.002179},{"claims":[22,21,39,64,54,102,167,184,228],"deltaPp":0.9399,"id":26,"keywords":["image","restoration","denoising","super-resolution","reconstruction"],"label":"Image Restoration","labelFull":"Image Restoration: Denoising, Super-Resolution, Deblurring","papers":521,"papersByYear":[10,13,21,30,30,54,94,114,155],"rel":0.9483,"reviewed":true,"share":[0.9911,0.9104,1.107,1.2853,1.1232,1.6781,2.3302,2.1425,1.931],"size":881,"slope":0.001716},{"claims":[2,2,3,14,13,43,108,216,465],"deltaPp":3.888,"id":27,"keywords":["token","attention","transformer","prediction","representations"],"label":"Dynamic Token Pruning","labelFull":"Dynamic Token Pruning for Efficient Transformer Inference","papers":610,"papersByYear":[2,2,3,12,9,33,70,151,328],"rel":19.6165,"reviewed":true,"share":[0.1982,0.1401,0.1581,0.5141,0.337,1.0255,1.7353,2.8378,4.0862],"size":866,"slope":0.004552},{"claims":[15,29,61,81,135,124,149,129,119],"deltaPp":-0.2544,"id":28,"keywords":["segmentation","semantic","instance","object","image"],"label":"Image Segmentation","labelFull":"Image Segmentation (Semantic, Instance, Unsupervised)","papers":505,"papersByYear":[13,17,39,47,76,65,85,80,83],"rel":-0.1975,"reviewed":true,"share":[1.2884,1.1905,2.0559,2.0137,2.8454,2.0199,2.1071,1.5035,1.034],"size":842,"slope":5e-06},{"claims":[13,23,48,51,60,67,102,184,284],"deltaPp":1.4886,"id":29,"keywords":["manipulation","navigation","robotic","locomotion","success"],"label":"Goal-Conditioned Robotic RL","labelFull":"Goal-Conditioned Reinforcement Learning for Robotic Manipulation & Locomotion","papers":574,"papersByYear":[11,18,26,35,42,50,66,119,207],"rel":1.3654,"reviewed":true,"share":[1.0902,1.2605,1.3706,1.4996,1.5724,1.5538,1.6361,2.2364,2.5788],"size":832,"slope":0.001578},{"claims":[0,2,1,5,23,122,221,208,210],"deltaPp":2.0929,"id":30,"keywords":["diffusion","text-to-image","denoising","video","generation"],"label":"Efficient Diffusion Models","labelFull":"Efficient Training and Sampling of Diffusion Models","papers":580,"papersByYear":[0,2,1,4,15,87,145,158,168],"rel":null,"reviewed":true,"share":[0.0,0.1401,0.0527,0.1714,0.5616,2.7035,3.5944,2.9694,2.0929],"size":792,"slope":0.004412},{"claims":[41,67,102,106,107,91,88,60,122],"deltaPp":-2.1607,"id":31,"keywords":["imagenet","cifar-10","accuracy","cifar-100","top-1"],"label":"Image Classification Accuracy","labelFull":"Top-1 Accuracy Gains on ImageNet, CIFAR, and Tiny-ImageNet","papers":674,"papersByYear":[35,59,83,86,88,82,80,56,105],"rel":-0.6229,"reviewed":true,"share":[3.4688,4.1317,4.3753,3.6847,3.2946,2.5482,1.9831,1.0524,1.3081],"size":784,"slope":-0.003967},{"claims":[47,66,66,86,81,88,90,99,152],"deltaPp":-2.0729,"id":32,"keywords":["complexity","algorithm","lower","bound","log"],"label":"Non-Asymptotic Complexity Bounds","labelFull":"Provable Non-Asymptotic Complexity Bounds for ML Algorithms","papers":628,"papersByYear":[36,53,58,71,71,75,72,72,120],"rel":-0.581,"reviewed":true,"share":[3.5679,3.7115,3.0575,3.042,2.6582,2.3306,1.7848,1.3531,1.495],"size":775,"slope":-0.003104},{"claims":[32,34,58,55,82,79,105,131,166],"deltaPp":-0.5024,"id":33,"keywords":["spiking","neural","neurons","synaptic","plasticity"],"label":"Spiking Neural Network Training","labelFull":"Gradient-Based Training of Spiking Neural Networks","papers":349,"papersByYear":[15,20,26,27,35,40,46,61,79],"rel":-0.338,"reviewed":true,"share":[1.4866,1.4006,1.3706,1.1568,1.3104,1.243,1.1403,1.1464,0.9842],"size":742,"slope":-0.000524},{"claims":[15,23,32,53,53,106,95,140,171],"deltaPp":0.2054,"id":34,"keywords":["combinatorial","vehicle routing","pareto","solutions","optimization"],"label":"Neural Combinatorial Optimization","labelFull":"Learning-Based Methods for Combinatorial Optimization","papers":409,"papersByYear":[11,16,22,28,37,50,58,83,104],"rel":0.1884,"reviewed":true,"share":[1.0902,1.1204,1.1597,1.1997,1.3852,1.5538,1.4378,1.5599,1.2956],"size":688,"slope":0.000508},{"claims":[24,16,27,27,57,64,75,104,266],"deltaPp":0.8181,"id":35,"keywords":["explanations","evidence","human","evaluation","theoretical"],"label":"Faithful Explanations","labelFull":"Faithful Explanations: Theory, Evaluation, and Improvement","papers":449,"papersByYear":[15,10,16,16,42,45,49,71,185],"rel":0.5503,"reviewed":true,"share":[1.4866,0.7003,0.8434,0.6855,1.5724,1.3984,1.2147,1.3343,2.3047],"size":660,"slope":0.001105},{"claims":[13,16,35,89,82,87,131,88,108],"deltaPp":-0.0188,"id":36,"keywords":["vision","transformers","visual","image","features"],"label":"Efficient Vision Transformers","labelFull":"Efficient Vision Transformers: Sparse & Parameter-Efficient","papers":477,"papersByYear":[11,14,29,58,64,64,84,67,86],"rel":-0.0172,"reviewed":true,"share":[1.0902,0.9804,1.5287,2.485,2.3961,1.9888,2.0823,1.2592,1.0714],"size":649,"slope":0.000229},{"claims":[44,70,49,66,38,65,59,82,125],"deltaPp":-1.493,"id":37,"keywords":["low-rank","matrix","recovery","sensing","matrices"],"label":"Low-Rank Matrix Recovery","labelFull":"Low-Rank Matrix Recovery, Sensing, and Completion Techniques","papers":362,"papersByYear":[26,35,30,31,23,37,36,57,87],"rel":-0.5794,"reviewed":true,"share":[2.5768,2.451,1.5814,1.3282,0.8611,1.1498,0.8924,1.0712,1.0838],"size":598,"slope":-0.001945},{"claims":[0,0,32,64,71,96,140,105,86],"deltaPp":0.598,"id":38,"keywords":["federated","learning","client","personalized","heterogeneous"],"label":"Heterogeneous Federated Learning","labelFull":"Robust Personalized Federated Learning for Data Heterogeneity","papers":301,"papersByYear":[0,0,12,28,39,50,68,56,48],"rel":null,"reviewed":true,"share":[0.0,0.0,0.6326,1.1997,1.4601,1.5538,1.6857,1.0524,0.598],"size":594,"slope":0.001335},{"claims":[33,50,61,76,76,102,80,46,63],"deltaPp":-1.5077,"id":39,"keywords":["boosting","loss","surrogate","classification","ensemble"],"label":"Margin Theory of Boosting","labelFull":"Boosting, Margins, and Generalization Theory","papers":407,"papersByYear":[22,35,39,50,58,62,49,38,54],"rel":-0.6915,"reviewed":true,"share":[2.1804,2.451,2.0559,2.1422,2.1715,1.9267,1.2147,0.7142,0.6727],"size":587,"slope":-0.00219},{"claims":[18,47,59,51,80,89,67,83,92],"deltaPp":-0.2566,"id":40,"keywords":["games","equilibrium","nash","equilibria","zero-sum"],"label":"Equilibrium Computation in Games","labelFull":"Algorithmic Learning of Nash and Correlated Equilibria","papers":282,"papersByYear":[9,18,28,26,35,35,40,40,51],"rel":-0.2877,"reviewed":true,"share":[0.892,1.2605,1.476,1.114,1.3104,1.0876,0.9916,0.7517,0.6354],"size":586,"slope":-0.000591},{"claims":[24,39,40,39,53,56,69,76,142],"deltaPp":-0.6118,"id":41,"keywords":["generative","score-based","modeling","latent","deep"],"label":"One-Step Generative Modeling","labelFull":"One-Step Generative Modeling and Efficient Sampling","papers":412,"papersByYear":[20,29,29,34,38,39,47,66,110],"rel":-0.3086,"reviewed":true,"share":[1.9822,2.0308,1.5287,1.4567,1.4227,1.2119,1.1651,1.2404,1.3704],"size":538,"slope":-0.000965},{"claims":[23,32,28,51,68,82,58,78,78],"deltaPp":-0.3665,"id":42,"keywords":["few-shot","zero-shot","classification","learning","clip"],"label":"Few-Shot Classification","labelFull":"Few-Shot and Zero-Shot Classification","papers":380,"papersByYear":[13,20,19,36,46,57,46,69,74],"rel":-0.2845,"reviewed":true,"share":[1.2884,1.4006,1.0016,1.5424,1.7222,1.7713,1.1403,1.2967,0.9219],"size":498,"slope":-0.000212},{"claims":[57,61,74,79,57,51,31,41,45],"deltaPp":-3.4915,"id":43,"keywords":["variational","inference","autoencoders","posterior","gaussian"],"label":"Deep Variational Inference","labelFull":"Variational Inference for Deep Probabilistic Models","papers":330,"papersByYear":[39,37,44,51,40,35,25,29,30],"rel":-0.9033,"reviewed":true,"share":[3.8652,2.591,2.3195,2.1851,1.4976,1.0876,0.6197,0.545,0.3737],"size":496,"slope":-0.0041},{"claims":[11,9,9,22,12,30,67,103,215],"deltaPp":1.3742,"id":44,"keywords":["question","answering","retrieval-augmented generation","qa","multi-hop"],"label":"Retrieval-Augmented QA","labelFull":"Retrieval-Augmented Generation for Question Answering","papers":361,"papersByYear":[7,8,7,12,10,24,45,82,166],"rel":1.9807,"reviewed":true,"share":[0.6938,0.5602,0.369,0.5141,0.3744,0.7458,1.1155,1.5411,2.068],"size":478,"slope":0.001694},{"claims":[20,23,23,24,20,25,42,89,212],"deltaPp":0.7791,"id":45,"keywords":["code","verification","program","synthesis","execution"],"label":"Verified Code Generation","labelFull":"Execution-Driven Verification for Code Generation","papers":329,"papersByYear":[12,11,17,13,13,15,31,59,158],"rel":0.6551,"reviewed":true,"share":[1.1893,0.7703,0.8962,0.557,0.4867,0.4661,0.7685,1.1088,1.9684],"size":478,"slope":0.000631},{"claims":[0,1,3,9,26,42,105,135,142],"deltaPp":1.4451,"id":46,"keywords":["vision-language","visual","benchmarks","adaptation","image"],"label":"Efficient Vision-Language Modeling","labelFull":"Efficient Training and Adaptation of Vision-Language Models","papers":358,"papersByYear":[0,1,3,8,18,32,78,102,116],"rel":null,"reviewed":true,"share":[0.0,0.07,0.1581,0.3428,0.6739,0.9944,1.9336,1.9169,1.4451],"size":463,"slope":0.002587},{"claims":[15,25,20,45,49,79,55,64,108],"deltaPp":0.1669,"id":47,"keywords":["manifold","equivariant","group","symmetries","geometric"],"label":"Group Equivariant Manifold Learning","labelFull":"Group Equivariant Neural Networks for Learning on Manifolds","papers":303,"papersByYear":[9,10,15,27,30,41,40,46,85],"rel":0.1871,"reviewed":true,"share":[0.892,0.7003,0.7907,1.1568,1.1232,1.2741,0.9916,0.8645,1.0589],"size":460,"slope":0.00028},{"claims":[5,19,21,12,20,47,45,124,166],"deltaPp":0.6116,"id":48,"keywords":["eeg","fmri","brain","activity","signals"],"label":"EEG Foundation Models","labelFull":"Foundation Models for EEG & Multimodal Brain Signal Learning","papers":193,"papersByYear":[2,11,8,8,11,23,18,47,65],"rel":3.0858,"reviewed":true,"share":[0.1982,0.7703,0.4217,0.3428,0.4118,0.7147,0.4462,0.8833,0.8098],"size":459,"slope":0.000534},{"claims":[2,6,28,14,12,17,31,90,248],"deltaPp":1.9321,"id":49,"keywords":["flow","matching","velocity","normalizing","rectified"],"label":"Flow Matching","labelFull":"Flow Matching for Training Normalizing and Continuous Flows","papers":305,"papersByYear":[2,5,16,12,9,11,20,59,171],"rel":9.7482,"reviewed":true,"share":[0.1982,0.3501,0.8434,0.5141,0.337,0.3418,0.4958,1.1088,2.1303],"size":448,"slope":0.001523},{"claims":[26,46,49,51,43,74,40,51,58],"deltaPp":-0.8773,"id":50,"keywords":["clustering","k-means","algorithm","approximation","k-median"],"label":"Scalable Approximate Clustering","labelFull":"Scalable Approximate Clustering Algorithms for k-Means and k-Median","papers":200,"papersByYear":[13,16,19,26,19,35,15,24,33],"rel":-0.6809,"reviewed":true,"share":[1.2884,1.1204,1.0016,1.114,0.7113,1.0876,0.3718,0.451,0.4111],"size":438,"slope":-0.001134},{"claims":[25,32,53,42,44,67,54,49,70],"deltaPp":-1.0245,"id":51,"keywords":["kernel","ridge","regression","distributionally robust","robust optimization"],"label":"Kernel Methods","labelFull":"Kernel Methods for Learning, Approximation, and Optimization","papers":303,"papersByYear":[17,25,35,32,30,45,32,34,53],"rel":-0.6081,"reviewed":true,"share":[1.6848,1.7507,1.845,1.371,1.1232,1.3984,0.7933,0.639,0.6603],"size":436,"slope":-0.001585},{"claims":[9,32,38,48,77,39,41,68,60],"deltaPp":-0.1337,"id":52,"keywords":["fairness","group","demographic","parity","guarantees"],"label":"Fairness-Accuracy Tradeoff","labelFull":"Fairness-Accuracy Tradeoff Optimization with Guarantees","papers":219,"papersByYear":[6,15,17,28,35,20,24,37,37],"rel":-0.2249,"reviewed":true,"share":[0.5946,1.0504,0.8962,1.1997,1.3104,0.6215,0.5949,0.6954,0.4609],"size":412,"slope":-0.000463},{"claims":[4,22,20,17,38,35,50,65,141],"deltaPp":0.7616,"id":53,"keywords":["quantization","quantized","post-training","bit","vector"],"label":"Post-Training Quantization","labelFull":"Post-Training Quantization for Efficient Low-Bit Inference","papers":235,"papersByYear":[3,11,10,10,23,21,33,39,85],"rel":2.5617,"reviewed":true,"share":[0.2973,0.7703,0.5271,0.4284,0.8611,0.6526,0.818,0.7329,1.0589],"size":392,"slope":0.000623},{"claims":[26,37,30,41,40,42,62,59,44],"deltaPp":-1.7812,"id":54,"keywords":["numerical","convergence","stationary","theoretical","experiments"],"label":"Optimization Convergence Guarantees","labelFull":"Convergence Guarantees for Optimization Algorithms","papers":337,"papersByYear":[23,27,27,36,38,37,57,52,40],"rel":-0.7814,"reviewed":true,"share":[2.2795,1.8908,1.4233,1.5424,1.4227,1.1498,1.413,0.9773,0.4983],"size":381,"slope":-0.001713},{"claims":[34,42,24,36,44,52,28,42,71],"deltaPp":-1.6311,"id":55,"keywords":["posterior","gaussian process","bayesian","likelihood","inference"],"label":"Black-Box Bayesian Inference","labelFull":"Posterior Approximation via Black-Box Bayesian Inference","papers":283,"papersByYear":[24,28,17,29,33,36,23,33,60],"rel":-0.6857,"reviewed":true,"share":[2.3786,1.9608,0.8962,1.2425,1.2355,1.1187,0.5702,0.6202,0.7475],"size":373,"slope":-0.001887},{"claims":[14,20,34,31,45,35,44,60,90],"deltaPp":-0.1802,"id":56,"keywords":["multi-task","transfer","learning","generalization","adaptation"],"label":"Multi-Task Learning","labelFull":"Multi-Task Learning for Transfer and Adaptation","papers":317,"papersByYear":[12,15,29,26,39,29,34,52,81],"rel":-0.1515,"reviewed":true,"share":[1.1893,1.0504,1.5287,1.114,1.4601,0.9012,0.8428,0.9773,1.0091],"size":373,"slope":-0.000421},{"claims":[11,20,22,21,41,42,68,64,61],"deltaPp":-0.1706,"id":57,"keywords":["detection","anomaly","out-of-distribution","outlier","normal"],"label":"Out-of-Distribution Detection","labelFull":"Out-of-Distribution Detection in Deep Learning Models","papers":200,"papersByYear":[7,10,13,15,20,26,38,29,42],"rel":-0.2459,"reviewed":true,"share":[0.6938,0.7003,0.6853,0.6427,0.7488,0.808,0.942,0.545,0.5232],"size":350,"slope":-7.8e-05},{"claims":[6,29,35,17,29,34,43,55,101],"deltaPp":0.2021,"id":58,"keywords":["transport","optimal","unbalanced","entropic","plan"],"label":"Computational Optimal Transport","labelFull":"Scalable Algorithms and Theory for Optimal Transport Variants","papers":187,"papersByYear":[5,13,18,10,17,21,20,27,56],"rel":0.4079,"reviewed":true,"share":[0.4955,0.9104,0.9489,0.4284,0.6365,0.6526,0.4958,0.5074,0.6976],"size":349,"slope":-0.00018},{"claims":[0,6,11,7,21,18,52,74,155],"deltaPp":1.0216,"id":59,"keywords":["medical","segmentation","pathology","image","slide"],"label":"Medical Image Segmentation","labelFull":"Medical Image Segmentation with Foundation Models","papers":191,"papersByYear":[0,2,7,6,12,10,26,46,82],"rel":null,"reviewed":true,"share":[0.0,0.1401,0.369,0.2571,0.4493,0.3108,0.6445,0.8645,1.0216],"size":344,"slope":0.001144},{"claims":[27,8,30,53,47,35,42,44,42],"deltaPp":-1.0506,"id":60,"keywords":["domain","adaptation","generalization","source","target"],"label":"Unsupervised Domain Adaptation","labelFull":"Unsupervised Domain Adaptation and Domain Generalization","papers":224,"papersByYear":[15,5,15,29,31,26,34,34,35],"rel":-0.7067,"reviewed":true,"share":[1.4866,0.3501,0.7907,1.2425,1.1606,0.808,0.8428,0.639,0.436],"size":328,"slope":-0.000611},{"claims":[0,0,5,1,14,19,66,104,118],"deltaPp":0.4983,"id":61,"keywords":["watermark","image","detection","robustness","text"],"label":"Robust Generative Watermarking","labelFull":"Robust Watermarking Techniques for Generative AI Models","papers":108,"papersByYear":[0,0,1,1,5,7,22,32,40],"rel":null,"reviewed":true,"share":[0.0,0.0,0.0527,0.0428,0.1872,0.2175,0.5454,0.6014,0.4983],"size":327,"slope":0.000826},{"claims":[1,0,0,1,2,17,21,70,214],"deltaPp":1.3834,"id":62,"keywords":["diffusion","masked","autoregressive","tokens","decoding"],"label":"Diffusion Language Modeling","labelFull":"Diffusion Language Modeling: Training and Efficient Generation","papers":190,"papersByYear":[1,0,0,1,1,7,16,45,119],"rel":13.9596,"reviewed":true,"share":[0.0991,0.0,0.0,0.0428,0.0374,0.2175,0.3966,0.8457,1.4825],"size":326,"slope":0.001506},{"claims":[18,13,16,17,34,31,57,58,81],"deltaPp":-0.2697,"id":63,"keywords":["recommendation","user","item","collaborative","systems"],"label":"Recommender Systems","labelFull":"Recommender System Models and Performance Enhancements","papers":168,"papersByYear":[8,6,11,10,14,17,27,33,42],"rel":-0.3401,"reviewed":true,"share":[0.7929,0.4202,0.5799,0.4284,0.5241,0.5283,0.6693,0.6202,0.5232],"size":325,"slope":-3.3e-05},{"claims":[1,1,3,17,31,34,73,71,94],"deltaPp":0.8228,"id":64,"keywords":["transformers","in-context","linear","layers","architectures"],"label":"Efficient Transformers","labelFull":"Efficient Transformer Architectures and Training","papers":232,"papersByYear":[1,1,3,15,17,22,46,53,74],"rel":8.3027,"reviewed":true,"share":[0.0991,0.07,0.1581,0.6427,0.6365,0.6837,1.1403,0.9961,0.9219],"size":325,"slope":0.001346},{"claims":[13,10,25,33,42,33,56,50,61],"deltaPp":0.0034,"id":65,"keywords":["compression","lossless","image","lossy","coding"],"label":"Neural Compression","labelFull":"Learned Neural Compression for Data and Model Efficiency","papers":206,"papersByYear":[6,8,15,22,22,21,33,31,48],"rel":0.0057,"reviewed":true,"share":[0.5946,0.5602,0.7907,0.9426,0.8237,0.6526,0.818,0.5826,0.598],"size":323,"slope":-2.6e-05},{"claims":[4,12,29,30,27,38,36,73,74],"deltaPp":0.3505,"id":66,"keywords":["continual learning","unlearning","forgetting","catastrophic","learning benchmarks"],"label":"Continual Learning","labelFull":"Continual Learning: Benchmarks and Forgetting Mitigation","papers":224,"papersByYear":[3,10,22,17,18,24,26,52,52],"rel":1.1789,"reviewed":true,"share":[0.2973,0.7003,1.1597,0.7284,0.6739,0.7458,0.6445,0.9773,0.6478],"size":323,"slope":0.000203},{"claims":[12,18,32,34,38,41,44,45,51],"deltaPp":-0.4933,"id":67,"keywords":["detection","object","detectors","coco","state-of-the-art"],"label":"Object Detection","labelFull":"State-of-the-Art Object Detection Methods and Benchmarks","papers":214,"papersByYear":[9,9,19,20,28,29,34,34,32],"rel":-0.553,"reviewed":true,"share":[0.892,0.6303,1.0016,0.8569,1.0483,0.9012,0.8428,0.639,0.3987],"size":315,"slope":-0.00037},{"claims":[0,9,2,10,1,13,17,105,155],"deltaPp":1.2583,"id":68,"keywords":["reasoning","visual","spatial","vision-language","perception"],"label":"Vision-Language Spatial Reasoning","labelFull":"Spatial Reasoning in Vision-Language Models","papers":199,"papersByYear":[0,7,1,8,1,7,12,62,101],"rel":null,"reviewed":true,"share":[0.0,0.4902,0.0527,0.3428,0.0374,0.2175,0.2975,1.1652,1.2583],"size":312,"slope":0.001237},{"claims":[7,3,13,12,21,31,37,58,129],"deltaPp":0.9247,"id":69,"keywords":["trajectory","prediction","inference","learns","framework"],"label":"Trajectory Prediction","labelFull":"Trajectory Prediction and Forecasting Methods","papers":239,"papersByYear":[5,3,11,7,12,20,25,42,114],"rel":1.8662,"reviewed":true,"share":[0.4955,0.2101,0.5799,0.2999,0.4493,0.6215,0.6197,0.7893,1.4202],"size":311,"slope":0.000973},{"claims":[4,18,13,30,41,39,41,40,73],"deltaPp":0.1263,"id":70,"keywords":["auctions","pricing","revenue","welfare","bidding"],"label":"Learning-Augmented Auctions","labelFull":"Learning-Augmented Auction Design with Regret Guarantees","papers":140,"papersByYear":[3,8,5,14,18,16,19,23,34],"rel":0.4248,"reviewed":true,"share":[0.2973,0.5602,0.2636,0.5998,0.6739,0.4972,0.471,0.4322,0.4236],"size":299,"slope":7.2e-05},{"claims":[6,0,3,3,4,43,72,65,98],"deltaPp":0.5623,"id":71,"keywords":["editing","text-to-image","generation","visual","video"],"label":"Controllable Text-to-Image Generation","labelFull":"Controllable Text-to-Image Generation and Editing","papers":217,"papersByYear":[3,0,3,2,3,36,47,54,69],"rel":1.8914,"reviewed":true,"share":[0.2973,0.0,0.1581,0.0857,0.1123,1.1187,1.1651,1.0148,0.8596],"size":294,"slope":0.00139},{"claims":[2,20,22,30,16,25,27,47,95],"deltaPp":0.6614,"id":72,"keywords":["calibration","uncertainty","epistemic","quantification","aleatoric"],"label":"Uncertainty Calibration","labelFull":"Calibrated Predictive Uncertainty in Machine Learning","papers":211,"papersByYear":[2,13,18,19,13,20,20,37,69],"rel":3.337,"reviewed":true,"share":[0.1982,0.9104,0.9489,0.8141,0.4867,0.6215,0.4958,0.6954,0.8596],"size":284,"slope":0.00015},{"claims":[6,11,9,21,27,25,23,63,87],"deltaPp":0.4632,"id":73,"keywords":["video","frames","temporal","memory","long"],"label":"Long-Form Video Modeling","labelFull":"Efficient Long-Form Video Modeling and Prediction","papers":197,"papersByYear":[4,6,7,12,17,17,16,49,69],"rel":1.1685,"reviewed":true,"share":[0.3964,0.4202,0.369,0.5141,0.6365,0.5283,0.3966,0.9209,0.8596],"size":272,"slope":0.000571},{"claims":[3,6,9,10,10,20,45,67,99],"deltaPp":0.7616,"id":74,"keywords":["generation","video","image","synthesis","quality"],"label":"Diffusion Visual Generation","labelFull":"Diffusion-Based Image and Video Generation","papers":224,"papersByYear":[3,4,7,8,8,17,38,54,85],"rel":2.5617,"reviewed":true,"share":[0.2973,0.2801,0.369,0.3428,0.2995,0.5283,0.942,1.0148,1.0589],"size":269,"slope":0.001097},{"claims":[1,0,0,1,8,41,54,66,95],"deltaPp":0.9598,"id":75,"keywords":["prompts","semantic","optimization","learning"],"label":"Prompt Optimization","labelFull":"Prompt Optimization and Selection for Large Language Models","papers":214,"papersByYear":[1,0,0,1,6,27,41,53,85],"rel":9.6852,"reviewed":true,"share":[0.0991,0.0,0.0,0.0428,0.2246,0.839,1.0164,0.9961,1.0589],"size":266,"slope":0.001609},{"claims":[10,17,14,7,34,40,48,58,34],"deltaPp":-0.4446,"id":76,"keywords":["face","facial","recognition","identity","images"],"label":"Identity-Preserving Face Generation","labelFull":"Identity-Preserving Face Generation, Editing, and Synthesis","papers":126,"papersByYear":[7,7,10,4,16,14,22,26,20],"rel":-0.6408,"reviewed":true,"share":[0.6938,0.4902,0.5271,0.1714,0.599,0.4351,0.5454,0.4886,0.2492],"size":262,"slope":-0.000247},{"claims":[0,9,1,11,7,28,37,61,103],"deltaPp":0.6229,"id":77,"keywords":["conformal","prediction","coverage","sets","intervals"],"label":"Conformal Prediction","labelFull":"Conformal Prediction Methods for Valid Prediction Sets","papers":125,"papersByYear":[0,3,1,4,5,14,18,30,50],"rel":null,"reviewed":true,"share":[0.0,0.2101,0.0527,0.1714,0.1872,0.4351,0.4462,0.5638,0.6229],"size":257,"slope":0.000767},{"claims":[12,24,18,36,25,31,28,37,34],"deltaPp":-0.5681,"id":78,"keywords":["recurrent","neural networks","dynamics","memory","networks trained"],"label":"RNN Training and Dynamics","labelFull":"Training Algorithms, Dynamics, and Interpretability of RNNs","papers":144,"papersByYear":[9,14,12,16,14,18,15,20,26],"rel":-0.6369,"reviewed":true,"share":[0.892,0.9804,0.6326,0.6855,0.5241,0.5594,0.3718,0.3759,0.3239],"size":245,"slope":-0.000789},{"claims":[27,17,15,36,30,30,30,31,28],"deltaPp":-0.7544,"id":79,"keywords":["tensor","decomposition","low-rank","completion","approximation"],"label":"Low-Rank Tensor Decomposition","labelFull":"Low-Rank Tensor Decomposition and Completion","papers":121,"papersByYear":[10,8,8,13,17,15,14,17,19],"rel":-0.7612,"reviewed":true,"share":[0.9911,0.5602,0.4217,0.557,0.6365,0.4661,0.3471,0.3195,0.2367],"size":244,"slope":-0.000663},{"claims":[5,3,2,4,3,6,61,97,62],"deltaPp":0.3131,"id":80,"keywords":["preference","direct","optimization","dpo","pairs"],"label":"Direct Preference Optimization","labelFull":"Direct Preference Optimization and Variants for LLMs","papers":164,"papersByYear":[3,1,2,3,3,5,35,63,49],"rel":1.0531,"reviewed":true,"share":[0.2973,0.07,0.1054,0.1285,0.1123,0.1554,0.8676,1.184,0.6104],"size":243,"slope":0.001024},{"claims":[1,0,1,3,6,14,45,89,79],"deltaPp":0.748,"id":81,"keywords":["reinforcement","reward","agents","learning","planning"],"label":"LLM Reinforcement Learning","labelFull":"Reinforcement Learning for Large Language Model Reasoning","papers":192,"papersByYear":[1,0,1,2,5,10,34,71,68],"rel":7.5479,"reviewed":true,"share":[0.0991,0.0,0.0527,0.0857,0.1872,0.3108,0.8428,1.3343,0.8471],"size":238,"slope":0.001467},{"claims":[1,1,5,6,17,9,31,75,90],"deltaPp":0.5363,"id":82,"keywords":["driving","autonomous","lane","planning","end-to-end"],"label":"End-to-End Autonomous Driving","labelFull":"End-to-End Learning for Autonomous Driving","papers":139,"papersByYear":[1,1,5,5,10,7,16,43,51],"rel":5.4117,"reviewed":true,"share":[0.0991,0.07,0.2636,0.2142,0.3744,0.2175,0.3966,0.8081,0.6354],"size":235,"slope":0.000771},{"claims":[4,5,11,10,23,26,29,43,73],"deltaPp":0.2135,"id":83,"keywords":["quantum","circuits","classical","states","parameterized"],"label":"Quantum Differentiable Programming","labelFull":"Gradient-Based Training of Parameterized Quantum Circuits","papers":110,"papersByYear":[3,3,6,7,9,12,14,15,41],"rel":0.7181,"reviewed":true,"share":[0.2973,0.2101,0.3163,0.2999,0.337,0.3729,0.3471,0.2819,0.5108],"size":224,"slope":0.000201}],"totals":{"claims":[4706,6697,9062,11117,12801,15692,19755,26897,42208],"papers":[1009,1428,1897,2334,2671,3218,4034,5321,8027]},"years":[2018,2019,2020,2021,2022,2023,2024,2025,2026]}