{"noiseShare":0.5199,"topics":[{"claims":[12,25,38,56,102,181,293,594,663],"deltaPp":2.464,"id":0,"keywords":["protein","molecular","molecules","quantum","prediction"],"label":"Biomolecular Generation","labelFull":"Generative Modeling and Property Prediction for Biomolecules","papers":716,"papersByYear":[7,9,16,26,40,69,108,198,243],"rel":1.1862,"reviewed":true,"share":[2.0772,1.7928,2.329,3.0268,3.6563,4.3865,4.7788,5.347,4.5412],"size":1964,"slope":0.004463},{"claims":[18,58,126,110,185,280,356,397,319],"deltaPp":-0.1103,"id":1,"keywords":["graph","node","neural networks","classification","state-of-the-art"],"label":"Graph Neural Networks","labelFull":"Graph Neural Network Architectures, Training, and Scaling","papers":724,"papersByYear":[9,22,48,41,71,108,132,156,137],"rel":-0.0413,"reviewed":true,"share":[2.6706,4.3825,6.9869,4.773,6.4899,6.8659,5.8407,4.2128,2.5603],"size":1849,"slope":-0.000192},{"claims":[4,26,44,48,68,150,248,460,597],"deltaPp":4.0621,"id":2,"keywords":["3d","pose","rendering","gaussian","depth"],"label":"Neural 3D Reconstruction","labelFull":"Neural 3D Reconstruction With Multi-View Consistency","papers":727,"papersByYear":[3,14,20,23,33,63,110,196,265],"rel":4.5631,"reviewed":true,"share":[0.8902,2.7888,2.9112,2.6775,3.0165,4.0051,4.8673,5.293,4.9523],"size":1645,"slope":0.004833},{"claims":[80,97,148,152,184,162,228,267,325],"deltaPp":-6.2998,"id":3,"keywords":["adversarial","attacks","robustness","backdoor","examples"],"label":"Adversarial Robustness","labelFull":"Adversarial Robustness: Training, Attacks, and Defenses","papers":751,"papersByYear":[32,51,54,61,69,77,99,137,171],"rel":-0.6635,"reviewed":true,"share":[9.4955,10.1594,7.8603,7.1013,6.3071,4.8951,4.3805,3.6997,3.1957],"size":1643,"slope":-0.008957},{"claims":[6,8,21,12,14,64,111,308,1072],"deltaPp":9.6476,"id":4,"keywords":["reasoning","mathematical","math","chain-of-thought","benchmarks"],"label":"Efficient Multi-step Reasoning","labelFull":"Efficient Multi-step Reasoning for Large Language Models","papers":830,"papersByYear":[2,6,14,6,9,33,60,152,548],"rel":16.2554,"reviewed":true,"share":[0.5935,1.1952,2.0378,0.6985,0.8227,2.0979,2.6549,4.1048,10.2411],"size":1616,"slope":0.008325},{"claims":[29,23,29,25,58,56,234,556,582],"deltaPp":2.8396,"id":5,"keywords":["fine-tuning","evaluation","framework"],"label":"LLM Benchmarking & Fine-Tuning","labelFull":"Benchmarking and Fine-Tuning Large Language Models","papers":1071,"papersByYear":[16,16,22,19,36,41,163,352,406],"rel":0.5981,"reviewed":true,"share":[4.7478,3.1873,3.2023,2.2119,3.2907,2.6065,7.2124,9.5058,7.5874],"size":1592,"slope":0.006455},{"claims":[16,27,43,57,55,143,165,329,619],"deltaPp":5.0612,"id":6,"keywords":["benchmarks","state-of-the-art","baselines","outperforms","memory"],"label":"Performance Benchmarking","labelFull":"Empirical Benchmarking of State-of-the-Art Performance","papers":1236,"papersByYear":[15,23,40,46,48,127,151,277,509],"rel":1.1371,"reviewed":true,"share":[4.451,4.5817,5.8224,5.3551,4.3876,8.0737,6.6814,7.4804,9.5122],"size":1454,"slope":0.005563},{"claims":[2,0,1,7,29,77,177,417,492],"deltaPp":4.7865,"id":7,"keywords":["diffusion","denoising","sampling","process","guidance"],"label":"Efficient Diffusion Models","labelFull":"Efficient Training and Sampling for Diffusion Models","papers":638,"papersByYear":[1,0,1,4,14,43,101,202,272],"rel":16.1325,"reviewed":true,"share":[0.2967,0.0,0.1456,0.4657,1.2797,2.7336,4.469,5.455,5.0832],"size":1202,"slope":0.007738},{"claims":[24,29,47,123,164,231,210,161,168],"deltaPp":-2.5426,"id":8,"keywords":["label","domain","contrastive","generalization","learning"],"label":"Learning with Noisy Labels","labelFull":"Learning with Noisy Labels for Domain Shift Robustness","papers":703,"papersByYear":[16,20,28,71,86,125,133,106,118],"rel":-0.5355,"reviewed":true,"share":[4.7478,3.9841,4.0757,8.2654,7.8611,7.9466,5.885,2.8625,2.2052],"size":1157,"slope":-0.001706},{"claims":[28,31,34,28,39,54,71,225,566],"deltaPp":0.8354,"id":9,"keywords":["agents","ai","game","llm","systems"],"label":"Autonomous LLM Agents","labelFull":"Autonomous Large Language Model Agents and Benchmarks","papers":540,"papersByYear":[14,19,21,17,26,30,37,109,267],"rel":0.2011,"reviewed":true,"share":[4.1543,3.7849,3.0568,1.979,2.3766,1.9072,1.6372,2.9436,4.9897],"size":1076,"slope":-0.000349},{"claims":[69,97,120,91,104,154,117,136,128],"deltaPp":-9.6688,"id":10,"keywords":["relu","networks","neural","tangent","activation"],"label":"Neural Tangent Kernel","labelFull":"Neural Tangent Kernel and Infinite-Width Network Theory","papers":598,"papersByYear":[38,53,62,54,58,86,78,83,86],"rel":-0.8575,"reviewed":true,"share":[11.276,10.5578,9.0247,6.2864,5.3016,5.4673,3.4513,2.2414,1.6072],"size":1016,"slope":-0.012598},{"claims":[19,40,25,67,45,70,118,235,333],"deltaPp":-0.1336,"id":11,"keywords":["speech","audio","music","recognition","synthesis"],"label":"Audio Foundation Models","labelFull":"Audio Foundation Models for Representation and Generation","papers":303,"papersByYear":[7,17,9,21,18,24,33,70,104],"rel":-0.0643,"reviewed":true,"share":[2.0772,3.3865,1.31,2.4447,1.6453,1.5257,1.4602,1.8904,1.9436],"size":952,"slope":-0.00094},{"claims":[5,11,15,28,57,64,163,169,316],"deltaPp":1.3524,"id":12,"keywords":["series","time","forecasting","multivariate","temporal"],"label":"Time Series Forecasting","labelFull":"Neural Time Series Forecasting with Large-Scale Models","papers":344,"papersByYear":[3,5,8,14,28,28,60,78,120],"rel":1.5192,"reviewed":true,"share":[0.8902,0.996,1.1645,1.6298,2.5594,1.78,2.6549,2.1064,2.2426],"size":828,"slope":0.001979},{"claims":[29,28,43,52,74,96,137,119,204],"deltaPp":-2.7227,"id":13,"keywords":["reinforcement","learning","offline","algorithms","model-based"],"label":"Offline Reinforcement Learning","labelFull":"Offline Reinforcement Learning Algorithms and Benchmarks","papers":561,"papersByYear":[19,24,34,38,49,65,92,84,156],"rel":-0.4829,"reviewed":true,"share":[5.638,4.7809,4.9491,4.4237,4.479,4.1322,4.0708,2.2684,2.9153],"size":782,"slope":-0.003413},{"claims":[11,3,26,26,43,59,149,196,239],"deltaPp":1.4084,"id":14,"keywords":["causal","counterfactual","treatment","discovery","variables"],"label":"Causal Discovery and Inference","labelFull":"Causal Discovery and Counterfactual Inference","papers":358,"papersByYear":[3,3,12,13,23,34,63,84,123],"rel":1.5821,"reviewed":true,"share":[0.8902,0.5976,1.7467,1.5134,2.1024,2.1615,2.7876,2.2684,2.2986],"size":752,"slope":0.002229},{"claims":[16,23,32,36,47,66,92,163,265],"deltaPp":-0.3488,"id":15,"keywords":["manipulation","robotic","imitation","locomotion","navigation"],"label":"Robotic Skill Learning","labelFull":"Learning for Robotic Manipulation, Locomotion, Navigation","papers":459,"papersByYear":[11,17,21,29,30,39,61,95,156],"rel":-0.1069,"reviewed":true,"share":[3.2641,3.3865,3.0568,3.376,2.7422,2.4793,2.6991,2.5655,2.9153],"size":740,"slope":-0.000912},{"claims":[3,10,26,61,58,93,85,189,207],"deltaPp":1.8151,"id":16,"keywords":["equations","pde","differential","neural","systems"],"label":"Neural PDE Solvers","labelFull":"Neural Solvers for Partial Differential Equations","papers":371,"papersByYear":[1,6,12,25,29,43,51,91,113],"rel":6.1176,"reviewed":true,"share":[0.2967,1.1952,1.7467,2.9104,2.6508,2.7336,2.2566,2.4575,2.1118],"size":732,"slope":0.001982},{"claims":[0,0,0,1,3,32,133,158,196],"deltaPp":2.7285,"id":17,"keywords":["diffusion","text-to-image","video","generation","quality"],"label":"Efficient Diffusion Models","labelFull":"Efficient Diffusion Models for Image, Video, Text Generation","papers":370,"papersByYear":[0,0,0,1,3,22,80,118,146],"rel":null,"reviewed":true,"share":[0.0,0.0,0.0,0.1164,0.2742,1.3986,3.5398,3.1866,2.7285],"size":523,"slope":0.004806},{"claims":[12,12,11,16,25,33,75,133,194],"deltaPp":0.1467,"id":18,"keywords":["attention","heads","linear","softmax","mechanism"],"label":"Efficient Attention","labelFull":"Efficient Attention Mechanisms (Linear, Sparse, and Hybrid)","papers":309,"papersByYear":[7,6,8,11,15,22,43,78,119],"rel":0.0706,"reviewed":true,"share":[2.0772,1.1952,1.1645,1.2806,1.3711,1.3986,1.9027,2.1064,2.2239],"size":511,"slope":0.000819},{"claims":[1,0,2,0,8,29,72,170,217],"deltaPp":2.3944,"id":19,"keywords":["vision-language","visual","zero-shot","image","evaluation"],"label":"Vision-Language Model Evaluation","labelFull":"Evaluation and Benchmarking of Large Vision-Language Models","papers":314,"papersByYear":[1,0,2,0,6,18,45,98,144],"rel":8.0701,"reviewed":true,"share":[0.2967,0.0,0.2911,0.0,0.5484,1.1443,1.9912,2.6465,2.6911],"size":499,"slope":0.003677},{"claims":[1,2,12,9,9,16,43,121,285],"deltaPp":3.553,"id":20,"keywords":["token","importance","decoding","prediction","information"],"label":"Token Pruning","labelFull":"Token Pruning and Efficient Token Selection for LLMs","papers":372,"papersByYear":[1,2,8,7,7,14,34,93,206],"rel":11.9751,"reviewed":true,"share":[0.2967,0.3984,1.1645,0.8149,0.6399,0.89,1.5044,2.5115,3.8497],"size":498,"slope":0.003551},{"claims":[23,16,22,17,11,41,64,97,153],"deltaPp":-0.8556,"id":21,"keywords":["question","answering","summarization","retrieval","benchmarks"],"label":"Open-Domain QA","labelFull":"Open-Domain Retrieval-Augmented Question Answering","papers":306,"papersByYear":[10,9,18,14,9,27,38,68,113],"rel":-0.2883,"reviewed":true,"share":[2.9674,1.7928,2.6201,1.6298,0.8227,1.7165,1.6814,1.8363,2.1118],"size":444,"slope":-0.000847},{"claims":[4,21,8,16,25,53,57,96,159],"deltaPp":1.0534,"id":22,"keywords":["image","restoration","denoising","super-resolution","inverse"],"label":"Fast Diffusion Image Restoration","labelFull":"Fast Diffusion for Image Restoration and Super-Resolution","papers":278,"papersByYear":[3,12,6,9,14,34,32,64,104],"rel":1.1833,"reviewed":true,"share":[0.8902,2.3904,0.8734,1.0477,1.2797,2.1615,1.4159,1.7283,1.9436],"size":439,"slope":0.000738},{"claims":[8,2,16,28,21,47,60,121,129],"deltaPp":0.4203,"id":23,"keywords":["distillation","knowledge","student","teacher","distribution"],"label":"Knowledge Distillation","labelFull":"Knowledge Distillation for Efficient Model Compression","papers":259,"papersByYear":[4,2,7,13,14,28,41,64,86],"rel":0.3541,"reviewed":true,"share":[1.1869,0.3984,1.0189,1.5134,1.2797,1.78,1.8142,1.7283,1.6072],"size":432,"slope":0.001255},{"claims":[7,22,20,35,44,34,85,75,103],"deltaPp":-0.3601,"id":24,"keywords":["posterior","uncertainty","bayesian","estimator","calibration"],"label":"Bayesian Uncertainty Estimation","labelFull":"Bayesian Posterior Estimation and Uncertainty Calibration","papers":309,"papersByYear":[6,17,13,26,28,27,62,54,76],"rel":-0.2023,"reviewed":true,"share":[1.7804,3.3865,1.8923,3.0268,2.5594,1.7165,2.7434,1.4583,1.4203],"size":425,"slope":-0.001139},{"claims":[6,15,27,43,49,63,75,59,75],"deltaPp":0.2685,"id":25,"keywords":["few-shot","zero-shot","classification","learning","clip"],"label":"Few-Shot and Zero-Shot Learning","labelFull":"Few-Shot and Zero-Shot Learning for Classification","papers":284,"papersByYear":[3,7,14,21,29,47,52,49,62],"rel":0.3016,"reviewed":true,"share":[0.8902,1.3944,2.0378,2.4447,2.6508,2.9879,2.3009,1.3233,1.1587],"size":412,"slope":0.000322},{"claims":[4,10,18,24,40,49,40,96,129],"deltaPp":0.7894,"id":26,"keywords":["continual","forgetting","unlearning","catastrophic","learning"],"label":"Continual Learning","labelFull":"Continual Learning and Catastrophic Forgetting Mitigation","papers":232,"papersByYear":[2,7,7,11,20,27,24,60,74],"rel":1.3301,"reviewed":true,"share":[0.5935,1.3944,1.0189,1.2806,1.8282,1.7165,1.0619,1.6203,1.3829],"size":410,"slope":0.000726},{"claims":[6,20,18,26,31,57,64,106,78],"deltaPp":-0.5493,"id":27,"keywords":["gradient","descent","stochastic","convergence","convex"],"label":"Gradient Descent Convergence","labelFull":"Convergence Guarantees for Gradient-Based Optimization","papers":244,"papersByYear":[5,16,15,14,24,31,36,53,50],"rel":-0.3702,"reviewed":true,"share":[1.4837,3.1873,2.1834,1.6298,2.1938,1.9708,1.5929,1.4313,0.9344],"size":406,"slope":-0.001384},{"claims":[2,12,12,31,57,57,60,68,100],"deltaPp":0.752,"id":28,"keywords":["vision","transformers","visual","image","features"],"label":"Vision Transformers","labelFull":"Vision Transformers for Efficient Self-supervised Learning","papers":292,"papersByYear":[2,9,10,24,38,42,46,49,72],"rel":1.2671,"reviewed":true,"share":[0.5935,1.7928,1.4556,2.7939,3.4735,2.6701,2.0354,1.3233,1.3455],"size":399,"slope":0.000439},{"claims":[18,19,32,21,35,44,55,61,105],"deltaPp":-2.3439,"id":29,"keywords":["policy","gradient","optimization","markov decision","decision processes"],"label":"Off-Policy Policy Gradient","labelFull":"Off-Policy Policy Gradient Algorithms for Sample-Efficient RL","papers":281,"papersByYear":[13,14,20,17,25,32,40,39,81],"rel":-0.6076,"reviewed":true,"share":[3.8576,2.7888,2.9112,1.979,2.2852,2.0343,1.7699,1.0532,1.5137],"size":390,"slope":-0.002802},{"claims":[0,0,12,22,48,86,106,48,61],"deltaPp":0.6728,"id":30,"keywords":["federated","learning","client","personalized","heterogeneity"],"label":"Personalized Federated Learning","labelFull":"Personalized Federated Learning for Heterogeneous Client Data","papers":179,"papersByYear":[0,0,5,10,16,41,45,26,36],"rel":null,"reviewed":true,"share":[0.0,0.0,0.7278,1.1641,1.4625,2.6065,1.9912,0.7021,0.6728],"size":383,"slope":0.001461},{"claims":[0,7,7,19,32,31,44,109,121],"deltaPp":1.6819,"id":31,"keywords":["low-rank","matrix","matrices","sparsity","factorization"],"label":"Low-Rank Adaptation","labelFull":"Low-Rank Adaptation for Parameter-Efficient Fine-Tuning","papers":241,"papersByYear":[0,5,7,10,18,18,30,63,90],"rel":null,"reviewed":true,"share":[0.0,0.996,1.0189,1.1641,1.6453,1.1443,1.3274,1.7013,1.6819],"size":370,"slope":0.001573},{"claims":[5,4,8,7,8,14,39,64,200],"deltaPp":1.2448,"id":32,"keywords":["reinforcement","learning","planning","agents","rewards"],"label":"LLM Reinforcement Learning","labelFull":"Reinforcement Learning for Large Language Models","papers":257,"papersByYear":[5,2,5,5,5,9,30,50,146],"rel":0.839,"reviewed":true,"share":[1.4837,0.3984,0.7278,0.5821,0.457,0.5722,1.3274,1.3503,2.7285],"size":349,"slope":0.001504},{"claims":[7,11,18,31,16,45,56,87,77],"deltaPp":-0.7877,"id":33,"keywords":["pretraining","reduces","small","fine-tuning","parameters"],"label":"Efficient Training","labelFull":"Efficient Training of Large-Scale Pretrained Models","papers":301,"papersByYear":[7,11,14,24,14,39,48,75,69],"rel":-0.3792,"reviewed":true,"share":[2.0772,2.1912,2.0378,2.7939,1.2797,2.4793,2.1239,2.0254,1.2895],"size":348,"slope":-0.000632},{"claims":[1,10,0,10,10,22,56,102,132],"deltaPp":1.3479,"id":34,"keywords":["multimodal","modalities","fusion","information","features"],"label":"Multimodal Representation Alignment","labelFull":"Multimodal Representation Alignment and Fusion","papers":208,"papersByYear":[1,3,0,7,3,11,35,60,88],"rel":4.543,"reviewed":true,"share":[0.2967,0.5976,0.0,0.8149,0.2742,0.6993,1.5487,1.6203,1.6446],"size":343,"slope":0.001907},{"claims":[20,22,29,15,13,18,34,67,123],"deltaPp":-0.6404,"id":35,"keywords":["quantization","quantized","weight","post-training","error"],"label":"Post-Training Quantization","labelFull":"Post-Training Quantization for Low-bit Mixed-Precision","papers":160,"papersByYear":[6,8,10,8,7,10,14,36,61],"rel":-0.3597,"reviewed":true,"share":[1.7804,1.5936,1.4556,0.9313,0.6399,0.6357,0.6195,0.9722,1.14],"size":341,"slope":-0.001066},{"claims":[1,9,8,15,22,47,25,102,103],"deltaPp":0.7311,"id":36,"keywords":["combinatorial","optimization","vehicle routing","pareto","solution"],"label":"Neural Combinatorial Optimization","labelFull":"Neural Combinatorial Optimization: Solvers and Benchmarks","papers":168,"papersByYear":[1,3,3,10,13,20,15,48,55],"rel":2.4641,"reviewed":true,"share":[0.2967,0.5976,0.4367,1.1641,1.1883,1.2715,0.6637,1.2962,1.0278],"size":332,"slope":0.00093},{"claims":[13,7,17,13,8,20,57,47,147],"deltaPp":0.4786,"id":37,"keywords":["code","program","tool","test","generation"],"label":"Execution-Guided Code Generation","labelFull":"Execution-Guided Code Generation with Test-Based Validation","papers":206,"papersByYear":[5,5,7,6,6,10,31,31,105],"rel":0.3226,"reviewed":true,"share":[1.4837,0.996,1.0189,0.6985,0.5484,0.6357,1.3717,0.8372,1.9623],"size":329,"slope":0.000347},{"claims":[5,1,9,8,15,15,39,97,137],"deltaPp":1.2731,"id":38,"keywords":["retrieval","query","documents","accuracy","information"],"label":"Retrieval-Augmented Generation","labelFull":"Efficient Dense Retrieval for Retrieval-Augmented Generation","papers":203,"papersByYear":[1,1,8,5,10,11,27,56,84],"rel":4.2909,"reviewed":true,"share":[0.2967,0.1992,1.1645,0.5821,0.9141,0.6993,1.1947,1.5123,1.5698],"size":326,"slope":0.001535},{"claims":[8,4,22,14,18,34,63,66,87],"deltaPp":0.2124,"id":39,"keywords":["detection","out-of-distribution","anomaly","ood","detector"],"label":"Out-of-Distribution Detection","labelFull":"Out-of-Distribution Detection in Deep Neural Networks","papers":178,"papersByYear":[3,2,11,8,12,19,30,34,59],"rel":0.2386,"reviewed":true,"share":[0.8902,0.3984,1.6012,0.9313,1.0969,1.2079,1.3274,0.9182,1.1026],"size":316,"slope":0.000356},{"claims":[2,0,4,5,14,40,75,89,87],"deltaPp":0.563,"id":40,"keywords":["regret","bandits","bound","algorithm","sqrt"],"label":"Bandit Regret Minimization","labelFull":"Regret Bounds for Multi-Armed, Linear, and Federated Bandits","papers":156,"papersByYear":[1,0,2,3,11,22,35,36,46],"rel":1.8975,"reviewed":true,"share":[0.2967,0.0,0.2911,0.3492,1.0055,1.3986,1.5487,0.9722,0.8597],"size":316,"slope":0.001456},{"claims":[1,4,4,15,14,18,38,85,127],"deltaPp":1.3665,"id":41,"keywords":["driving","trajectory","autonomous","guidance","prediction"],"label":"Autonomous Driving Trajectory Prediction","labelFull":"Trajectory Prediction and Generation for Autonomous Driving","papers":195,"papersByYear":[1,2,4,9,10,14,21,45,89],"rel":4.6057,"reviewed":true,"share":[0.2967,0.3984,0.5822,1.0477,0.9141,0.89,0.9292,1.2152,1.6632],"size":306,"slope":0.001409},{"claims":[14,10,8,36,30,23,38,68,74],"deltaPp":-2.1428,"id":42,"keywords":["generative","conditional","modeling","distribution","latent"],"label":"Model-Agnostic Generative Modeling","labelFull":"Model-Agnostic Generative Modeling Frameworks","papers":227,"papersByYear":[11,7,7,26,23,16,29,48,60],"rel":-0.6565,"reviewed":true,"share":[3.2641,1.3944,1.0189,3.0268,2.1024,1.0172,1.2832,1.2962,1.1213],"size":301,"slope":-0.001724},{"claims":[1,6,4,11,11,33,65,81,88],"deltaPp":0.7311,"id":43,"keywords":["transformers","standard","architectures","parameters","comparable"],"label":"Efficient Transformers","labelFull":"Efficient Transformer Architectures and Training Techniques","papers":196,"papersByYear":[1,3,4,7,8,20,37,61,55],"rel":2.4641,"reviewed":true,"share":[0.2967,0.5976,0.5822,0.8149,0.7313,1.2715,1.6372,1.6473,1.0278],"size":300,"slope":0.00144},{"claims":[2,3,5,5,8,16,15,40,206],"deltaPp":1.7051,"id":44,"keywords":["reasoning","visual","spatial","vision-language","perception"],"label":"Vision-Language Spatial Reasoning","labelFull":"Spatial Reasoning in Vision-Language Models","papers":186,"papersByYear":[2,3,4,5,5,9,9,26,123],"rel":2.873,"reviewed":true,"share":[0.5935,0.5976,0.5822,0.5821,0.457,0.5722,0.3982,0.7021,2.2986],"size":300,"slope":0.001126},{"claims":[3,2,8,12,25,28,42,73,94],"deltaPp":0.1914,"id":45,"keywords":["privacy","differential","private","utility","guarantees"],"label":"Differentially Private Machine Learning","labelFull":"Differential Privacy for Machine Learning","papers":136,"papersByYear":[2,2,4,6,10,13,20,37,42],"rel":0.3225,"reviewed":true,"share":[0.5935,0.3984,0.5822,0.6985,0.9141,0.8264,0.885,0.9992,0.7849],"size":287,"slope":0.00055},{"claims":[1,4,10,22,22,50,52,68,55],"deltaPp":0.3948,"id":46,"keywords":["segmentation","semantic","instance","object","detection"],"label":"Image Segmentation","labelFull":"Semantic, Instance, and Promptable Image Segmentation Models","papers":173,"papersByYear":[1,3,7,12,14,34,27,38,37],"rel":1.3306,"reviewed":true,"share":[0.2967,0.5976,1.0189,1.397,1.2797,2.1615,1.1947,1.0262,0.6915],"size":284,"slope":0.000664},{"claims":[0,2,4,4,2,13,48,73,136],"deltaPp":1.9996,"id":47,"keywords":["generation","video","editing","text-to-image","synthesis"],"label":"Text-to-Video Generation","labelFull":"Text-to-Video Generation and Editing with Diffusion Models","papers":216,"papersByYear":[0,1,2,3,2,9,33,59,107],"rel":null,"reviewed":true,"share":[0.0,0.1992,0.2911,0.3492,0.1828,0.5722,1.4602,1.5933,1.9996],"size":282,"slope":0.002457},{"claims":[0,0,0,0,0,2,11,89,176],"deltaPp":2.2239,"id":48,"keywords":["multimodal","understanding","evaluation","visual","benchmark"],"label":"Multimodal LLM Benchmarks","labelFull":"Comprehensive Evaluation of Multimodal Large Language Models","papers":184,"papersByYear":[0,0,0,0,0,2,10,53,119],"rel":null,"reviewed":true,"share":[0.0,0.0,0.0,0.0,0.0,0.1271,0.4425,1.4313,2.2239],"size":278,"slope":0.002367},{"claims":[64,63,36,42,40,17,7,5,1],"deltaPp":-7.9932,"id":49,"keywords":["gan","discriminator","generator","wasserstein","samples"],"label":"Stable GAN Training","labelFull":"Stabilizing GAN Training Via Regularization And Augmentation","papers":130,"papersByYear":[27,26,16,16,24,12,3,5,1],"rel":-0.9977,"reviewed":true,"share":[8.0119,5.1793,2.329,1.8626,2.1938,0.7629,0.1327,0.135,0.0187],"size":275,"slope":-0.008766},{"claims":[0,11,23,17,29,29,50,52,49],"deltaPp":0.5046,"id":50,"keywords":["equivariant","invariant","group","permutation","simplicial"],"label":"Equivariant Graph Neural Networks","labelFull":"Equivariant Graph Neural Networks for Geometric Deep Learning","papers":140,"papersByYear":[0,3,9,10,15,21,28,27,27],"rel":null,"reviewed":true,"share":[0.0,0.5976,1.31,1.1641,1.3711,1.335,1.2389,0.7291,0.5046],"size":260,"slope":0.000407},{"claims":[2,0,1,4,2,5,21,69,156],"deltaPp":1.3874,"id":51,"keywords":["human","llm","evaluation","judges","responses"],"label":"LLM-as-a-Judge Evaluation","labelFull":"Benchmarks and Bias Analysis for LLM-as-a-Judge","papers":175,"papersByYear":[2,0,1,4,2,5,14,41,106],"rel":2.3377,"reviewed":true,"share":[0.5935,0.0,0.1456,0.4657,0.1828,0.3179,0.6195,1.1072,1.9809],"size":260,"slope":0.001612},{"claims":[0,4,5,6,9,9,31,75,115],"deltaPp":1.4016,"id":52,"keywords":["reward","preference","intrinsic","modeling","human"],"label":"Reward Modeling","labelFull":"Reward Modeling for Large Language Model Alignment","papers":162,"papersByYear":[0,4,4,4,9,6,21,39,75],"rel":null,"reviewed":true,"share":[0.0,0.7968,0.5822,0.4657,0.8227,0.3814,0.9292,1.0532,1.4016],"size":254,"slope":0.001164},{"claims":[5,5,13,3,13,17,55,54,88],"deltaPp":0.098,"id":53,"keywords":["spiking","neural networks","neuron","plasticity","synaptic"],"label":"Spiking Neural Network Training","labelFull":"Efficient Training Methods for Spiking Neural Networks","papers":107,"papersByYear":[2,3,4,3,6,7,25,20,37],"rel":0.1651,"reviewed":true,"share":[0.5935,0.5976,0.5822,0.3492,0.5484,0.445,1.1062,0.5401,0.6915],"size":253,"slope":0.000227},{"claims":[0,3,4,5,13,22,31,55,108],"deltaPp":0.9344,"id":54,"keywords":["eeg","brain","recordings","activity","fmri"],"label":"EEG Foundation Models","labelFull":"Foundation Models for EEG and fMRI Neural Recordings","papers":116,"papersByYear":[0,2,2,2,5,10,17,28,50],"rel":null,"reviewed":true,"share":[0.0,0.3984,0.2911,0.2328,0.457,0.6357,0.7522,0.7561,0.9344],"size":241,"slope":0.001023},{"claims":[9,17,25,34,23,35,32,30,21],"deltaPp":-1.3879,"id":55,"keywords":["cifar-10","cifar-100","imagenet","cifar10","mnist"],"label":"Image Classification Benchmarks","labelFull":"Image Classification Benchmarks (CIFAR, ImageNet, etc.)","papers":205,"papersByYear":[6,16,23,30,22,31,28,28,21],"rel":-0.7795,"reviewed":true,"share":[1.7804,3.1873,3.3479,3.4924,2.011,1.9708,1.2389,0.7561,0.3925],"size":226,"slope":-0.003097},{"claims":[0,0,0,1,1,3,19,91,108],"deltaPp":1.2521,"id":56,"keywords":["preference","optimization","direct","user","alignment"],"label":"Direct Preference Optimization","labelFull":"Direct Preference Optimization for LLM Alignment","papers":145,"papersByYear":[0,0,0,1,1,2,13,61,67],"rel":null,"reviewed":true,"share":[0.0,0.0,0.0,0.1164,0.0914,0.1271,0.5752,1.6473,1.2521],"size":223,"slope":0.001852},{"claims":[0,0,0,2,5,34,31,67,74],"deltaPp":1.0652,"id":57,"keywords":["prompts","tuning","incorporates","tokens","knowledge"],"label":"Prompt Optimization","labelFull":"Prompt Optimization and Adaptive Prompt Tuning for LLMs","papers":158,"papersByYear":[0,0,0,1,4,21,26,49,57],"rel":null,"reviewed":true,"share":[0.0,0.0,0.0,0.1164,0.3656,1.335,1.1504,1.3233,1.0652],"size":213,"slope":0.001958},{"claims":[27,29,22,40,9,16,19,13,35],"deltaPp":-4.0212,"id":58,"keywords":["recurrent","neural networks","memory","linear","hidden"],"label":"RNN Theory","labelFull":"Theoretical Analysis of Recurrent Neural Networks","papers":113,"papersByYear":[15,15,10,18,5,9,10,8,23],"rel":-0.9034,"reviewed":true,"share":[4.451,2.988,1.4556,2.0955,0.457,0.5722,0.4425,0.216,0.4298],"size":210,"slope":-0.004658},{"claims":[5,4,14,14,16,48,33,32,40],"deltaPp":-0.8131,"id":59,"keywords":["games","nash","equilibrium","zero-sum","equilibria"],"label":"Nash Equilibrium Learning","labelFull":"Learning Nash Equilibria in Multi-Agent Games","papers":99,"papersByYear":[4,3,5,6,9,20,16,16,20],"rel":-0.6851,"reviewed":true,"share":[1.1869,0.5976,0.7278,0.6985,0.8227,1.2715,0.708,0.4321,0.3738],"size":206,"slope":-0.000536},{"claims":[1,2,1,10,6,9,14,39,120],"deltaPp":0.9928,"id":60,"keywords":["flow","matching","velocity","normalizing","field"],"label":"Flow Matching","labelFull":"Flow Matching for Continuous and Discrete Normalizing Flows","papers":119,"papersByYear":[1,1,1,3,4,5,9,26,69],"rel":3.3461,"reviewed":true,"share":[0.2967,0.1992,0.1456,0.3492,0.3656,0.3179,0.3982,0.7021,1.2895],"size":202,"slope":0.000992},{"claims":[28,23,25,24,27,25,22,17,11],"deltaPp":-3.7268,"id":61,"keywords":["translation","machine","multilingual","non-autoregressive","wmt"],"label":"Neural Machine Translation","labelFull":"Neural Machine Translation: Scaling, Multilingual, Efficient","papers":102,"papersByYear":[13,14,16,13,9,11,12,7,7],"rel":-0.9661,"reviewed":true,"share":[3.8576,2.7888,2.329,1.5134,0.8227,0.6993,0.531,0.189,0.1308],"size":202,"slope":-0.004519},{"claims":[14,32,14,11,19,23,24,37,25],"deltaPp":-1.9628,"id":62,"keywords":["variational","autoencoders","inference","bayesian","latent"],"label":"Variational Autoencoders","labelFull":"Variational Autoencoders: Training, Inference, and Extensions","papers":143,"papersByYear":[8,19,11,11,15,20,15,22,22],"rel":-0.8268,"reviewed":true,"share":[2.3739,3.7849,1.6012,1.2806,1.3711,1.2715,0.6637,0.5941,0.4111],"size":199,"slope":-0.003218},{"claims":[1,15,13,20,4,13,12,35,81],"deltaPp":0.5256,"id":63,"keywords":["dialogue","conversational","multi-turn","systems","generates"],"label":"Multi-Turn Dialogue Modeling","labelFull":"Multi-Turn Dialogue Modeling & Evaluation for LLMs","papers":94,"papersByYear":[1,6,5,5,2,5,7,19,44],"rel":1.7715,"reviewed":true,"share":[0.2967,1.1952,0.7278,0.5821,0.1828,0.3179,0.3097,0.5131,0.8223],"size":194,"slope":-0.000174},{"claims":[0,0,1,1,2,1,43,45,92],"deltaPp":0.542,"id":64,"keywords":["watermark","robustness","attacks","removal","detection"],"label":"Robust AI Watermarking","labelFull":"Robust Watermarking for AI-Generated Text, Images, and Code","papers":59,"papersByYear":[0,0,1,1,1,1,11,15,29],"rel":null,"reviewed":true,"share":[0.0,0.0,0.1456,0.1164,0.0914,0.0636,0.4867,0.4051,0.542],"size":185,"slope":0.000669},{"claims":[2,0,7,0,6,11,38,57,64],"deltaPp":0.3596,"id":65,"keywords":["captions","image-text","alignment","video","clip"],"label":"Vision-Language Alignment","labelFull":"Alignment and Captioning for Text-to-Image Models","papers":143,"papersByYear":[2,0,5,0,4,10,29,42,51],"rel":0.6059,"reviewed":true,"share":[0.5935,0.0,0.7278,0.0,0.3656,0.6357,1.2832,1.1342,0.9531],"size":185,"slope":0.001098},{"claims":[1,8,11,7,10,22,28,50,47],"deltaPp":0.5069,"id":66,"keywords":["convergence","numerical","guarantees","theoretical","rate"],"label":"Convergence Guarantees","labelFull":"Convergence Guarantees for Optimization Algorithms","papers":163,"papersByYear":[1,7,11,7,8,20,24,42,43],"rel":1.7085,"reviewed":true,"share":[0.2967,1.3944,1.6012,0.8149,0.7313,1.2715,1.0619,1.1342,0.8036],"size":184,"slope":0.000104},{"claims":[2,3,12,15,13,23,25,28,45],"deltaPp":0.1705,"id":67,"keywords":["knowledge graph","entity","graph completion","temporal knowledge","reasoning"],"label":"Knowledge Graph Completion","labelFull":"Knowledge Graph Completion and Reasoning","papers":93,"papersByYear":[1,2,9,7,8,15,12,14,25],"rel":0.5747,"reviewed":true,"share":[0.2967,0.3984,1.31,0.8149,0.7313,0.9536,0.531,0.3781,0.4672],"size":166,"slope":-0.000133},{"claims":[1,0,1,1,2,6,9,57,79],"deltaPp":0.6751,"id":68,"keywords":["decoding","speculative","draft","sampling","parallel"],"label":"Speculative Decoding","labelFull":"Speculative Decoding for Accelerated LLM Inference","papers":109,"papersByYear":[1,0,1,1,1,6,7,40,52],"rel":2.2754,"reviewed":true,"share":[0.2967,0.0,0.1456,0.1164,0.0914,0.3814,0.3097,1.0802,0.9718],"size":156,"slope":0.001089},{"claims":[4,0,6,0,10,7,15,55,58],"deltaPp":-0.0305,"id":69,"keywords":["alignment","feedback","compared","preference","target"],"label":"Preference Optimization","labelFull":"Preference Optimization For Large Language Model Alignment","papers":117,"papersByYear":[3,0,5,0,5,4,12,42,46],"rel":-0.0343,"reviewed":true,"share":[0.8902,0.0,0.7278,0.0,0.457,0.2543,0.531,1.1342,0.8597],"size":155,"slope":0.000524},{"claims":[0,0,7,18,9,35,39,11,34],"deltaPp":0.2803,"id":70,"keywords":["fairness","sensitive","group","clustering","attributes"],"label":"Fair Classification","labelFull":"Fair Classification: Group Fairness and Accuracy Trade-offs","papers":75,"papersByYear":[0,0,3,8,6,15,21,7,15],"rel":null,"reviewed":true,"share":[0.0,0.0,0.4367,0.9313,0.5484,0.9536,0.9292,0.189,0.2803],"size":153,"slope":0.000449},{"claims":[19,38,17,19,7,9,8,23,13],"deltaPp":-2.4277,"id":71,"keywords":["word","embeddings","sentence","text classification","topic"],"label":"Training-Free Text Embeddings","labelFull":"Training-Free Text Embeddings with Post-Processing","papers":88,"papersByYear":[9,17,10,11,4,7,5,12,13],"rel":-0.909,"reviewed":true,"share":[2.6706,3.3865,1.4556,1.2806,0.3656,0.445,0.2212,0.3241,0.2429],"size":153,"slope":-0.0037},{"claims":[1,0,3,0,8,5,19,39,75],"deltaPp":0.9928,"id":72,"keywords":["generation","quality","guidance","maintaining","compared"],"label":"Quality-Aware Decoding","labelFull":"Methods for Quality-Aware Decoding and Efficient Generation","papers":138,"papersByYear":[1,0,3,0,7,5,17,36,69],"rel":3.3461,"reviewed":true,"share":[0.2967,0.0,0.4367,0.0,0.6399,0.3179,0.7522,0.9722,1.2895],"size":150,"slope":0.001306},{"claims":[1,1,0,1,1,2,11,29,101],"deltaPp":0.9928,"id":73,"keywords":["vision-language-action","embodied","robot","visual","instructions"],"label":"Vision-Language-Action","labelFull":"Vision-Language-Action Models for Robotic Manipulation","papers":101,"papersByYear":[1,1,0,1,1,2,9,17,69],"rel":3.3461,"reviewed":true,"share":[0.2967,0.1992,0.0,0.1164,0.0914,0.1271,0.3982,0.4591,1.2895],"size":147,"slope":0.000926},{"claims":[0,0,0,0,0,0,11,54,80],"deltaPp":0.7849,"id":74,"keywords":["hallucination","vision-language","reduces","lvlm","mitigates"],"label":"Vision-Language Hallucination Mitigation","labelFull":"Hallucination Detection and Mitigation in Vision-Language Models","papers":76,"papersByYear":[0,0,0,0,0,0,7,27,42],"rel":null,"reviewed":true,"share":[0.0,0.0,0.0,0.0,0.0,0.0,0.3097,0.7291,0.7849],"size":145,"slope":0.000991},{"claims":[6,10,8,16,20,11,13,23,36],"deltaPp":-0.4043,"id":75,"keywords":["multi-task","vectors","task-specific","transfer","learning"],"label":"Multi-Task Meta-Learning","labelFull":"Meta-Learning for Multi-Task Learning and Task Weighting","papers":107,"papersByYear":[3,8,5,10,14,10,13,18,26],"rel":-0.4542,"reviewed":true,"share":[0.8902,1.5936,0.7278,1.1641,1.2797,0.6357,0.5752,0.4861,0.4859],"size":143,"slope":-0.000962},{"claims":[0,0,2,3,1,8,14,40,66],"deltaPp":1.1587,"id":76,"keywords":["inference","speedup","reduces","speed","faster"],"label":"Inference Acceleration","labelFull":"Inference Acceleration and Efficiency Techniques","papers":128,"papersByYear":[0,0,2,3,1,8,14,38,62],"rel":null,"reviewed":true,"share":[0.0,0.0,0.2911,0.3492,0.0914,0.5086,0.6195,1.0262,1.1587],"size":134,"slope":0.001422},{"claims":[3,2,0,1,3,3,16,49,57],"deltaPp":0.397,"id":77,"keywords":["question","answering","visual","pairs","captioning"],"label":"Visual Question Answering","labelFull":"Visual Question Answering and Multimodal Evaluation Benchmarks","papers":114,"papersByYear":[2,2,0,1,3,3,12,38,53],"rel":0.6689,"reviewed":true,"share":[0.5935,0.3984,0.0,0.1164,0.2742,0.1907,0.531,1.0262,0.9905],"size":134,"slope":0.000768},{"claims":[6,5,5,6,25,11,15,24,36],"deltaPp":-0.9417,"id":78,"keywords":["compression","ratios","pruning","error","techniques"],"label":"Neural Model Compression","labelFull":"Efficient Inference via Neural Model Compression","papers":93,"papersByYear":[5,4,5,4,15,6,9,16,29],"rel":-0.6347,"reviewed":true,"share":[1.4837,0.7968,0.7278,0.4657,1.3711,0.3814,0.3982,0.4321,0.542],"size":133,"slope":-0.000934},{"claims":[2,0,6,12,19,22,16,30,25],"deltaPp":-0.145,"id":79,"keywords":["ensemble","prediction","decision tree","selection","sets"],"label":"Ensemble Learning","labelFull":"Ensemble Learning: Selection, Diversity, and Robustness","papers":97,"papersByYear":[2,0,5,8,10,12,12,24,24],"rel":-0.2443,"reviewed":true,"share":[0.5935,0.0,0.7278,0.9313,0.9141,0.7629,0.531,0.6481,0.4485],"size":132,"slope":0.000134},{"claims":[8,6,17,12,22,24,11,10,21],"deltaPp":-1.4253,"id":80,"keywords":["imagenet","top-1","accuracy","resnet-50","flops"],"label":"Efficient ImageNet Classification","labelFull":"Model Compression, Pruning, and Low-Precision for ImageNet","papers":120,"papersByYear":[6,6,16,12,20,21,10,10,19],"rel":-0.8006,"reviewed":true,"share":[1.7804,1.1952,2.329,1.397,1.8282,1.335,0.4425,0.2701,0.3551],"size":131,"slope":-0.002052},{"claims":[8,0,1,0,2,6,23,41,49],"deltaPp":-0.0889,"id":81,"keywords":["motion","human","sequences","generation","video"],"label":"Human Motion Generation","labelFull":"Human Motion Generation and Representation from Text and Video","papers":73,"papersByYear":[2,0,1,0,2,4,11,26,27],"rel":-0.1498,"reviewed":true,"share":[0.5935,0.0,0.1456,0.0,0.1828,0.2543,0.4867,0.7021,0.5046],"size":130,"slope":0.000448},{"claims":[0,0,0,0,9,11,24,34,52],"deltaPp":0.8036,"id":82,"keywords":["tokens","visual","masked image","image modeling","vision"],"label":"Efficient Visual Tokenization","labelFull":"Efficient Visual Tokenization for Vision Transformers","papers":91,"papersByYear":[0,0,0,0,6,8,11,23,43],"rel":null,"reviewed":true,"share":[0.0,0.0,0.0,0.0,0.5484,0.5086,0.4867,0.6211,0.8036],"size":130,"slope":0.001093},{"claims":[0,1,4,3,10,17,8,33,53],"deltaPp":0.6167,"id":83,"keywords":["log","algorithm","complexity","approximation","polylogarithmic"],"label":"Polylogarithmic Complexity Guarantees","labelFull":"Algorithmic Complexity Bounds Up to Polylogarithmic Factors","papers":99,"papersByYear":[0,1,4,3,9,15,8,26,33],"rel":null,"reviewed":true,"share":[0.0,0.1992,0.5822,0.3492,0.8227,0.9536,0.354,0.7021,0.6167],"size":129,"slope":0.000687},{"claims":[8,14,8,7,5,23,7,33,23],"deltaPp":-0.8879,"id":84,"keywords":["adam","sgd","adamw","convergence","mini-batch"],"label":"Adam Optimizer Variants","labelFull":"Adam Optimizer Variants and Convergence Analyses","papers":77,"papersByYear":[4,7,6,5,4,9,6,20,16],"rel":-0.7481,"reviewed":true,"share":[1.1869,1.3944,0.8734,0.5821,0.3656,0.5722,0.2655,0.5401,0.299],"size":128,"slope":-0.001223}],"totals":{"claims":[1639,2416,3272,4113,5323,7651,11105,18692,27860],"papers":[337,502,687,859,1094,1573,2260,3703,5351]},"years":[2018,2019,2020,2021,2022,2023,2024,2025,2026]}