{"noiseShare":0.5041,"topics":[{"claims":[198,210,305,260,278,337,444,356,592],"deltaPp":-10.0622,"id":0,"keywords":["convergence","gradient","stochastic","optimization","descent"],"label":"Accelerated Stochastic Gradient Methods","labelFull":"Accelerated Stochastic Gradient Methods for Fast Convergence","papers":1558,"papersByYear":[96,98,132,132,147,181,218,201,353],"rel":-0.6509,"reviewed":true,"share":[15.4589,12.6779,12.1771,11.1675,11.9221,9.9015,8.3846,6.0469,5.3967],"size":2980,"slope":-0.011499},{"claims":[58,107,152,134,202,368,379,491,504],"deltaPp":-0.3637,"id":1,"keywords":["graph","node","neural","networks","classification"],"label":"Graph Neural Networks","labelFull":"Graph Neural Networks for Node and Graph Classification","papers":958,"papersByYear":[24,42,56,60,70,130,149,198,229],"rel":-0.0941,"reviewed":true,"share":[3.8647,5.4334,5.1661,5.0761,5.6772,7.1116,5.7308,5.9567,3.501],"size":2395,"slope":0.000547},{"claims":[143,152,189,180,165,227,238,235,348],"deltaPp":-7.1657,"id":2,"keywords":["networks","neural","deep","relu","recurrent"],"label":"Efficient Deep Neural Networks","labelFull":"Efficient Training and Architecture for Deep Neural Networks","papers":1075,"papersByYear":[67,68,98,106,96,130,136,137,237],"rel":-0.6642,"reviewed":true,"share":[10.789,8.7969,9.0406,8.9679,7.7859,7.1116,5.2308,4.1215,3.6233],"size":1877,"slope":-0.008694},{"claims":[10,22,44,53,85,146,266,431,690],"deltaPp":3.3379,"id":3,"keywords":["protein","molecular","design","molecules","generation"],"label":"Generative Biomolecular Design","labelFull":"Generative Modeling for Molecular, Protein, Peptide Design","papers":678,"papersByYear":[5,14,22,22,33,56,101,154,271],"rel":4.1454,"reviewed":true,"share":[0.8052,1.8111,2.0295,1.8613,2.6764,3.0635,3.8846,4.633,4.1431],"size":1747,"slope":0.004455},{"claims":[77,90,97,154,168,213,255,208,417],"deltaPp":-2.7904,"id":4,"keywords":["reinforcement","learning","offline","policy","imitation"],"label":"Reinforcement Learning Benchmarks","labelFull":"Benchmarking and Empirical Evaluation of RL Algorithms","papers":1073,"papersByYear":[46,55,59,101,101,120,154,135,302],"rel":-0.3767,"reviewed":true,"share":[7.4074,7.1151,5.4428,8.5448,8.1914,6.5646,5.9231,4.0614,4.617],"size":1679,"slope":-0.003557},{"claims":[5,11,4,24,27,154,259,360,759],"deltaPp":5.7922,"id":5,"keywords":["diffusion","denoising","sampling","image","discrete"],"label":"Diffusion Model Training & Sampling","labelFull":"Training, Fine-Tuning, and Sampling for Diffusion Models","papers":856,"papersByYear":[4,4,4,12,13,75,123,200,421],"rel":8.9927,"reviewed":true,"share":[0.6441,0.5175,0.369,1.0152,1.0543,4.1028,4.7308,6.0168,6.4363],"size":1603,"slope":0.00858},{"claims":[9,7,21,15,19,80,288,462,668],"deltaPp":5.9278,"id":6,"keywords":["fine-tuning","evaluation","framework"],"label":"LLM Calibration and Benchmarking","labelFull":"Calibration, Benchmarking, and Fine-Tuning of LLMs","papers":1037,"papersByYear":[8,6,14,10,14,44,177,292,472],"rel":4.6016,"reviewed":true,"share":[1.2882,0.7762,1.2915,0.846,1.1354,2.407,6.8077,8.7846,7.216],"size":1569,"slope":0.010055},{"claims":[53,119,133,133,150,171,197,229,354],"deltaPp":-0.6308,"id":7,"keywords":["adversarial","attacks","robustness","backdoor","poisoning"],"label":"Adversarial Machine Learning","labelFull":"Adversarial Machine Learning: Attacks and Defenses","papers":758,"papersByYear":[23,47,64,61,65,74,102,121,201],"rel":-0.1703,"reviewed":true,"share":[3.7037,6.0802,5.9041,5.1607,5.2717,4.0481,3.9231,3.6402,3.0729],"size":1539,"slope":-0.002486},{"claims":[3,7,5,7,11,25,90,212,850],"deltaPp":7.8653,"id":8,"keywords":["reasoning","mathematical","chain-of-thought","benchmarks","accuracy"],"label":"Mathematical Reasoning","labelFull":"Mathematical Reasoning Benchmarks and Chain-of-Thought","papers":734,"papersByYear":[1,2,3,5,6,16,50,126,525],"rel":48.8528,"reviewed":true,"share":[0.161,0.2587,0.2768,0.423,0.4866,0.8753,1.9231,3.7906,8.0263],"size":1210,"slope":0.007634},{"claims":[1,0,5,14,23,38,136,216,758],"deltaPp":6.2906,"id":9,"keywords":["vision-language","visual","multimodal","reasoning","question"],"label":"Vision-Language Model Benchmarks","labelFull":"Evaluation and Benchmarking of Large Vision-Language Models","papers":668,"papersByYear":[1,0,3,11,13,22,73,123,422],"rel":39.072,"reviewed":true,"share":[0.161,0.0,0.2768,0.9306,1.0543,1.2035,2.8077,3.7004,6.4516],"size":1191,"slope":0.006933},{"claims":[19,23,68,45,77,117,203,201,351],"deltaPp":1.0335,"id":10,"keywords":["dynamics","pde","equations","neural","differential"],"label":"Neural Differential Equation Solvers","labelFull":"Neural Differential Equation Solvers for PDEs and ODEs","papers":553,"papersByYear":[12,15,36,26,33,58,89,90,194],"rel":0.5348,"reviewed":true,"share":[1.9324,1.9405,3.321,2.1997,2.6764,3.1729,3.4231,2.7076,2.9659],"size":1104,"slope":0.001269},{"claims":[23,36,68,53,73,141,189,211,295],"deltaPp":-0.3822,"id":11,"keywords":["causal","treatment","effect","discovery","variables"],"label":"Causal Discovery","labelFull":"Causal Discovery Algorithms and Evaluation Benchmarks","papers":488,"papersByYear":[15,19,26,24,37,65,75,94,133],"rel":-0.1582,"reviewed":true,"share":[2.4155,2.458,2.3985,2.0305,3.0008,3.5558,2.8846,2.8279,2.0333],"size":1089,"slope":0.000346},{"claims":[11,12,23,36,23,27,86,178,669],"deltaPp":4.3765,"id":12,"keywords":["agents","llm","systems","ai","memory"],"label":"LLM Agents","labelFull":"LLM Agents: Frameworks, Benchmarks, and Memory Evaluation","papers":595,"papersByYear":[7,7,16,28,20,22,48,87,360],"rel":3.8826,"reviewed":true,"share":[1.1272,0.9056,1.476,2.3689,1.6221,1.2035,1.8462,2.6173,5.5037],"size":1065,"slope":0.003703},{"claims":[18,18,30,39,36,61,133,208,387],"deltaPp":0.8806,"id":13,"keywords":["time series","forecasting","multivariate","autoregressive","temporal"],"label":"Time Series Foundation Models","labelFull":"Foundation Models for Time Series Forecasting","papers":445,"papersByYear":[12,11,18,23,17,33,58,89,184],"rel":0.4557,"reviewed":true,"share":[1.9324,1.423,1.6605,1.9459,1.3788,1.8053,2.2308,2.6775,2.813],"size":930,"slope":0.001381},{"claims":[10,14,14,29,25,42,97,185,500],"deltaPp":3.7181,"id":14,"keywords":["memory","inference","routing","cache","latency"],"label":"LLM Inference Acceleration","labelFull":"Accelerating LLM Inference via Memory, Routing, and Caching","papers":653,"papersByYear":[9,10,11,24,19,34,69,139,338],"rel":2.5654,"reviewed":true,"share":[1.4493,1.2937,1.0148,2.0305,1.541,1.86,2.6538,4.1817,5.1674],"size":916,"slope":0.004441},{"claims":[24,57,78,107,94,128,116,129,174],"deltaPp":-0.641,"id":15,"keywords":["regret","bound","algorithm","sqrt","online"],"label":"Online Regret Minimization","labelFull":"Regret Minimization in Online Learning","papers":522,"papersByYear":[13,33,52,59,59,71,65,75,95],"rel":-0.3062,"reviewed":true,"share":[2.0934,4.2691,4.797,4.9915,4.7851,3.884,2.5,2.2563,1.4524],"size":907,"slope":-0.002384},{"claims":[36,54,86,74,78,128,123,138,171],"deltaPp":-3.4111,"id":16,"keywords":["complexity","algorithm","theoretical","bounds","approximation"],"label":"Provable Guarantees","labelFull":"Provable Guarantees and Complexity Bounds for ML Algorithms","papers":708,"papersByYear":[34,44,61,63,61,101,97,112,135],"rel":-0.623,"reviewed":true,"share":[5.475,5.6921,5.6273,5.3299,4.9473,5.5252,3.7308,3.3694,2.0639],"size":888,"slope":-0.004035},{"claims":[27,47,56,69,75,129,138,119,156],"deltaPp":-0.8317,"id":17,"keywords":["privacy","private","differential","mechanism","guarantees"],"label":"Differential Privacy Mechanisms","labelFull":"Differential Privacy Mechanisms for Machine Learning","papers":355,"papersByYear":[12,15,19,29,32,57,63,56,72],"rel":-0.4304,"reviewed":true,"share":[1.9324,1.9405,1.7528,2.4535,2.5953,3.1182,2.4231,1.6847,1.1007],"size":816,"slope":-0.000348},{"claims":[24,38,23,36,52,54,122,142,314],"deltaPp":0.2701,"id":18,"keywords":["audio","speech","music","emotion","recognition"],"label":"Audio Foundation Models","labelFull":"Audio Foundation Models for Representation and Generation","papers":289,"papersByYear":[10,15,10,13,15,20,36,47,123],"rel":0.1677,"reviewed":true,"share":[1.6103,1.9405,0.9225,1.0998,1.2165,1.0941,1.3846,1.414,1.8804],"size":805,"slope":7e-05},{"claims":[36,41,80,97,98,122,81,101,68],"deltaPp":-2.1871,"id":19,"keywords":["bandits","regret","contextual","multi-armed","linear"],"label":"Contextual Bandits","labelFull":"Contextual Bandits: Regret Bounds and Adaptive Algorithms","papers":368,"papersByYear":[17,22,42,43,49,66,38,55,36],"rel":-0.7989,"reviewed":true,"share":[2.7375,2.8461,3.8745,3.6379,3.974,3.6105,1.4615,1.6546,0.5504],"size":724,"slope":-0.002863},{"claims":[22,34,91,87,79,85,113,105,103],"deltaPp":-1.1995,"id":20,"keywords":["label","semi-supervised","noisy","learning","unlabeled"],"label":"Noisy Label Learning","labelFull":"Learning from Noisy, Partial, and Complementary Labels","papers":398,"papersByYear":[14,21,44,51,39,45,57,58,69],"rel":-0.5321,"reviewed":true,"share":[2.2544,2.7167,4.059,4.3147,3.163,2.4617,2.1923,1.7449,1.0549],"size":719,"slope":-0.002217},{"claims":[0,9,10,40,30,38,84,121,351],"deltaPp":3.1188,"id":21,"keywords":["attention","heads","token","linear","layers"],"label":"Efficient Attention","labelFull":"Efficient and Sparse Attention Mechanisms for Transformers","papers":398,"papersByYear":[0,5,7,16,17,20,56,73,204],"rel":null,"reviewed":true,"share":[0.0,0.6468,0.6458,1.3536,1.3788,1.0941,2.1538,2.1961,3.1188],"size":683,"slope":0.003313},{"claims":[7,11,26,19,15,48,74,139,339],"deltaPp":2.1923,"id":22,"keywords":["3d","point","cloud","geometric","shape"],"label":"Neural 3D Scene Representation","labelFull":"Neural 3D Scene Representation and Reconstruction","papers":337,"papersByYear":[3,5,8,12,7,23,38,66,175],"rel":4.538,"reviewed":true,"share":[0.4831,0.6468,0.738,1.0152,0.5677,1.2582,1.4615,1.9856,2.6754],"size":678,"slope":0.002413},{"claims":[34,48,48,73,72,87,99,69,135],"deltaPp":-1.7305,"id":23,"keywords":["policy","gradient","mdps","markov decision","decision processes"],"label":"Off-Policy Policy Gradient","labelFull":"Off-Policy Policy Gradient Methods for Reinforcement Learning","papers":445,"papersByYear":[21,24,31,51,45,56,58,51,108],"rel":-0.5117,"reviewed":true,"share":[3.3816,3.1048,2.8598,4.3147,3.6496,3.0635,2.2308,1.5343,1.6511],"size":665,"slope":-0.002357},{"claims":[58,76,87,67,57,67,74,64,78],"deltaPp":-4.3886,"id":24,"keywords":["variational","gaussian","inference","process","posterior"],"label":"Variational Gaussian Processes","labelFull":"Variational Inference for Gaussian Process Models","papers":371,"papersByYear":[32,44,52,38,34,38,44,39,50],"rel":-0.8517,"reviewed":true,"share":[5.153,5.6921,4.797,3.2149,2.7575,2.0788,1.6923,1.1733,0.7644],"size":628,"slope":-0.006409},{"claims":[36,31,51,45,41,70,64,79,131],"deltaPp":-2.7956,"id":25,"keywords":["estimator","asymptotic","variance","error","rate"],"label":"Efficient Estimation","labelFull":"Efficient Estimation with Asymptotic Guarantees","papers":390,"papersByYear":[26,23,34,35,33,47,47,54,91],"rel":-0.6677,"reviewed":true,"share":[4.1868,2.9754,3.1365,2.9611,2.6764,2.5711,1.8077,1.6245,1.3912],"size":548,"slope":-0.003047},{"claims":[0,3,5,5,7,6,54,136,312],"deltaPp":3.6386,"id":26,"keywords":["token","decoding","prediction","reduces","inference"],"label":"Dynamic Token Pruning","labelFull":"Dynamic Token Pruning for Efficient Transformer Inference","papers":396,"papersByYear":[0,3,4,5,5,5,42,94,238],"rel":null,"reviewed":true,"share":[0.0,0.3881,0.369,0.423,0.4055,0.2735,1.6154,2.8279,3.6386],"size":528,"slope":0.004036},{"claims":[0,7,13,26,70,80,100,111,110],"deltaPp":0.9784,"id":27,"keywords":["federated","learning","clients","personalized","communication"],"label":"Federated Learning","labelFull":"Federated Learning: Efficient Heterogeneity-Aware Methods","papers":253,"papersByYear":[0,2,5,16,29,43,45,49,64],"rel":null,"reviewed":true,"share":[0.0,0.2587,0.4613,1.3536,2.352,2.3523,1.7308,1.4741,0.9784],"size":517,"slope":0.00185},{"claims":[12,10,14,17,20,50,67,88,211],"deltaPp":1.1976,"id":28,"keywords":["distillation","teacher","student","knowledge","loss"],"label":"Knowledge Distillation","labelFull":"Knowledge Distillation for Efficient Training and Compression","papers":272,"papersByYear":[5,5,8,8,11,23,35,46,131],"rel":1.4873,"reviewed":true,"share":[0.8052,0.6468,0.738,0.6768,0.8921,1.2582,1.3462,1.3839,2.0028],"size":489,"slope":0.001467},{"claims":[10,5,3,12,1,27,63,104,263],"deltaPp":1.1803,"id":29,"keywords":["generation","video","image","editing","quality"],"label":"Text-to-Image & Video Generation","labelFull":"High-Fidelity Text-to-Image and Video Generation","papers":328,"papersByYear":[9,3,3,7,1,22,40,71,172],"rel":0.8144,"reviewed":true,"share":[1.4493,0.3881,0.2768,0.5922,0.0811,1.2035,1.5385,2.136,2.6296],"size":488,"slope":0.002183},{"claims":[49,38,42,51,23,38,56,78,106],"deltaPp":-2.6264,"id":30,"keywords":["matrix","low-rank","sensing","subspace","factorization"],"label":"Low-Rank Matrix Factorization","labelFull":"Low-Rank Matrix Factorization for Efficient Training","papers":299,"papersByYear":[24,24,26,30,19,23,23,49,81],"rel":-0.6796,"reviewed":true,"share":[3.8647,3.1048,2.3985,2.5381,1.541,1.2582,0.8846,1.4741,1.2383],"size":481,"slope":-0.003284},{"claims":[17,25,21,33,29,43,74,81,151],"deltaPp":-0.5972,"id":31,"keywords":["synthetic","real-world","state-of-the-art","outperforms","baselines"],"label":"Empirical Benchmarking","labelFull":"Empirical Benchmarking of Models on Synthetic and Real Data","papers":456,"papersByYear":[17,25,20,31,28,42,73,80,140],"rel":-0.2182,"reviewed":true,"share":[2.7375,3.2342,1.845,2.6227,2.2709,2.2976,2.8077,2.4067,2.1403],"size":474,"slope":-0.000545},{"claims":[10,8,18,25,20,33,48,55,191],"deltaPp":0.3313,"id":32,"keywords":["manipulation","robotic","locomotion","navigation","control"],"label":"Robotic Skill Learning","labelFull":"Skill Learning for Manipulation, Locomotion, and Navigation","papers":288,"papersByYear":[10,8,15,22,15,19,36,36,127],"rel":0.2057,"reviewed":true,"share":[1.6103,1.0349,1.3838,1.8613,1.2165,1.0394,1.3846,1.083,1.9416],"size":408,"slope":0.000108},{"claims":[7,16,23,32,39,54,51,74,107],"deltaPp":0.0734,"id":33,"keywords":["equivariant","group","manifold","networks","invariant"],"label":"Equivariant Neural Networks","labelFull":"Equivariant Neural Networks for Lie Groups and Manifold Data","papers":233,"papersByYear":[6,10,10,16,18,33,28,44,68],"rel":0.076,"reviewed":true,"share":[0.9662,1.2937,0.9225,1.3536,1.4599,1.8053,1.0769,1.3237,1.0396],"size":403,"slope":0.000191},{"claims":[1,4,32,16,28,20,45,86,170],"deltaPp":1.1079,"id":34,"keywords":["quantization","post-training","quantized","weight","error"],"label":"Low-bit Quantization","labelFull":"Low-bit Quantization and Error Modeling for Neural Networks","papers":205,"papersByYear":[1,3,12,7,12,12,27,48,83],"rel":6.8814,"reviewed":true,"share":[0.161,0.3881,1.107,0.5922,0.9732,0.6565,1.0385,1.444,1.2689],"size":402,"slope":0.001254},{"claims":[5,8,17,29,45,40,56,72,101],"deltaPp":0.5871,"id":35,"keywords":["detection","anomaly","out-of-distribution","ood","outlier"],"label":"Out-of-Distribution Detection","labelFull":"Out-of-Distribution Detection and Anomaly Detection in Machine Learning","papers":203,"papersByYear":[3,5,9,18,15,18,29,36,70],"rel":1.2153,"reviewed":true,"share":[0.4831,0.6468,0.8303,1.5228,1.2165,0.9847,1.1154,1.083,1.0702],"size":373,"slope":0.000615},{"claims":[28,32,37,48,33,38,44,61,47],"deltaPp":-1.5808,"id":36,"keywords":["clustering","k-means","approximation","correlation","k-median"],"label":"Approximation Algorithms for Clustering","labelFull":"Approximation Algorithms for Efficient Large-Scale Clustering","papers":175,"papersByYear":[12,16,15,22,15,16,22,34,23],"rel":-0.8181,"reviewed":true,"share":[1.9324,2.0699,1.3838,1.8613,1.2165,0.8753,0.8462,1.0229,0.3516],"size":368,"slope":-0.001921},{"claims":[3,3,9,12,7,29,77,82,141],"deltaPp":1.0763,"id":37,"keywords":["transformer","in-context","learning","linear","trained"],"label":"Transformer Training Efficiency","labelFull":"Training Efficiency and Dynamics of Transformers","papers":239,"papersByYear":[3,3,5,8,7,13,44,54,102],"rel":2.2279,"reviewed":true,"share":[0.4831,0.3881,0.4613,0.6768,0.5677,0.7112,1.6923,1.6245,1.5594],"size":363,"slope":0.001752},{"claims":[0,1,0,0,24,31,39,75,191],"deltaPp":1.804,"id":38,"keywords":["multimodal","modality","fusion","unimodal","missing"],"label":"Missing-Modality Multimodal Learning","labelFull":"Multimodal Representation Learning with Missing Modalities","papers":215,"papersByYear":[0,1,0,0,9,16,23,48,118],"rel":null,"reviewed":true,"share":[0.0,0.1294,0.0,0.0,0.7299,0.8753,0.8846,1.444,1.804],"size":361,"slope":0.002301},{"claims":[1,1,0,2,1,5,27,75,239],"deltaPp":2.3616,"id":39,"keywords":["human","evaluation","judgments","llm","llms"],"label":"LLM-as-Judge Evaluation","labelFull":"LLM-as-Judge Evaluation: Human-Aligned Metrics, Calibration","papers":243,"papersByYear":[1,1,0,2,1,5,16,52,165],"rel":14.6683,"reviewed":true,"share":[0.161,0.1294,0.0,0.1692,0.0811,0.2735,0.6154,1.5644,2.5226],"size":351,"slope":0.002514},{"claims":[12,25,34,15,17,54,30,56,97],"deltaPp":-0.7552,"id":40,"keywords":["generative","modeling","latent","deep","flow"],"label":"Deep Generative Modeling","labelFull":"Deep Generative Modeling: Frameworks and Efficient Training","papers":264,"papersByYear":[12,16,22,13,16,41,23,44,77],"rel":-0.3908,"reviewed":true,"share":[1.9324,2.0699,2.0295,1.0998,1.2976,2.2429,0.8846,1.3237,1.1772],"size":340,"slope":-0.001068},{"claims":[25,32,41,14,50,33,54,41,37],"deltaPp":-2.6774,"id":41,"keywords":["kernel","ridge","regression","lasso","random features"],"label":"Scalable Kernel Ridge Regression","labelFull":"Scalable Kernel Ridge Regression via Random Features","papers":205,"papersByYear":[19,17,21,10,31,23,35,24,25],"rel":-0.8751,"reviewed":true,"share":[3.0596,2.1992,1.9373,0.846,2.5142,1.2582,1.3462,0.722,0.3822],"size":327,"slope":-0.002652},{"claims":[20,21,31,32,23,43,35,43,69],"deltaPp":-1.0446,"id":42,"keywords":["fairness","group","ranking","classifier","constraints"],"label":"Fairness-Constrained Learning","labelFull":"Fairness-Constrained Learning with Multi-Group Guarantees","papers":157,"papersByYear":[10,12,14,14,12,19,20,19,37],"rel":-0.6487,"reviewed":true,"share":[1.6103,1.5524,1.2915,1.1844,0.9732,1.0394,0.7692,0.5716,0.5657],"size":317,"slope":-0.001385},{"claims":[16,26,39,33,46,48,42,28,35],"deltaPp":-1.9263,"id":43,"keywords":["imagenet","cifar-10","cifar-100","mnist","accuracy"],"label":"Image Classification Benchmarks","labelFull":"Performance on Standard Image Classification Benchmarks","papers":268,"papersByYear":[15,20,29,26,39,42,38,27,32],"rel":-0.7975,"reviewed":true,"share":[2.4155,2.5873,2.6753,2.1997,3.163,2.2976,1.4615,0.8123,0.4892],"size":313,"slope":-0.00256},{"claims":[16,20,26,42,34,33,44,56,40],"deltaPp":-1.2668,"id":44,"keywords":["feature","trees","decision","selection","ensemble"],"label":"Tree Ensemble Feature Selection","labelFull":"Tree Ensemble Feature Selection for Robust Classification","papers":217,"papersByYear":[11,15,18,27,24,23,32,34,33],"rel":-0.7152,"reviewed":true,"share":[1.7713,1.9405,1.6605,2.2843,1.9465,1.2582,1.2308,1.0229,0.5045],"size":311,"slope":-0.001618},{"claims":[4,16,18,28,18,34,45,59,63],"deltaPp":0.3272,"id":45,"keywords":["few-shot","zero-shot","classification","generalization","unseen"],"label":"Few-Shot Classification","labelFull":"Few-Shot Classification and Generalization Methods","papers":204,"papersByYear":[3,8,9,16,14,24,34,43,53],"rel":0.6773,"reviewed":true,"share":[0.4831,1.0349,0.8303,1.3536,1.1354,1.3129,1.3077,1.2936,0.8103],"size":285,"slope":0.0005},{"claims":[6,3,3,14,20,38,37,33,125],"deltaPp":0.8775,"id":46,"keywords":["super-resolution","image","restoration","reconstruction","denoising"],"label":"Deep Image Restoration","labelFull":"Deep Learning for Image Restoration and Reconstruction","papers":193,"papersByYear":[3,3,3,12,15,16,25,27,89],"rel":1.8164,"reviewed":true,"share":[0.4831,0.3881,0.2768,1.0152,1.2165,0.8753,0.9615,0.8123,1.3606],"size":279,"slope":0.001002},{"claims":[6,2,5,14,15,21,43,75,98],"deltaPp":0.1121,"id":47,"keywords":["forgetting","unlearning","continual learning","catastrophic","memorization"],"label":"Catastrophic Forgetting Mitigation","labelFull":"Catastrophic Forgetting Mitigation in Continual Learning","papers":172,"papersByYear":[5,2,4,10,10,15,25,41,60],"rel":0.1392,"reviewed":true,"share":[0.8052,0.2587,0.369,0.846,0.811,0.8206,0.9615,1.2335,0.9173],"size":279,"slope":0.000755},{"claims":[0,0,0,1,0,1,46,64,160],"deltaPp":1.3301,"id":48,"keywords":["ai","research","systems","human","technical"],"label":"AI Governance","labelFull":"AI Governance, Safety, and Responsible AI Frameworks","papers":136,"papersByYear":[0,0,0,1,0,1,22,25,87],"rel":null,"reviewed":true,"share":[0.0,0.0,0.0,0.0846,0.0,0.0547,0.8462,0.7521,1.3301],"size":272,"slope":0.00154},{"claims":[20,14,23,22,31,32,41,38,46],"deltaPp":-1.1903,"id":49,"keywords":["multi-task","transfer","learning","merging","task-specific"],"label":"Parameter-Efficient Multi-Task Learning","labelFull":"Parameter-Efficient Multi-Task Learning for Transfer","papers":194,"papersByYear":[11,8,14,16,22,25,31,29,38],"rel":-0.672,"reviewed":true,"share":[1.7713,1.0349,1.2915,1.3536,1.7843,1.3676,1.1923,0.8724,0.581],"size":267,"slope":-0.000906},{"claims":[2,0,3,3,12,8,24,49,142],"deltaPp":0.5952,"id":50,"keywords":["brain","eeg","sleep","fmri","recordings"],"label":"Neuroimaging Foundation Models","labelFull":"Self-supervised Foundation Models for Brain Signals","papers":111,"papersByYear":[2,0,2,2,6,5,13,21,60],"rel":1.8479,"reviewed":true,"share":[0.3221,0.0,0.1845,0.1692,0.4866,0.2735,0.5,0.6318,0.9173],"size":243,"slope":0.000835},{"claims":[3,4,9,9,10,12,38,40,118],"deltaPp":0.8928,"id":51,"keywords":["question","answering","summarization","baselines","benchmarks"],"label":"LLM Question Answering","labelFull":"LLM Question Answering: Prompting, Evaluation, Benchmarks","papers":188,"papersByYear":[3,2,5,8,6,7,33,34,90],"rel":1.8481,"reviewed":true,"share":[0.4831,0.2587,0.4613,0.6768,0.4866,0.3829,1.2692,1.0229,1.3759],"size":243,"slope":0.001198},{"claims":[1,0,2,0,7,21,39,49,120],"deltaPp":1.3372,"id":52,"keywords":["prompt","knowledge","tuning","domain","optimization"],"label":"Prompt Optimization","labelFull":"Automatic Prompt Optimization for LLMs","papers":186,"papersByYear":[1,0,2,0,4,13,30,38,98],"rel":8.3056,"reviewed":true,"share":[0.161,0.0,0.1845,0.0,0.3244,0.7112,1.1538,1.1432,1.4982],"size":239,"slope":0.001905},{"claims":[0,0,1,10,11,27,28,78,78],"deltaPp":0.6727,"id":53,"keywords":["conformal","prediction","coverage","sets","quantile"],"label":"Conformal Prediction","labelFull":"Conformal Prediction: Coverage Guarantees and Efficient Sets","papers":121,"papersByYear":[0,0,1,3,5,14,18,36,44],"rel":null,"reviewed":true,"share":[0.0,0.0,0.0923,0.2538,0.4055,0.7659,0.6923,1.083,0.6727],"size":233,"slope":0.001275},{"claims":[14,24,23,26,19,21,22,29,55],"deltaPp":-1.1598,"id":54,"keywords":["posterior","bayesian","dirichlet","distributions","inference"],"label":"Bayesian Posterior Approximation","labelFull":"Bayesian Posterior Approximation for Deep Neural Networks","papers":163,"papersByYear":[11,14,18,17,12,15,13,23,40],"rel":-0.6548,"reviewed":true,"share":[1.7713,1.8111,1.6605,1.4382,0.9732,0.8206,0.5,0.6919,0.6115],"size":233,"slope":-0.001823},{"claims":[6,8,25,27,32,28,25,42,39],"deltaPp":-0.53,"id":55,"keywords":["games","nash","equilibrium","zero-sum","equilibria"],"label":"Equilibrium Computation in Games","labelFull":"Computing Nash and Correlated Equilibria in Games","papers":124,"papersByYear":[5,5,12,16,18,19,12,19,18],"rel":-0.6582,"reviewed":true,"share":[0.8052,0.6468,1.107,1.3536,1.4599,1.0394,0.4615,0.5716,0.2752],"size":232,"slope":-0.000658},{"claims":[6,7,3,3,4,17,34,62,94],"deltaPp":0.4719,"id":56,"keywords":["reward","preference","feedback","reinforcement learning","human"],"label":"Reward Modeling","labelFull":"Reward Modeling for Reinforcement Learning from Human Feedback","papers":159,"papersByYear":[4,5,3,3,4,10,19,38,73],"rel":0.7327,"reviewed":true,"share":[0.6441,0.6468,0.2768,0.2538,0.3244,0.547,0.7308,1.1432,1.116],"size":230,"slope":0.000763},{"claims":[0,0,4,1,3,7,26,86,101],"deltaPp":1.0396,"id":57,"keywords":["preference","direct","optimization","alignment","human"],"label":"Direct Preference Optimization","labelFull":"Direct Preference Optimization for Human Preference Alignment","papers":152,"papersByYear":[0,0,2,1,3,3,17,58,68],"rel":null,"reviewed":true,"share":[0.0,0.0,0.1845,0.0846,0.2433,0.1641,0.6538,1.7449,1.0396],"size":228,"slope":0.001735},{"claims":[6,2,3,10,5,28,37,63,73],"deltaPp":0.266,"id":58,"keywords":["segmentation","semantic","medical","object","image"],"label":"Image Segmentation","labelFull":"Image Segmentation Methods and Applications","papers":137,"papersByYear":[3,1,1,5,5,15,22,36,49],"rel":0.5506,"reviewed":true,"share":[0.4831,0.1294,0.0923,0.423,0.4055,0.8206,0.8462,1.083,0.7491],"size":227,"slope":0.000972},{"claims":[4,15,23,31,23,34,27,41,29],"deltaPp":-0.2925,"id":59,"keywords":["loss","functions","bounds","generalization","surrogate"],"label":"Calibrated Surrogate Losses","labelFull":"Theory and Bounds for Calibrated Surrogate Losses","papers":167,"papersByYear":[4,13,18,18,17,22,20,32,23],"rel":-0.4541,"reviewed":true,"share":[0.6441,1.6818,1.6605,1.5228,1.3788,1.2035,0.7692,0.9627,0.3516],"size":227,"slope":-0.000905},{"claims":[3,0,1,8,9,11,46,36,110],"deltaPp":0.527,"id":60,"keywords":["spiking","neural networks","temporal","neuron","energy"],"label":"Spiking Neural Network Training","labelFull":"Training and Efficient Inference for Spiking Neural Networks","papers":92,"papersByYear":[1,0,1,3,3,5,15,19,45],"rel":3.2733,"reviewed":true,"share":[0.161,0.0,0.0923,0.2538,0.2433,0.2735,0.5769,0.5716,0.688],"size":224,"slope":0.000802},{"claims":[4,9,19,17,22,28,23,28,69],"deltaPp":0.1437,"id":61,"keywords":["transport","optimal","unbalanced","maps","entropic"],"label":"Optimal Transport","labelFull":"Optimal Transport Algorithms, Regularization, and Extensions","papers":116,"papersByYear":[3,3,8,9,8,15,15,14,41],"rel":0.2975,"reviewed":true,"share":[0.4831,0.3881,0.738,0.7614,0.6488,0.8206,0.5769,0.4212,0.6268],"size":219,"slope":6.9e-05},{"claims":[13,15,27,16,8,23,27,24,65],"deltaPp":-0.6084,"id":62,"keywords":["latent","space","variable","representations","distribution"],"label":"Latent Variable Modeling","labelFull":"Latent Variable Modeling and Disentangled Representations","papers":178,"papersByYear":[9,14,18,14,8,20,21,19,55],"rel":-0.4198,"reviewed":true,"share":[1.4493,1.8111,1.6605,1.1844,0.6488,1.0941,0.8077,0.5716,0.8409],"size":218,"slope":-0.001325},{"claims":[0,4,3,16,2,32,32,60,67],"deltaPp":0.5198,"id":63,"keywords":["quantum","schrödinger","bridge","classical","states"],"label":"Quantum Machine Learning","labelFull":"Quantum Machine Learning Algorithms and Speedup Analyses","papers":100,"papersByYear":[0,2,2,4,2,16,13,27,34],"rel":null,"reviewed":true,"share":[0.0,0.2587,0.1845,0.3384,0.1622,0.8753,0.5,0.8123,0.5198],"size":216,"slope":0.000818},{"claims":[6,11,19,34,13,21,41,23,45],"deltaPp":-0.2395,"id":64,"keywords":["risk","minimization","distributionally robust","excess","robust optimization"],"label":"Robust Empirical Risk Minimization","labelFull":"Robust Empirical Risk Minimization: Excess Risk Guarantees","papers":152,"papersByYear":[5,11,14,21,11,15,20,18,37],"rel":-0.2974,"reviewed":true,"share":[0.8052,1.423,1.2915,1.7766,0.8921,0.8206,0.7692,0.5415,0.5657],"size":213,"slope":-0.000934},{"claims":[1,1,4,15,0,16,23,35,117],"deltaPp":1.1538,"id":65,"keywords":["code","program","verification","repair","execution"],"label":"Code Generation Evaluation","labelFull":"Execution-based Evaluation & Verification of Code Generation","papers":136,"papersByYear":[1,1,3,6,0,8,13,18,86],"rel":7.1665,"reviewed":true,"share":[0.161,0.1294,0.2768,0.5076,0.0,0.4376,0.5,0.5415,1.3148],"size":212,"slope":0.001038},{"claims":[3,12,25,22,24,26,23,45,29],"deltaPp":-0.162,"id":66,"keywords":["domain","adaptation","source","generalization","unsupervised"],"label":"Domain Adaptation","labelFull":"Unsupervised Multi-Source Domain Adaptation and Generalization","papers":131,"papersByYear":[3,6,15,11,14,15,17,29,21],"rel":-0.3353,"reviewed":true,"share":[0.4831,0.7762,1.3838,0.9306,1.1354,0.8206,0.6538,0.8724,0.3211],"size":209,"slope":-0.000322},{"claims":[0,0,0,5,3,16,28,55,101],"deltaPp":0.5657,"id":67,"keywords":["watermark","detection","text","attacks","robustness"],"label":"LLM Watermarking","labelFull":"Robust Watermarking and Detection for LLM-Generated Text","papers":72,"papersByYear":[0,0,0,1,1,5,9,19,37],"rel":null,"reviewed":true,"share":[0.0,0.0,0.0,0.0846,0.0811,0.2735,0.3462,0.5716,0.5657],"size":208,"slope":0.00081},{"claims":[0,8,6,15,42,44,24,23,36],"deltaPp":0.3669,"id":68,"keywords":["contrastive","augmentation","learning","supervised","features"],"label":"Contrastive Representation Learning","labelFull":"Theoretical and Augmentation Advances in Contrastive Learning","papers":115,"papersByYear":[0,4,5,7,20,23,17,15,24],"rel":null,"reviewed":true,"share":[0.0,0.5175,0.4613,0.5922,1.6221,1.2582,0.6538,0.4513,0.3669],"size":198,"slope":0.000387},{"claims":[0,1,9,15,8,7,13,56,85],"deltaPp":0.8409,"id":69,"keywords":["flow","matching","normalizing","velocity","conditional"],"label":"Flow Matching","labelFull":"Flow Matching Methods for Training Normalizing Flow Models","papers":105,"papersByYear":[0,1,4,7,5,5,4,24,55],"rel":null,"reviewed":true,"share":[0.0,0.1294,0.369,0.5922,0.4055,0.2735,0.1538,0.722,0.8409],"size":194,"slope":0.000732},{"claims":[8,2,12,11,13,19,28,23,62],"deltaPp":-0.2018,"id":70,"keywords":["uncertainty","calibration","quantification","epistemic","estimation"],"label":"Uncertainty Calibration","labelFull":"Uncertainty Calibration and Quantification for ML Models","papers":143,"papersByYear":[6,2,11,9,12,17,19,17,50],"rel":-0.2089,"reviewed":true,"share":[0.9662,0.2587,1.0148,0.7614,0.9732,0.93,0.7308,0.5114,0.7644],"size":178,"slope":-7.5e-05},{"claims":[38,32,37,26,13,13,8,2,4],"deltaPp":-2.868,"id":71,"keywords":["gan","generative adversarial","discriminator","adversarial networks","wasserstein"],"label":"Stable GAN Training","labelFull":"Theoretical and Practical Methods for Stable GAN Training","papers":89,"papersByYear":[18,16,19,13,6,8,6,1,2],"rel":-0.9894,"reviewed":true,"share":[2.8986,2.0699,1.7528,1.0998,0.4866,0.4376,0.2308,0.0301,0.0306],"size":173,"slope":-0.00355},{"claims":[10,3,23,14,24,15,28,18,32],"deltaPp":-0.3536,"id":72,"keywords":["tensor","decomposition","factorization","completion","network"],"label":"Low-Rank Tensor Learning","labelFull":"Low-Rank Tensor Factorization, Completion, and Learning","papers":85,"papersByYear":[4,2,10,8,11,10,13,8,19],"rel":-0.549,"reviewed":true,"share":[0.6441,0.2587,0.9225,0.6768,0.8921,0.547,0.5,0.2407,0.2905],"size":167,"slope":-0.000407},{"claims":[0,3,1,4,11,9,15,24,99],"deltaPp":0.8714,"id":73,"keywords":["retrieval","query","entity","documents","embedding"],"label":"Efficient Dense Retrieval","labelFull":"Efficient Dense Retrieval and Retrieval-Augmented Methods","papers":103,"papersByYear":[0,3,1,3,7,4,10,18,57],"rel":null,"reviewed":true,"share":[0.0,0.3881,0.0923,0.2538,0.5677,0.2188,0.3846,0.5415,0.8714],"size":166,"slope":0.000749},{"claims":[3,5,0,1,7,10,19,42,76],"deltaPp":0.527,"id":74,"keywords":["multi-view","clustering","incomplete","cross-view","missing"],"label":"Incomplete Multi-View Clustering","labelFull":"Incomplete Multi-View Clustering with Missing-View Imputation","papers":90,"papersByYear":[1,2,0,1,5,5,9,22,45],"rel":3.2733,"reviewed":true,"share":[0.161,0.2587,0.0,0.0846,0.4055,0.2735,0.3462,0.6619,0.688],"size":163,"slope":0.0007},{"claims":[6,13,6,15,11,16,24,30,40],"deltaPp":-0.2854,"id":75,"keywords":["compression","codes","perceptual","distortion","decoder"],"label":"Neural Compression","labelFull":"Neural Compression: Learned Data and Model Compression","papers":110,"papersByYear":[5,6,4,6,8,11,16,20,34],"rel":-0.3544,"reviewed":true,"share":[0.8052,0.7762,0.369,0.5076,0.6488,0.6018,0.6154,0.6017,0.5198],"size":161,"slope":-0.00018},{"claims":[1,2,12,9,27,20,26,20,44],"deltaPp":0.2059,"id":76,"keywords":["explanations","rationale","faithfulness","counterfactual","framework"],"label":"Post-hoc Explanations","labelFull":"Robust, Faithful, and Efficient Post-hoc Explanations","papers":93,"papersByYear":[1,1,5,8,12,11,17,14,24],"rel":1.2789,"reviewed":true,"share":[0.161,0.1294,0.4613,0.6768,0.9732,0.6018,0.6538,0.4212,0.3669],"size":161,"slope":0.000335},{"claims":[4,5,10,14,9,23,14,26,47],"deltaPp":-0.2619,"id":77,"keywords":["recommendation","collaborative filtering","user","systems","social"],"label":"Recommender Systems","labelFull":"Collaborative Filtering, Social, Sequential Recommendation","papers":87,"papersByYear":[4,3,7,6,5,13,8,16,25],"rel":-0.4066,"reviewed":true,"share":[0.6441,0.3881,0.6458,0.5076,0.4055,0.7112,0.3077,0.4813,0.3822],"size":152,"slope":-0.000207},{"claims":[1,2,0,4,3,9,32,37,62],"deltaPp":0.5728,"id":78,"keywords":["text","visual","image","clip","captions"],"label":"Fine-Grained Vision-Language Alignment","labelFull":"Fine-Grained Vision-Language Alignment and Evaluation","papers":116,"papersByYear":[1,2,0,2,3,8,22,30,48],"rel":3.5578,"reviewed":true,"share":[0.161,0.2587,0.0,0.1692,0.2433,0.4376,0.8462,0.9025,0.7338],"size":150,"slope":0.001031},{"claims":[1,15,6,15,11,26,14,20,34],"deltaPp":0.3129,"id":79,"keywords":["exploration","planning","exploitation","active","policies"],"label":"Exploration Strategies","labelFull":"Exploration Strategies for Sample-Efficient Reinforcement Learning","papers":112,"papersByYear":[1,10,5,13,8,19,11,14,31],"rel":1.9435,"reviewed":true,"share":[0.161,1.2937,0.4613,1.0998,0.6488,1.0394,0.4231,0.4212,0.4739],"size":142,"slope":-0.00025},{"claims":[2,3,1,5,4,3,15,31,76],"deltaPp":0.5035,"id":80,"keywords":["alignment","sequence","direct","llm","human"],"label":"Direct Preference Optimization","labelFull":"Direct Preference Optimization for LLM Alignment","papers":101,"papersByYear":[2,2,1,4,2,3,10,23,54],"rel":1.5632,"reviewed":true,"share":[0.3221,0.2587,0.0923,0.3384,0.1622,0.1641,0.3846,0.6919,0.8256],"size":140,"slope":0.000621}],"totals":{"claims":[2870,3559,5010,5555,5866,8652,12340,16076,33132],"papers":[621,773,1084,1182,1233,1828,2600,3324,6541]},"years":[2018,2019,2020,2021,2022,2023,2024,2025,2026]}