用户建模论文趋势 2023–2026

主要结论

• 用户建模论文从 2023 年的 49 篇增长到 2026 年的 142 篇，三年间扩大到约 2.9 倍。
• NeurIPS 是发用户建模论文最多的会议，2023–2026 年共 121 篇，占四个会议总量的 36%。
• 交叉方向里涨得最快的是对齐与安全：同时属于两个主题的论文从 2023 年的 1 篇增至 2026 年的 26 篇。
• 在「教育与交互」的 5 个主题里，它的 2026 年论文数排第 1；同组增长最快的界面理解是 1.8 倍，用户建模是 2.9 倍。

代表论文

按年份和会议均匀挑选，方便快速看到每一届的样貌

• NeurIPS 2026BehaviorBench: Modeling Real-World User Decisions from Behavioral Traces
• ICML 2026Eating for a Sustainable Planet: Personalized Sustainable Diet Recommendation via Constraint-Aware Decision-Making Modeling
• ICLR 2026A Scalable Inter-edge Correlation Modeling in CopulaGNN for Link Sign Prediction
• CVPR 2026D-Prism Differentiable Primitives for Structured Dynamic Modeling
• NeurIPS 2025Beyond Node-Centric Modeling: Sketching Signed Networks with Simplicial Complexes
• ICML 2025How Transformers Learn Structured Data: Insights From Hierarchical Filtering
• ICLR 2025Logic-Logit: A Logic-Based Approach to Choice Modeling
• CVPR 2025Bridging Gait Recognition and Large Language Models Sequence Modeling
• NeurIPS 20244DBInfer: A 4D Benchmarking Toolbox for Graph-Centric Predictive Modeling on RDBs
• ICML 2024An amortized approach to non-linear mixed-effects modeling based on neural posterior estimation
• ICLR 2024Be Aware of the Neighborhood Effect: Modeling Selection Bias under Interference
• CVPR 2024Steganographic Passport: An Owner and User Verifiable Credential for Deep Model IP Protection Without Retraining
• NeurIPS 2023FairLISA: Fair User Modeling with Limited Sensitive Attributes Information
• ICML 2023dugMatting: Decomposed-Uncertainty-Guided Matting
• ICLR 2023Augmentation Component Analysis: Modeling Similarity via the Augmentation Overlaps
• CVPR 2023BAAM: Monocular 3D Pose and Shape Reconstruction With Bi-Contextual Attention Module and Attention-Guided Modeling
• NeurIPS 2026Causally Structured Differential Network Modeling for Single-Cell Perturbation Prediction
• ICML 2026HyperMLP: An Integrated Perspective for Sequence Modeling
• ICLR 2026Ads that Stick: Near-Optimal Ad Optimization through Psychological Behavior Models
• CVPR 2026Learning Personalized Photographic Style from Pairwise User Preferences

全部用户建模论文

共 142 篇，第 2 / 3 页

51. ICLR 2026Robust Preference Alignment via Directional Neighborhood Consensus
52. ICLR 2026Routing, Cascades, and User Choice for LLMs
53. ICLR 2026Steerable Adversarial Scenario Generation through Test-Time Preference Alignment
54. ICLR 2026Swap-guided Preference Learning for Personalized Reinforcement Learning from Human Feedback
55. ICLR 2026Theoretical Modeling of Large Language Model Self-Improvement Training Dynamics Through Solver-Verifier Gap
56. ICLR 2026Towards Cognitively-Faithful Decision-Making Models to Improve AI Alignment
57. ICLR 2026Towards Persistent Noise-Tolerant Active Learning of Regular Languages with Class Query
58. ICLR 2026What's In My Human Feedback? Learning Interpretable Descriptions of Preference Data
59. ICLR 2026Your Language Model Secretly Contains Personality Subnetworks
60. ICML 2026A Regret Minimization Framework on Preference Learning in Large Language Models
61. ICML 2026Approximating Drift-Diffusion Models for User Decisions under Nudging and External Information
62. ICML 2026Asymmetric Multi-View Clustering with Hyperbolic Uncertainty Modeling
63. ICML 2026Beyond Single Embedding: Modeling User Preferences as Distribution in Federated Recommendation
64. ICML 2026CAMEL: Confidence-Gated Reflection for Reward Modeling
65. ICML 2026Can LLM Agents Stick to the Script? Modeling Commitment in Interactive Narratives
66. ICML 2026Characterizing the Predictive Impact of Modalities with Supervised Latent-Variable Modeling
67. ICML 2026Cross-Subject Modeling for Widefield Calcium Imaging via Atlas-Aligned Spatiotemporal Tokenization
68. ICML 2026Eating for a Sustainable Planet: Personalized Sustainable Diet Recommendation via Constraint-Aware Decision-Making Modeling
69. ICML 2026Federated Variational Preference Alignment with Gumbel-Softmax Prior for Personalized User Preferences
70. ICML 2026HyperMLP: An Integrated Perspective for Sequence Modeling
71. ICML 2026Machine Learning Hamiltonians are Accurate Energy-Force Predictors
72. ICML 2026MASH: Modeling Abstention via Selective Help-Seeking
73. ICML 2026MTNL: A Unified Modeling Perspective for Enhancing Tensor Network Learning
74. ICML 2026On the Sample Efficiency of Inverse Dynamics Models for Semi-Supervised Imitation Learning
75. ICML 2026Personalized Additive Modeling for Multi-level Federated Learning
76. ICML 2026Phase-Type Variational Autoencoders for Heavy-Tailed Data
77. ICML 2026Query Circuits: Explaining How Language Models Answer User Prompts
78. ICML 2026Rethinking Personalization in Large Language Models at the Token Level
79. ICML 2026Stochastic Neural Ray Tracing for Radio Frequency Channel Modeling
80. ICML 2026Strategy-Aware Optimization Modeling with Reasoning LLMs
81. ICML 2026The Sign Estimator: Preference Modeling for LLM Alignment under Heterogeneity
82. ICML 2026Thoughtbubbles: an Unsupervised Method for Parallel Thinking in Latent Space
83. ICML 2026TokSuite: Measuring the Impact of Tokenizer Choice on Language Model Behavior
84. ICML 2026TOM-SWE: User Mental Modeling For Software Engineering Agents
85. ICML 2026Transitivity Meets Cyclicity: Explicit Preference Decomposition for Dynamic Large Language Model Alignment
86. ICML 2026Understanding the Performance Gap in Preference Learning: A Dichotomy of RLHF and DPO
87. ICML 2026Unraveling Syntax: Language Modeling and the Substructure of Grammars
88. ICML 2026User-Aware Active Knowledge Acquisition for Emotional Support Dialogue
89. ICML 2026VenusBench-Mobile: A Challenging and User-Centric Benchmark for Mobile GUI Agents with Capability Diagnostics
90. ICML 2026What Does Preference Learning Recover from Pairwise Comparison Data?
91. ICML 2026When Does Predictive Inverse Dynamics Outperform Behavior Cloning?
92. NeurIPS 2026AMPS: Adaptive Modality Preference Steering via Functional Entropy
93. NeurIPS 2026APM: Evaluating Style Personalization in LLMs with Arbitrary Preference Mappings
94. NeurIPS 2026Atomic Trajectory Modeling with State Space Models for Biomolecular Dynamics
95. NeurIPS 2026BehaviorBench: Modeling Real-World User Decisions from Behavioral Traces
96. NeurIPS 2026Borda-Based Fair Multi-User Dueling Bandit in Tabular and Generalized Linear Settings
97. NeurIPS 2026Bridging Textual Profiles and Latent User Embeddings for Personalization
98. NeurIPS 2026Causally Structured Differential Network Modeling for Single-Cell Perturbation Prediction
99. NeurIPS 2026Class-Domain Incremental Learning with Extensible Multi-Center Modeling
100. NeurIPS 2026Compute Allocation Under Model-Provider Competition

常见问题

2023–2026 年四大会议一共有多少篇用户建模论文？

共 335 篇，按年份分别是 2023 年 49 篇、2024 年 54 篇、2025 年 90 篇、2026 年 142 篇。

NeurIPS 2026 有多少篇用户建模论文？

NeurIPS 2026 有 51 篇用户建模论文，上一届是 42 篇。

ICLR 2026 有多少篇用户建模论文？

ICLR 2026 有 40 篇用户建模论文，上一届是 17 篇。

ICML 2026 有多少篇用户建模论文？

ICML 2026 有 32 篇用户建模论文，上一届是 20 篇。

CVPR 2026 有多少篇用户建模论文？

CVPR 2026 有 19 篇用户建模论文，上一届是 11 篇。

哪个会议的用户建模论文最多？

NeurIPS，2023–2026 年共 121 篇。