对话系统论文趋势 2023–2026

主要结论

• 对话系统论文从 2023 年的 49 篇增长到 2026 年的 171 篇，三年间扩大到约 3.5 倍。
• NeurIPS 是发对话系统论文最多的会议，2023–2026 年共 189 篇，占四个会议总量的 45%。
• 交叉方向里涨得最快的是多智能体 RL：同时属于两个主题的论文从 2023 年的 0 篇增至 2026 年的 17 篇。
• 在「自然语言」的 5 个主题里，它的 2026 年论文数排第 4；同组增长最快的检索增强生成是 5.4 倍，对话系统是 3.5 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 2026A Quantitative Visual Taxonomy of Worldwide Writing Systems
• ICML 2026An Exponential Separation Between Quantum and Quantum-Inspired Classical Algorithms for Linear Systems
• ICLR 2026A Spectral-Grassmann Wasserstein metric for operator representations of dynamical systems
• CVPR 2026Thermally Activated Dual-Modal Adversarial Clothing against AI Surveillance Systems
• NeurIPS 2025FlowNet: Modeling Dynamic Spatio-Temporal Systems via Flow Propagation
• ICML 2025Discovering Physics Laws of Dynamical Systems via Invariant Function Learning
• ICLR 2025ContextGNN: Beyond Two-Tower Recommendation Systems
• CVPR 2025Efficient Depth Estimation for Unstable Stereo Camera Systems on AR Glasses
• NeurIPS 2024Benchmarking Structural Inference Methods for Interacting Dynamical Systems with Synthetic Data
• ICML 2024Amortized Equation Discovery in Hybrid Dynamical Systems
• ICLR 2024Safe Collaborative Filtering
• NeurIPS 2023Moral Responsibility for AI Systems
• ICML 2023Graph Switching Dynamical Systems
• ICLR 2023An Equal-Size Hard EM Algorithm for Diverse Dialogue Generation
• CVPR 2023expOSE: Accurate Initialization-Free Projective Factorization Using Exponential Regularization
• NeurIPS 2026Adaptive auditing of AI systems with anytime-valid guarantees
• ICML 2026Can Recommender Systems Teach Themselves? A Recursive Self-Improving Framework with Fidelity Control
• ICLR 2026ClarifyVC: Clarifying Ambiguous Commands in Vehicle Control with a Hybrid Data Augmentation Pipeline
• CVPR 2026Task-Driven Implicit Representations for Automated Design of LiDAR Systems
• NeurIPS 2025In-context Learning of Linear Dynamical Systems with Transformers: Approximation Bounds and Depth-separation

全部对话系统论文

共 76 篇，第 2 / 2 页

51. NeurIPS 2026Prompt–Activation Duality: Improving Activation Steering via Attention-Level Interventions
52. NeurIPS 2026Quantifying Concentration Phenomena of Mean-Field Transformers in the Low-Temperature Regime
53. NeurIPS 2026RAG in a Trenchcoat: When Minimal Memory Is Enough for Agentic Systems, and When It Isn’t
54. NeurIPS 2026Recursive Multi-Agent Systems
55. NeurIPS 2026Runtime Monitoring of Perception-Based Autonomous Systems via Embedding Temporal Logic
56. NeurIPS 2026SCDM: Scalable Causal Discovery in Nonlinear Temporal Systems with Meta-Learning
57. NeurIPS 2026Self-EvolveRec: Self-Evolving Recommender Systems with LLM-based Directional Feedback
58. NeurIPS 2026Sparse Koopman Autoencoders Identify Local Dynamical Regimes in Multibasin Systems
59. NeurIPS 2026Spectral Estimation with Deformed Decompression
60. NeurIPS 2026SplineFlow: Flow Matching for Dynamical Systems with B-Spline Interpolants
61. NeurIPS 2026SurvCancel: A Longitudinal Dataset and Benchmark for Dynamic Order Cancellation Prediction in On-Demand Ride-Sharing Systems
62. NeurIPS 2026Symplectic Neural Operators for Learning Infinite Dimensional Hamiltonian Systems
63. NeurIPS 2026Symplectic Reck: In-Situ Learning of Gaussian Quantum Operations
64. NeurIPS 2026Synsema: Syntax-Guided Learning of Semantically Valid Programs
65. NeurIPS 2026The balance between feature learning and collapse in generative dynamical systems
66. NeurIPS 2026The Constitutional Coverage Trilemma in AI Governance
67. NeurIPS 2026The Scaling Laws of Skills in LLM Agent Systems
68. NeurIPS 2026To discretize continually: Mean shift interacting particle systems for Bayesian inference
69. NeurIPS 2026To trust or not to trust: Attention-based Trust Management for LLM Multi-Agent Systems
70. NeurIPS 2026TRACE: Tourism Recommendation with Accountable Citation Evidence
71. NeurIPS 2026Two is better than one: designing heterogeneous scales in binary rating systems
72. NeurIPS 2026Universal Approximation Theorems for Dynamical Systems with Infinite-Time Horizon Guarantees
73. NeurIPS 2026Useful Memories Become Faulty When Continuously Updated by LLMs
74. NeurIPS 2026Validating Causal Abstraction Metrics on Simulated Complex Systems
75. NeurIPS 2026When the Most Disruptive Messages Are Safe to Prune in Aggregate: Functional Analysis of Communication Pruning in Two-Agent LLM Systems
76. NeurIPS 2026Youdunit: Single-Call Counterfactual Necessity in Multi-Agent LLM Systems

常见问题

2023–2026 年四大会议一共有多少篇对话系统论文？

共 419 篇，按年份分别是 2023 年 49 篇、2024 年 92 篇、2025 年 107 篇、2026 年 171 篇。

NeurIPS 2026 有多少篇对话系统论文？

NeurIPS 2026 有 76 篇对话系统论文，上一届是 46 篇。

ICLR 2026 有多少篇对话系统论文？

ICLR 2026 有 36 篇对话系统论文，上一届是 34 篇。

ICML 2026 有多少篇对话系统论文？

ICML 2026 有 54 篇对话系统论文，上一届是 20 篇。

CVPR 2026 有多少篇对话系统论文？

CVPR 2026 有 5 篇对话系统论文，上一届是 7 篇。

哪个会议的对话系统论文最多？

NeurIPS，2023–2026 年共 189 篇。