深度学习 · foundation models论文趋势 2023–2026

主要结论

• 深度学习 · foundation models论文从 2023 年的 72 篇增长到 2026 年的 177 篇，三年间扩大到约 2.5 倍。
• NeurIPS 是发深度学习 · foundation models论文最多的会议，2023–2026 年共 299 篇，占四个会议总量的 56%。
• 交叉方向里涨得最快的是基础模型适配：同时属于两个主题的论文从 2023 年的 7 篇增至 2026 年的 66 篇。
• 在「深度学习」的 11 个主题里，它的 2026 年论文数排第 8；同组增长最快的扩散模型是 4.1 倍，深度学习 · foundation models是 2.5 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 2026Inferotemporal Cortex Collaborates Before It Codes: Non-Serial Inter-Area Synergy in the Macaque Ventral Stream
• ICML 2026ACO-MoE-LoRA: Evolving-while-Training for Adapting Segment Anything Model 2 to Specialized Domains
• ICLR 2026Qronos: Correcting the Past by Shaping the Future... in Post-Training Quantization
• CVPR 2026Adapting a Pre-trained Single-Cell Foundation Model to Spatial Gene Expression
• NeurIPS 2025An Investigation of Memorization Risk in Healthcare Foundation Models
• ICML 2025Contrastive Localized Language-Image Pre-Training
• ICLR 2025Revisiting Convolution Architecture in the Realm of DNA Foundation Models
• CVPR 2025DEFOM-Stereo: Depth Foundation Model Based Stereo Matching
• NeurIPS 2024$\textit{Read-ME}$: Refactorizing LLMs as Router-Decoupled Mixture of Experts with System Co-Design
• ICML 2024Position: On the Societal Impact of Open Foundation Models
• CVPR 20243D-LFM: Lifting Foundation Model
• NeurIPS 2023Bridging the Domain Gap: Self-Supervised 3D Scene Understanding with Foundation Models
• ICML 2023Improving Hyperparameter Learning under Approximate Inference in Gaussian Process Models
• NeurIPS 2026Do Pathology Foundation Models Encode Disease Progression? A Pseudotime Analysis of Visual Representations
• ICML 2026Alethia: a Foundational Encoder for Voice Deepfakes
• ICLR 2026DND: Boosting Large Language Models with Dynamic Nested Depth
• CVPR 2026UniSER A Foundation Model for Unified Soft Effects Removal
• NeurIPS 2025ChromFound: Towards A Universal Foundation Model for Single-Cell Chromatin Accessibiltiy Data
• ICML 2025Exploring Criteria of Loss Reweighting to Enhance LLM Unlearning
• ICLR 2025Recite, Reconstruct, Recollect: Memorization in LMs as a Multifaceted Phenomenon

全部深度学习 · foundation models论文

共 206 篇，第 2 / 5 页

51. ICML 2026InfoAtlas: A Foundation Model for Zero-Shot Statistical Dependence Estimate
52. ICML 2026InfoDLM: an Information-Adaptive Framework for Discrete Diffusion Language Model Pretraining
53. ICML 2026INT vs. FP: A Comprehensive Study of Fine-Grained Low-bit Quantization Formats
54. ICML 2026Interpretable Discovery of One-parameter Subgroups: A Modular Framework for Elliptical, Hyperbolic, and Parabolic Symmetries
55. ICML 2026Is One Layer Enough? Understanding Inference Dynamics in Tabular Foundation Models
56. ICML 2026IsoCompute Playbook: Optimally Scaling Sampling Compute for LLM RL
57. ICML 2026Just Y-Prediction: Enabling Historical Cumulative Inconsistency in Label Diffusion for Learning with Noisy Label
58. ICML 2026KAST-BAR: Knowledge-Anchored Semantically-Dynamic Topology Brain Autoregressive Modeling for Universal Neural Interpretation
59. ICML 2026KromHC: Manifold-Constrained Hyper-Connections with Kronecker-Product Residual Matrices
60. ICML 2026Learning 3D-Gaussian Simulators from RGB Videos
61. ICML 2026Learning Rewrite-Invariant Reasoning with Targeted Alternation Training
62. ICML 2026LEC: Linear Expectation Constraints for Selection-Conditioned Risk Control in Selective Prediction and Routing Systems
63. ICML 2026LiftQuant: Continuous Bit-Width LLM via Dimensional Lifting and Projection
64. ICML 2026LimiX-2M: Mitigating Low-Rank Collapse and Attention Bottlenecks in Tabular Foundation Models
65. ICML 2026Long Grounded Thoughts: Synthesizing Visual Problems and Reasoning Chains at Scale
66. ICML 2026LUCID: Attention with Preconditioned Representations
67. ICML 2026MemoryLLM: Plug-n-Play Interpretable Feed-Forward Memory for Transformers
68. ICML 2026MentisOculi: Revealing the Limits of Reasoning with Mental Imagery
69. ICML 2026Mitigating Hallucinations in Large Vision-Language Models via Causal Route Gating
70. ICML 2026Mitigating Label Shift in Tabular In-Context Learning via Test-Time Posterior Adjustment
71. ICML 2026Mitigating Mask Prior Drift and Positional Attention Collapse in Large Diffusion Vision-Language Models
72. ICML 2026mmBERT: A Modern Multilingual Encoder with Annealed Language Learning
73. ICML 2026MODEL SOUPS NEED ONLY ONE INGREDIENT
74. ICML 2026Model-Dowser: Data-Free Importance Probing to Mitigate Catastrophic Forgetting in Multimodal Large Language Models
75. ICML 2026Model-Preserving Adaptive Rounding
76. ICML 2026MODUS: Decoder-only Any-to-Any Modeling of Diverse Modalities
77. ICML 2026MORE: A Multilingual Document Parsing Benchmark and Evaluation
78. ICML 2026Multimodal Latent Language Modeling with Next-Token Diffusion
79. ICML 2026NanoFLUX: Distillation-Driven Compression of Large Text-to-Image Generation Models for Mobile Devices
80. ICML 2026No Need to Train Your RDB Foundation Model
81. ICML 2026Olivia: Harmonizing Time Series Foundation Models with Power Spectral Density
82. ICML 2026Omni-fMRI: A Universal Atlas-Free fMRI Foundation Model
83. ICML 2026PluRel: Synthetic Data unlocks Scaling Laws for Relational Foundation Models
84. ICML 2026Preconditioning Neural Tangent Kernel for Adaptive Optimization
85. ICML 2026Preference Goal Tuning: Post-Training as Latent Control for Frozen Policies
86. ICML 2026PretrainZero: Reinforcement Active Learning on Pretraining Data
87. ICML 2026PRIM：Cooperative Dynamic Token Compression for Efficient Large Multimodal Models
88. ICML 2026PRISM: Synergizing Vision Foundation Models via Self-organized Expert Specialization
89. ICML 2026Pull Requests as a Training Signal for Repo-Level Code Editing
90. ICML 2026Quantifying the Uncertainty of Foundation Models with Singular Value Ensembles
91. ICML 2026R1-SyntheticVL: Is Synthetic Data from Generative Models Ready for Multimodal Large Language Model?
92. ICML 2026RAGEN-2: Reasoning Collapse in Agentic RL
93. ICML 2026Ramba: Selective State-Space Models for Relational Deep Learning
94. ICML 2026ReasonEdit: Editing Vision--Language Models using Human Reasoning
95. ICML 2026Regression Language Models for Code
96. ICML 2026Relational In-Context Learning via Synthetic Pre-training with Structural Prior
97. ICML 2026Rethinking KV Cache Eviction via a Unified Information-Theoretic Objective
98. ICML 2026Scaling Laws and Architectural Frontiers in Metagenomic Foundation Models
99. ICML 2026ScDiVa: Masked Discrete Diffusion for Joint Modeling of Single-Cell Identity and Expression
100. ICML 2026ScreenParse: Moving Beyond Sparse Grounding with Complete Screen Parsing Supervision

常见问题

2023–2026 年四大会议一共有多少篇深度学习 · foundation models论文？

共 536 篇，按年份分别是 2023 年 72 篇、2024 年 209 篇、2025 年 78 篇、2026 年 177 篇。

NeurIPS 2026 有多少篇深度学习 · foundation models论文？

NeurIPS 2026 有 27 篇深度学习 · foundation models论文，上一届是 15 篇。

ICLR 2026 有多少篇深度学习 · foundation models论文？

ICLR 2026 有 8 篇深度学习 · foundation models论文，上一届是 6 篇。

ICML 2026 有多少篇深度学习 · foundation models论文？

ICML 2026 有 135 篇深度学习 · foundation models论文，上一届是 53 篇。

CVPR 2026 有多少篇深度学习 · foundation models论文？

CVPR 2026 有 7 篇深度学习 · foundation models论文，上一届是 4 篇。

哪个会议的深度学习 · foundation models论文最多？

NeurIPS，2023–2026 年共 299 篇。