仿真到真实论文趋势 2023–2026

主要结论

• 仿真到真实论文从 2023 年的 159 篇增长到 2026 年的 204 篇，三年间扩大到约 1.3 倍。
• CVPR 是发仿真到真实论文最多的会议，2023–2026 年共 223 篇，占四个会议总量的 35%。
• 交叉方向里涨得最快的是多模态大模型：同时属于两个主题的论文从 2023 年的 2 篇增至 2026 年的 27 篇。
• 在「机器人与具身」的 5 个主题里，它的 2026 年论文数排第 3；同组增长最快的人形控制是 3.8 倍，仿真到真实是 1.3 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 2026AET-Bench: Pixel-Accurate Atomic Tomography Is Not Atom-Accurate
• ICML 2026RTInfer: Real-Time Inference of Multiple DNNs on Edge GPUs
• ICLR 2026Manipulation as in Simulation: Enabling Accurate Geometry Perception in Robots
• CVPR 2026AnthroTAP Learning Point Tracking with Real-World Motion
• NeurIPS 2025Algorithms and SQ Lower Bounds for Robustly Learning Real-valued Multi-Index Models
• ICML 2025HyperIV: Real-time Implied Volatility Smoothing
• ICLR 2025Do Contemporary Causal Inference Models Capture Real-World Heterogeneity? Findings from a Large-Scale Benchmark
• CVPR 2025Analyzing the Synthetic-to-Real Domain Gap in 3D Hand Pose Estimation
• NeurIPS 2024EgoSim: An Egocentric Multi-view Simulator and Real Dataset for Body-worn Cameras during Motion and Activity
• ICML 2024A Global Geometric Analysis of Maximal Coding Rate Reduction
• ICLR 2024Extending Power of Nature from Binary to Real-Valued Graph Learning in Real World
• CVPR 2024Density-guided Translator Boosts Synthetic-to-Real Unsupervised Domain Adaptive Segmentation of 3D Point Clouds
• NeurIPS 2023[Re] Pure Noise to the Rescue of Insufficient Data
• ICML 2023Adversarially Robust PAC Learnability of Real-Valued Functions
• ICLR 2023Provable Sim-to-real Transfer in Continuous Domain with Partial Observations
• CVPR 2023Cloud-Device Collaborative Adaptation to Continual Changing Environments in the Real-World
• NeurIPS 2026Coarse-to-Real: Generative Rendering for Populated Dynamic Scenes
• ICML 2026Server-Proximal Aggregation for Federated Domain-Incremental Learning under Partial Participation: Task-Uniform Convergence and Backward Transfer
• ICLR 2026PatchRefiner V2: Fast and Lightweight Real-Domain High-Resolution Metric Depth Estimation
• CVPR 2026BiProLoRA Bilevel Prompt LoRA for Real Scene Recovery

全部仿真到真实论文

共 203 篇，第 3 / 5 页

101. ICML 2026Breaking the Synthetic-Real Domain Shortcut for Training-Free Generative Replay-based Class Incremental Learning
102. ICML 2026Closing the Sim-to-Real Gap in Non-Markovian Spreading Processes via GPU-Accelerated Distributional RL
103. ICML 2026Domain Transfer Becomes Identifiable via a Single Alignment
104. ICML 2026Domain-Shift-Aware Conformal Prediction for Large Language Models
105. ICML 2026Evidential Reasoning Advances Interpretable Real-World Disease Screening
106. ICML 2026Focus-Then-Contact: Speeding Up Robotic Contact-Rich Task Learning with Affordance-Guided Real-World Residual Reinforcement Learning
107. ICML 2026From Welfare to Utility: Generalized Objectives in Budget-Feasible Procurement
108. ICML 2026Hyperparameter Transfer with Mixture-of-Expert Layers
109. ICML 2026Incorporating Importance Weighting in Optimal Transport Based Domain Alignment
110. ICML 2026LIMMT: Less Is More for Motion Tracking
111. ICML 2026MER-DG: Modality-Entropy Regularization for Multimodal Domain Generalization
112. ICML 2026MoSA: Motion-constrained Stress Adaptation for Mitigating Real-to-Sim Gap in Continuum Dynamics via Learning Residual Anisotropy
113. ICML 2026Near-Universal Multiplicative Updates for Nonnegative Einsum Factorization
114. ICML 2026Quantifying Cross-Domain Knowledge Distillation in the Presence of Domain Shift
115. ICML 2026Real-Time Monitoring and Calibration of Chain-of-Thought Sycophancy in Large Reasoning Models
116. ICML 2026REAL: Regression-Aware Reinforcement Learning for LLM-as-a-Judge
117. ICML 2026RoboTwin 2.0: A Scalable Data Generator and Benchmark with Strong Domain Randomization for Robust Bimanual Robotic Manipulation
118. ICML 2026RTInfer: Real-Time Inference of Multiple DNNs on Edge GPUs
119. ICML 2026Same Graph Cross-Task Transfer in GNNs: Protocols and Predictors
120. ICML 2026Scaling Unsupervised Multi-Source Federated Domain Adaptation through Group-Wise Discrepancy Minimization
121. ICML 2026Server-Proximal Aggregation for Federated Domain-Incremental Learning under Partial Participation: Task-Uniform Convergence and Backward Transfer
122. ICML 2026Sobolev Regularized Score Difference Estimation in Diffusion Models
123. ICML 2026SoMA: A Real-to-Sim Neural Simulator for Robotic Soft-Body Manipulation
124. ICML 2026SorryDB: Can AI Provers Complete Real-World Lean Theorems?
125. ICML 2026SWE-Perf: Can Language Models Optimize Code Performance on Real-World Repositories?
126. ICML 2026Train Once, Reuse Everywhere: Generalizable Implicit ICL by Routing Attention
127. ICML 2026Transfer Learning in Nonparametric Regression with Deep ReLU Networks
128. ICML 2026Transferable Reinforcement Learning via Probabilistic Latent Embeddings and Dynamic Policy Adaptation for Sim-to-Real Deployment
129. NeurIPS 2026AET-Bench: Pixel-Accurate Atomic Tomography Is Not Atom-Accurate
130. NeurIPS 2026Are Agents Ready to Teach? A Multi-Stage Benchmark for Real-World Teaching Workflows
131. NeurIPS 2026Auditing AI peer reviewers: a dose-response and false-positive benchmark on real scientific papers
132. NeurIPS 2026BehaviorBench: Modeling Real-World User Decisions from Behavioral Traces
133. NeurIPS 2026Beyond Real or Fake: A Dual-Channel Authenticity and Reasoning Protocol for Photographic Assessment
134. NeurIPS 2026Bypassing PC1 Makes SAEs More Reproducible
135. NeurIPS 2026Can Entry-Wise Clipping Give Spectral Control of Stochastic Gradients?
136. NeurIPS 2026Cheap Talk, Real Stakes: Commitment and Exploitation in Human-LLM Strategic Interaction
137. NeurIPS 2026Closed Loop Dynamic Driving Data Mixture for Real-Synthetic Co-Training
138. NeurIPS 2026Coarse-to-Real: Generative Rendering for Populated Dynamic Scenes
139. NeurIPS 2026Continual Learning Bench: Evaluating Frontier AI Systems in Real-World Stateful Environments
140. NeurIPS 2026DeEscalWild: A Real-World Benchmark for Automated De-Escalation Training with SLMs
141. NeurIPS 2026Distributionally Robust Domain Randomization with Learned Risk-Sensitive Dynamics Samplers
142. NeurIPS 2026DoFP-Aligned Lookup Tables for Real-Time Polarization Demosaicking
143. NeurIPS 2026Dynamics-Informed Adaptive Offline RL for Real-Time Tokamak Plasma Control
144. NeurIPS 2026Emergent Misalignment as Data-Mediated Transfer
145. NeurIPS 2026Evaluating Synthetic ECG Pretraining: When Can Patient-Free Simulators Substitute for Real ECG Data?
146. NeurIPS 2026ExploitGym: Can AI Agents Turn Security Vulnerabilities into Real Attacks?
147. NeurIPS 2026ExtrapAir: Air Quality Inference at Unmonitored Locations via Weather-Bridged Spatial Attention
148. NeurIPS 2026FashionChameleon: Towards Real-Time and Interactive Human-Garment Video Customization
149. NeurIPS 2026FASTER: Rethinking Real-Time Flow VLAs
150. NeurIPS 2026FlashPlanner: Real-Time Goal-Conditioned Flow-Matching Planning for Autonomous Driving with Online RL Fine-Tuning

常见问题

2023–2026 年四大会议一共有多少篇仿真到真实论文？

共 629 篇，按年份分别是 2023 年 159 篇、2024 年 148 篇、2025 年 118 篇、2026 年 204 篇。

NeurIPS 2026 有多少篇仿真到真实论文？

NeurIPS 2026 有 75 篇仿真到真实论文，上一届是 33 篇。

ICLR 2026 有多少篇仿真到真实论文？

ICLR 2026 有 33 篇仿真到真实论文，上一届是 16 篇。

ICML 2026 有多少篇仿真到真实论文？

ICML 2026 有 33 篇仿真到真实论文，上一届是 18 篇。

CVPR 2026 有多少篇仿真到真实论文？

CVPR 2026 有 63 篇仿真到真实论文，上一届是 51 篇。

哪个会议的仿真到真实论文最多？

CVPR，2023–2026 年共 223 篇。