arXiv：cs.LG（机器学习，全量分类）· Joonhyeong Park, Byoungwoo Park, Chang-Bae Bang, Jungwon Choi, Hyungjin Chung, Byung-Hoon Kim, Juho Lee·· 15 小时前AI 评分33
随机最优控制框架实现连续时间 fMRI 表征学习
Stochastic Optimal Control for Continuous-Time fMRI Representation Learning
AI 导读
研究者提出将自监督学习重构为随机最优控制（SOC）问题的框架，把大脑活动建模为连续时间潜在动力学，通过学习对时间不规则性不敏感的控制策略来获取鲁棒表征。该框架统一了掩码自编码（MAE）与联合嵌入预测（JEPA），并采用无模拟推理策略以保证大规模 fMRI 数据集上的计算效率与可扩展性。模型在多项下游任务上取得 SOTA 表现，论文已被 ICLR 2026 接收。
正文
Abstract:Learning robust representations from functional magnetic resonance imaging (fMRI) is fundamentally challenged by the temporal irregularity and noise inherent in data from heterogeneous sources. Existing self-supervised learning (SSL) methods often discard critical temporal information by discretizing or averaging fMRI signals. To address this, we introduce a novel framework that reframes SSL as a Stochastic Optimal Control (SOC) problem. Our approach models brain activity as continuous-time latent dynamics, learning a robust representation of brain dynamics by optimizing a control policy that is agnostic to the temporal irregularity. This SOC framework naturally unifies masked autoencoding (MAE) and joint-embedding prediction (JEPA) to extract compact, control-derived representations. Furthermore, a simulation-free inference strategy ensures computational efficiency and scalability for large-scale fMRI datasets. Our model demonstrates state-of-the-art performance across diverse downstream applications, highlighting the potential of the SOC-based continuous-time representation learning framework.
| Comments: | ICLR 2026 |
| Subjects: | Machine Learning (cs.LG); Neurons and Cognition (q-bio.NC); Machine Learning (stat.ML) |
| Cite as: | arXiv:2502.04892 [cs.LG] |
| (or arXiv:2502.04892v2 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2502.04892 arXiv-issued DOI via DataCite |
Submission history
From: Byung-Hoon Kim [view email]
[v1]
Fri, 7 Feb 2025 12:57:26 UTC (20,241 KB)
[v2]
Thu, 1 Oct 2026 03:20:20 UTC (28,229 KB)
来源：arXiv：cs.LG（机器学习，全量分类） · arxiv.org