Identifying the brain hubs that are crucial for integrating information and distribution is key to understanding how the brain works. In recent years, although various hub identification methods have been proposed in the field of brain imaging, they typically rely on static network representations and analyze using univariate node metrics, thereby neglecting the hub nodes that play a critical role in dynamic global information integration. Additionally, there is an urgent need for an efficient learning method to handle complex brain networks. In this paper, we propose a new reinforcement learning framework, named HubRL, to identify hub nodes that play a central role in coordinating information flow and static topological structures. The agent identifies the most critical brain network nodes by combining simulated information propagation to assess dynamic influence with graph theory metrics to evaluate static topological importance. The experimental results demonstrate that we have successfully identified 37 Task-General hubs in the brain network. Topologically, these hubs exhibit a core advantage over nonhub nodes, with a distribution ratio of approximately$2: 1$in the cerebral cortex gyri and sulci. They also feature significantly longer structural connection fiber bundles and overlap with the regions of the brain with the strongest functional connectivity by up to 80 %. This work frames hub identification as a data-driven sequential decision-making problem without relying on heuristic rules, representing a powerful new paradigm for exploring brain hubs and understanding the working mechanism of the brain.

Human-in-the-loop reinforcement learning integrates human expertise to accelerate agent learning and provide critical guidance and feedback in complex fields. However, many existing approaches focus on single-agent tasks and require continuous human involvement during the training process, significantly increasing the human workload and limiting scalability. In this paper, we propose HARP (HumanAssisted Regrouping with Permutation Invariant Critic), a multi-agent reinforcement learning framework designed for group-oriented tasks. HARP integrates automatic agent regrouping with strategic human assistance during deployment, enabling and allowing non-experts to offer effective guidance with minimal intervention. During training, agents dynamically adjust their groupings to optimize collaborative task completion. When deployed, they actively seek human assistance and utilize the Permutation Invariant Group Critic to evaluate and refine human-proposed groupings, allowing non-expert users to contribute valuable suggestions. In multiple collaboration scenarios, our approach is able to leverage limited guidance from non-experts and enhance performance. The project can be found at https://github.com/huawen-hu/HARP.

Electroencephalogram (EEG) motor imagery decoding, serving as the primary non-invasive modality for exploring brain-computer interfaces, has gained increasing attention. Previous research has achieved significant breakthroughs in the extraction and classification of features related to motor imagery. However, effectively integrating temporal-frequency patterns and capturing long-term dependencies across the entire sequence remain open challenges. To address these issues, we propose a novel Temporal-Frequency Hierarchical Transformer Network (TF-HiTNet) for EEG motor imagery classification. TF-HiTNet leverages a hierarchical transformer architecture and a feature fusion module to effectively extract and integrate temporal and frequency features from EEG signals. This approach captures both local features within EEG segments and global patterns across segments, while simultaneously considering information in both the time and frequency domains. Evaluation on the BCI4-2A and GigaDB datasets demonstrates the effectiveness of TF-HiTNet, achieving an average performance of 79.7% and 83.1%, respectively. Our experiments validate that the hierarchical transformer architecture can effectively learn the relationships between low-level and high-level features, while the time-frequency fusion module significantly improves the accuracy of motor imagery classify-cation.