联邦学习（Federated Learning）是人工智能领域的一项前沿研究方向，旨在解决多个参与方拥有分布式数据，又不方便集中存储或共享数据的问题。在联邦学习中，每个参与方（如移动设备、边缘服务器、个人设备等）都持有自己的本地数据，而模型的训练在本地进行，仅通过交换模型参数的方式进行信息共享，从而实现全局模型的训练。

其中，联邦学习中的数据异构指的是在同一问题领域内，不同数据之间在特征分布、数据结构、性质等方面存在显著差异的情况。换句话说，数据异构问题出现在不同数据源、不同数据集或不同时间点收集的数据之间存在差异，这些差异可能会影响数据的处理、分析和建模过程。可信联邦学习旨在分布式数据环境中进行模型训练，确保数据隐私和安全，同时保持高模型性能，并探讨在联邦学习框架下的个性化和公平性。我们的工作关注于数据异质性、长尾分布、标签缺失和噪声问题，提高模型在异构数据和噪声环境下保证模型的鲁棒性和性能，包括长尾联邦学习、极端噪声标签下的联邦学习和半监督联邦学习等方向。

相关成果：

• 长尾联邦学习

  [1] Fine-Tuning Impairs the Balancedness of Foundation Models in Long-tailed Personalized Federated Learning, Shihao Hou, Chikai Shang, Zhiheng Yang, Jiacheng Yang, Xinyi Shang, Junlong Gao, Yiqun Zhang, and Yang Lu*, IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Colorado, USA, June 3-7, 2026. (CCF-A)

  [2] Asynchronous Federated Clustering with Unknown Number of Clusters, Yunfan Zhang, Yiqun Zhang, Yang Lu, Mengke Li, Xi Chen, and Yiu-ming Cheung, AAAI Conference on Artificial Intelligence (AAAI), Pennsylvania, USA, February 25 - March 4, 2025. (CCF-A)

  [3] MOOD: Leveraging Out-of-Distribution Data to Enhance Imbalanced Semi-Supervised Learning, Yang Lu, Xiaolin Huang, Yizhou Chen, Mengke Li, Yan Yan, Chen Gong, and Hanzi Wang, IEEE Transactions on Neural Networks and Learning Systems (TNNLS), vol. 31, no. 9, pp. 3525-3539, 2020. (JCR 1区 / CCF-B)

  [4] CLIP-guided Federated Learning on Heterogeneous and Long-Tailed Data, Jiangming Shi, Shanshan Zheng, Xiangbo Yin, Yang Lu, Yuan Xie, and Yanyun Qu, AAAI Conference on Artificial Intelligence (AAAI), Vancouver, Canada, February 20–27, 2024. (CCF-A)

  [5] Personalized Federated Learning on Long-Tailed Data via Adversarial Feature Augmentation, Yang Lu, Pinxin Qian, Gang Huang, and Hanzi Wang, IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), Greece, June 4-10, 2023. (CCF-B)

  [6] Federated Learning on Heterogeneous and Long-Tailed Data via Classifier Re-Training with Federated Features, Xinyi Shang, Yang Lu*, Gang Huang, and Hanzi Wang, International Joint Conference on Artificial Intelligence (IJCAI), pp.2218-2224, Vienna, Austria, July 23-29, 2022. (CCF-A)

  [7] FEDIC: Federated Learning on Non-IID and Long-Tailed Data via Calibrated Distillation, Xinyi Shang, Yang Lu*, Yiu-ming Cheung, and Hanzi Wang, IEEE International Conference on Multimedia and Expo (ICME), pp.1-6, Taipei, Taiwan, July 18-22, 2022. (CCF-B)

• 极端噪声标签下的联邦学习:

  [1] Federated Learning with Extremely Noisy Clients via Negative Distillation, Yang Lu, Lin Chen, Yonggang Zhang, Yiliang Zhang, Bo Han, Yiu-ming Cheung, and Hanzi Wang, AAAI Conference on Artificial Intelligence (AAAI), Vancouver, Canada, February 20–27, 2024. (CCF-A)

• 半监督联邦学习:

  [1] Federated Semi-Supervised Learning with Annotation Heterogeneity, Xinyi Shang, Gang Huang, Yang Lu*, Jian Lou, Bo Han, Yiu-ming Cheung, and Hanzi Wang. IEEE Transactions on Artificial Intelligence  (TAI), 2026. (JCR 1区/CCF-A)

  [2] Mind the Gap: Confidence Discrepancy Can Guide Federated Semi-Supervised Learning Across Pseudo-Mismatch, Yijie Liu, Xinyi Shang, Yiqun Zhang, Yang Lu*, Chen Gong, Jing-Hao Xue, and Hanzi Wang, IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), Nashville, USA, June 11-15, 2025. (CCF-A)