FCIL-MSN: A Federated Class-Incremental Learning Method for Multisatellite Networks

• Published in: IEEE transactions on geoscience and remote sensing 2024, Vol.62, p.1-15
• Main Authors: Niu, Ziqing, Cheng, Peirui, Wang, Zhirui, Zhao, Liangjin, Sun, Zheng, Sun, Xian, Guo, Zhi
• Format: Article
• Language: English
• Subjects: Adaptation models; Aggregation; Bias; Classification; Collaboration; Computation; Computational modeling; Data models; Distillation; Federated class-incremental learning (FCIL); Federated learning; in-orbit update; Learning; Model updating; Modules; multisatellite networks (MSNs); Performance degradation; Remote sensing; Satellites; Task analysis
• ISSN: 0196-2892; 1558-0644
• DOI: 10.1109/TGRS.2024.3406817

Multisatellite networks (MSNs) have become the prevalent mode for remote sensing intelligent interpretation, with the onboard models requiring class-incremental updates to accommodate the new categories emerging in evolving data and tasks. Traditional model updating methods, which involve uploading models separately after ground-based updating, are inefficient due to limited uplink bandwidth and cumbersome ground update processes while underutilizing potential computing resources on satellites. To address the aforementioned problems, this article innovatively proposes a collaborative in-orbit incremental update method termed federated class-incremental learning (FCIL)-MSN, which leverages observational information and computing resources from MSNs. First, FCIL-MSN achieves collaborative onboard model updates by introducing FCIL into MSNs. Second, a bias calibration-guided relationship distillation module constructs a pseudo-feature set by collaborative MSNs, which alleviates the model bias caused by class imbalance from a global perspective, thereby enhancing model performance. Finally, a gradient information aggregation module is designed to facilitate the exclusion of unfavorable local updates by measuring the contribution of each terminal, thereby accelerating the convergence while obtaining the global model. We conduct extensive experiments on two datasets for scene classification tasks to verify the effectiveness of our proposed method. The experimental results demonstrate that FCIL-MSN outperforms existing general FCIL methods, improving average classification accuracy by 1.45% and decreasing the performance degradation rate by 6.40%.