MLFCIL：一种多级遗忘缓解框架，用于低轨卫星的联邦类增量学习
MLFCIL: A Multi-Level Forgetting Mitigation Framework for Federated Class-Incremental Learning in LEO Satellites
AI总结：
本文提出MLFCIL框架，通过多级方法缓解低轨卫星联邦类增量学习中的遗忘问题，提升模型稳定性和可塑性，实验表明其在准确性和遗忘缓解方面优于基线方法。
摘要：
低地球轨道（LEO）卫星星座正越来越多地执行 onboard 计算。然而，在严格的内存和通信限制下，新类的持续出现对协同训练提出了重大挑战。联邦类增量学习（FCIL）允许分布式增量学习而无需共享原始数据，但面临三个LEO特定挑战：由轨道动力学引起的非独立同分布数据异质性、聚合期间的遗忘加剧，以及在有限资源下平衡稳定性和可塑性。为解决这些挑战，我们提出了MLFCIL，一种多级遗忘缓解框架，将灾难性遗忘分解为三个来源并在不同层次上解决：类加权损失以减少局部偏差，知识蒸馏与特征重放和原型引导的漂移补偿以保留跨任务知识，以及类感知聚合以在联邦过程中缓解遗忘。此外，我们设计了一种双粒度协调策略，结合轮级自适应损失平衡与步级梯度投影，进一步增强稳定性-可塑性权衡。在NWPU-RESISC45数据集上的实验表明，MLFCIL在准确性和遗忘缓解方面显著优于基线方法，同时引入了最小的资源开销。
英文摘要：
Low-Earth-orbit (LEO) satellite constellations are increasingly performing on-board computing. However, the continuous emergence of new classes under strict memory and communication constraints poses major challenges for collaborative training. Federated class-incremental learning (FCIL) enables distributed incremental learning without sharing raw data, but faces three LEO-specific challenges: non-independent and identically distributed data heterogeneity caused by orbital dynamics, amplified catastrophic forgetting during aggregation, and the need to balance stability and plasticity under limited resources. To tackle these challenges, we propose MLFCIL, a multi-level forgetting mitigation framework that decomposes catastrophic forgetting into three sources and addresses them at different levels: class-reweighted loss to reduce local bias, knowledge distillation with feature replay and prototype-guided drift compensation to preserve cross-task knowledge, and class-aware aggregation to mitigate forgetting during federation. In addition, we design a dual-granularity coordination strategy that combines round-level adaptive loss balancing with step-level gradient projection to further enhance the stability-plasticity trade-off. Experiments on the NWPU-RESISC45 dataset show that MLFCIL significantly outperforms baselines in both accuracy and forgetting mitigation, while introducing minimal resource overhead.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
