arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2610.18404eess.SPcs.LG
收藏

克服灾难性遗忘:雷达波形分类的持续学习

Overcoming Catastrophic Forgetting: Continual Learning for Radar Waveform Classification

Wolfson School of Mechanical, Electrical and Manufacturing Engineering. Loughborough University, U.K.(拉夫堡大学)

Mohamed Rabie, Chinthana J. Panagamuwa, Konstantinos G. Kyriakopoulos

AI总结:

本研究构建首个雷达波形识别的大规模持续学习基准,评估18种持续学习方法,发现保留旧任务数据的方法表现最优,结合有效机制可显著提升性能,为相关系统设计提供依据。

摘要:

自动雷达波形识别用于对接收信号的波形和调制类型进行分类。随着新波形变体随时间出现,已部署的分类器必须在不从头重新训练的情况下学习这些新波形。当学习新波形时会覆盖先前的知识,从而发生灾难性遗忘。持续学习(Continual Learning, CL)方法可应对这一挑战,但现有雷达领域研究仅在狭窄的基准上评估了少数方法,从业者缺乏应部署哪种方法或哪些机制有效的依据。本研究提出首个用于雷达波形识别的大规模持续学习基准,在由四个公共数据集(涵盖54种波形)生成的顺序十任务问题上,对18种CL方法进行评估,评估场景包括任务增量(Task-Incremental, TIL)和类增量(Class-Incremental, CIL)设置,以及单epoch和多epoch regime。保留旧任务数据的方法在TIL和CIL设置中均表现领先;仅使用普通回放(Plain replay)方法在TIL下可缩小上下实际极限间77%-88%的差距,但在CIL下仅缩小46%-55%的差距。对领先方法的内部机制进行消融研究并开展统计检验,以识别有效机制;每项消融研究均在n=6个训练随机种子上与完整方法配对,结果不确定时增至n=12,且在每种方法的家族内对p值进行Holm-Bonferroni校正。消融结果显示,若干机制可带来较小但具有统计显著性的增益,将这些机制结合后,在TIL和CIL下的领先F1分数分别提升3.4%-5.5%和2.7%-4.9%,且组合模型在所有信噪比(SNR)下均超过所有基准方法。这些结果为雷达波形分类系统的研究人员提供了设计层面的机制依据。

英文摘要:

Automated radar waveform recognition classifies the waveform and modulation type of a received signal. As new waveform variants appear over time, a deployed classifier must learn them without retraining from scratch. Catastrophic Forgetting occurs when learning new waveforms overwrites prior knowledge. Continual Learning (CL) methods tackle this challenge, but prior radar studies only evaluate a few methods on narrow benchmarks. Practitioners lack evidence on which method to deploy or which mechanisms are effective. This work presents the first large-scale benchmark of continual learning for radar waveform recognition. Eighteen CL methods are evaluated on a sequential ten-task problem, drawn from four public datasets spanning 54 waveforms, under task- and class-incremental (TIL, CIL) settings and single- and multi-epoch regimes. Methods that retain data from old tasks lead both TIL and CIL settings. Plain replay alone closes 77-88% of the gap between the lower and upper practical limits under TIL, but only 46-55% under CIL. The internal mechanisms of the leading methods are ablated and statistically tested to identify effective mechanisms. Each ablation is paired against its full method over n=6 training seeds, topped up to n=12 when inconclusive, and p-values are Holm-Bonferroni corrected within each method's family. Ablation results show that several mechanisms confer smaller but statistically significant gains. Combining these mechanisms improves the leading F1 score by 3.4-5.5% and 2.7-4.9% under TIL and CIL, respectively, and the combined models exceed every benchmarked method in both settings for all SNRs. These results give researchers of radar waveform classification systems a mechanism-level basis for design.

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。

↑