Scholarlink

<p>Recent advances in AI are pushing the limits of traditional hardware, making physical neural networks (PNNs) a promising alternative. However, training PNNs remains challenging: in&#xa0;silico training suffers from model-reality mismatch, while in&#xa0;situ training produces device-specific models that do not transfer across fabrication variations. Both approaches are further compromised by post-deployment perturbations, such as thermal drift or misalignment, which invalidate trained models and require retraining. We address these challenges through sharpness-aware training (SAT), inspired by sharpness-aware minimization, which links loss landscape geometry to generalization. We establish a connection between loss landscape sharpness and robustness in physical systems and leverage it to improve PNN training. SAT is compatible with both in&#xa0;silico and in&#xa0;situ settings: it mitigates model-reality gaps, enables cross-device transfer, and provides strong resilience to post-deployment perturbations without retraining. We demonstrate SAT across three PNN platforms and multiple tasks, including classification, compression, reconstruction, and generation, showing its broad applicability.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Physical neural networks using sharpness-aware training

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年1月19日

查看全文
  • Tengji Xu,
  • Zeyu Luo,
  • Shaojie Liu,
  • Li Fan,
  • Qiarong Xiao,
  • Benshan Wang,
  • Dongliang Wang,
  • Chaoran Huang

摘要

Recent advances in AI are pushing the limits of traditional hardware, making physical neural networks (PNNs) a promising alternative. However, training PNNs remains challenging: in silico training suffers from model-reality mismatch, while in situ training produces device-specific models that do not transfer across fabrication variations. Both approaches are further compromised by post-deployment perturbations, such as thermal drift or misalignment, which invalidate trained models and require retraining. We address these challenges through sharpness-aware training (SAT), inspired by sharpness-aware minimization, which links loss landscape geometry to generalization. We establish a connection between loss landscape sharpness and robustness in physical systems and leverage it to improve PNN training. SAT is compatible with both in silico and in situ settings: it mitigates model-reality gaps, enables cross-device transfer, and provides strong resilience to post-deployment perturbations without retraining. We demonstrate SAT across three PNN platforms and multiple tasks, including classification, compression, reconstruction, and generation, showing its broad applicability.