Training all-mechanical neural networks for task learning through in situ backpropagation

Shuaifeng Li, Xiaoming Mao

Recent advances unveiled physical neural networks as promising machine learning platforms, offering faster and more energy-efficient information processing. Compared with extensive…