Deep physical neural networks trained with backpropagation

🤖 Plain-English Summary

Abstract Deep-learning models have become pervasive tools in science and engineering. Physics-aware training combines the scalability of backpropagation with the automatic mitigation of imperfections and noise achievable with in situ algorithms.

🔑 Key Findings

• However, their energy requirements now increasingly limit their scalability 1 .
• Deep-learning accelerators 2–9 aim to perform deep learning energy-efficiently, usually targeting the inference phase and often by exploiting physical substrates beyond conventional electronics.
• Approaches so far 10–22 have been unable to apply the backpropagation algorithm to train unconventional novel hardware in situ.

🔗 Related Resources