Meta-Learning for On-chip Photonic Neural Network Training

Q: What type of study is this?

This is a Experimental Study study.

Matthew Ho; Zhanghao Sun; Carson Valdez; Olav Solgaard

doi:10.1364/cleo_at.2025.aa128_5

January 1, 2025

Meta-Learning for On-chip Photonic Neural Network Training

View Full Paper

Ask AI

Bookmark

Share

Key Points

Meta-learning reduces optimization steps in training photonic neural networks, enhancing efficiency.
The method achieves more than 4x acceleration in training speed, highlighting significant time savings.
Sparsified meta-learning optimizes less than 10% of hardware elements, increasing resource efficiency.
This optimization strategy may enable broader applications of photonic neural networks in customized tasks.

Implication

This approach shows over 4x optimization speed in photonic neural networks, indicating enhanced efficiency.

Abstract

We introduce photonics neural-network (PNN) meta-learning to reduce on-chip optimization steps for customized tasks. We demonstrate > 4x acceleration and higher accuracy. We further propose sparsified PNN meta-learning that only optimize < 10% of hardware elements.

Discussion

No takes yet. Share an insight, caveat, or question.

Cite This Study

Ho et al. (2025) studied this question.

synapsesocial.com/papers/68d464f131b076d99fa6458c https://doi.org/10.1364/cleo_at.2025.aa128_5

View Full Paper

Ask AI

Bookmark

Share

Meta-Learning for On-chip Photonic Neural Network Training

Key Points

Implication

Abstract

Discussion

Cite This Study

Meta-Learning for On-chip Photonic Neural Network Training

Key Points

Implication

Abstract

Discussion

Cite This Study

Keep reading

Keep reading