PNN

Photonic neural network framework

PNN – Photonic Neural Network Framework

This repository contains a hardware-aware photonic neural network (PNN) framework integrating experimentally measured nonlinear 2D material responses for weight update and activation modeling.

Objectives

• Integrate experimentally extracted phase–voltage characteristics of 2D materials
• Model hardware non-idealities in photonic neural networks
• Enable co-design between photonic devices and AI training
• Develop next-generation energy-efficient photonic AI systems

Structure

• device_models/ – Nonlinear 2D material models (CIPS, α-In₂Se₃, etc.)
• training/ – ONN training pipelines
• notebooks/ – Experimental notebooks and simulations
• experiments/ – Reproducible experiments

Status

Early-stage development – actively evolving.