An ITO Graphene hybrid integrated absorption modulator on Si-photonics for neuromorphic nonlinear activation

TL;DR

This paper presents a novel graphene-ITO hybrid integrated modulator on silicon photonics that enables efficient nonlinear activation functions for photonic neural networks, improving inference accuracy and energy efficiency.

Contribution

It introduces a new electro-optic device with diode-like nonlinearity using heterostructures, enhancing photonic neural network performance.

Findings

Demonstrated a ReLU-like optical nonlinearity in the device.

Showed improved inference accuracy in photonic neural networks.

Achieved better energy efficiency with the new nonlinearity.

Abstract

The high demand for machine intelligence of doubling every three months is driving novel hardware solutions beyond charging of electrical wires given a resurrection to application specific integrated circuit (ASIC)-based accelerators. These innovations include photonic ASICs (P-ASIC) due to prospects of performing optical linear (and also nonlinear) operations, such as multiply-accumulate for vector matrix multiplications or convolutions, without iterative architectures. Such photonic linear algebra enables picosecond delay when photonic integrated circuits are utilized, via on-the-fly mathematics. However, the neurons full function includes providing a nonlinear activation function, knowns as thresholding, to enable decision making on inferred data. Many P-ASIC solutions performing this nonlinearity in the electronic domain, which brings challenges in terms of data throughput and delay, thus breaking the optical link and introducing increased system complexity via domain crossings. This work follows the notion of utilizing enhanced light-matter-interactions to provide efficient, compact, and engineerable electro-optic neuron nonlinearity. Here, we introduce and demonstrate a novel electro-optic device to engineer the shape of this optical nonlinearity to resemble a rectifying linear unit (ReLU) - the most-commonly used nonlinear activation function in neural networks. We combine the counter-directional transfer functions from heterostructures made out of two electro-optic materials to design a diode-like nonlinear response of the device. Integrating this nonlinearity into a photonic neural network, we show how the electrostatics of this thresholders gating junction improves machine learning inference accuracy and the energy efficiency of the neural network.