Evolutionary Approximation of Ternary Neurons for On-sensor Printed Neural Networks

TL;DR

This paper introduces a novel evolutionary approach to design printed neural networks with ternary neurons, achieving significant improvements in area and power efficiency for on-sensor applications.

Contribution

It presents the first open-source digital printed neural network classifiers using ternary neurons optimized via evolutionary methods, suitable for printed electronics.

Findings

At least 6x reduction in area compared to previous designs.

Up to 19x power savings.

First open-source printed neural network classifiers compatible with energy harvesters.

Abstract

Printed electronics offer ultra-low manufacturing costs and the potential for on-demand fabrication of flexible hardware. However, significant intrinsic constraints stemming from their large feature sizes and low integration density pose design challenges that hinder their practicality. In this work, we conduct a holistic exploration of printed neural network accelerators, starting from the analog-to-digital interface - a major area and power sink for sensor processing applications - and extending to networks of ternary neurons and their implementation. We propose bespoke ternary neural networks using approximate popcount and popcount-compare units, developed through a multi-phase evolutionary optimization approach and interfaced with sensors via customizable analog-to-binary converters. Our evaluation results show that the presented designs outperform the state of the art, achieving at least 6x improvement in area and 19x in power. To our knowledge, they represent the first open-source digital printed neural network classifiers capable of operating with existing printed energy harvesters.