Arbitrary Precision Printed Ternary Neural Networks with Holistic Evolutionary Approximation

2025-08-27

AI Summary

To address the trade-off between classification accuracy and area/power efficiency in printed ternary neural networks, this paper proposes the first automated design framework for printed ternary neural networks supporting arbitrary input precision. The method integrates analog-digital co-design, multi-objective evolutionary optimization, and full-system approximate computing to enable end-to-end joint optimization—from analog-to-digital interface to ternary classifier. Innovatively, it unifies analog front-end modeling with digital neural architecture search. Under a <5% accuracy degradation constraint, the framework achieves, on average, a 17× reduction in silicon area and a 59× reduction in power consumption. It enables, for the first time, a battery-powered, near-sensor intelligent processing system implemented entirely using printed electronics. This work establishes a scalable design paradigm for flexible, low-cost edge AI hardware.

Technology Category

• Machine Learning: Hardware-aware ML
• Search and Optimization: Learning to Search
• Natural Language Processing: Learning & Optimization for NLP

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems

Abstract

Printed electronics offer a promising alternative for applications beyond silicon-based systems, requiring properties like flexibility, stretchability, conformality, and ultra-low fabrication costs. Despite the large feature sizes in printed electronics, printed neural networks have attracted attention for meeting target application requirements, though realizing complex circuits remains challenging. This work bridges the gap between classification accuracy and area efficiency in printed neural networks, covering the entire processing-near-sensor system design and co-optimization from the analog-to-digital interface-a major area and power bottleneck-to the digital classifier. We propose an automated framework for designing printed Ternary Neural Networks with arbitrary input precision, utilizing multi-objective optimization and holistic approximation. Our circuits outperform existing approximate printed neural networks by 17x in area and 59x in power on average, being the first to enable printed-battery-powered operation with under 5% accuracy loss while accounting for analog-to-digital interfacing costs.

Problem

• Bridging accuracy and area efficiency in printed neural networks
• Addressing analog-to-digital interface as major bottleneck
• Enabling printed-battery-powered operation with minimal accuracy loss

Innovation

• Holistic evolutionary approximation for neural networks
• Arbitrary precision printed ternary neural networks
• Multi-objective optimization covering analog-digital interface