Design of Oscillatory Neural Networks by Machine Learning - arXiv 2309.02532

Aug 2, 2026

Upload by Anda

Machine learning is used to design Oscillatory Neural Networks (ONNs) with resistively coupled ring oscillators. A circuit model is built in a machine-learning-enabled simulator, and Backpropagation through time (BPTT) determines coupling resistances. The approach yields associative memories and multi-layer ONN classifiers, showing superior performance versus methods such as Hebbian learning and enabling simpler circuit topologies. Results also demonstrate that multi-layer ONNs outperform single-layer designs, supporting the view that machine learning unlocks the hardware computing potential of ONNs.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages12

Reading Time30 min

AccessFree to download

Introduction

Analog computing and neuromorphic motivation

Oscillatory neural networks and limitations of Hebbian rules

Machine-learning-based design approach

Circuit model and learning framework

Differential equations and parameter learning

Application examples

Auto-associative memory

Multi-layer ONN classifier

Ring oscillator phase-based computation

arXiv :2309 .02532v1 [ cond-mat .dis-nn] 5 Sep 2023 Design of Oscillatory Neural Networks by Machine Learning Tams Rudner 1 , Wolfgang Porod2 , and Gyrgy Csaba 1 1 Pzmny Pter Catholic University, Faculty of Information Technology and Bionics, Budapest, Hungary 2 Center for Nano Science and Technology University of Notre Dame (NDnano), Notre Dame, IN, USA * [gcsaba@gmail.com](gcsaba@gmail.com) ABSTRACT We demonstrate the utility of machine learning algorithms for the design of Oscillatory Neural Networks (ONNs) . After constructing a circuit model of the oscillators in a machine-learning-enabled simulator and performing Backpropagation through time (BPTT) for determining the coupling resistances between the ring oscillators, we show the design of associative memories and multi-layered ONN classifiers. The machine-learning-designed ONNs show superior performance compared to other design methods (such as Hebbian learning) and they also enable significant simplifications in the circuit topology. We demonstrate the design of multi-layered ONNs that show superior performance compared to single-layer ones. We argue Machine learning can unlock the true computing potential of ONNs hardware. Introduction Throughout the history of electronics, analog computing devices were often looked at as the future of computing-only tobe turning up to be runner-ups behind their digital counterparts 1. Despite this history, research on analog computing is again on the rise, for multiple reasons. With the end of Dennard scaling2 , transistor resources are less plentiful and analog devices fit much better to sensory preprocessing functions then digital ones. New sensory processing pipelines require less and less digital number crunching and more neuromorphic, edge-AI functions, where analog computing devices may have an edge. For this reason, research is flourishing in analog hardware accelerators3 (aka neuromorphic hardware devices) that have the potential to boost the energy efficiency of AI processing pipelines by several orders of magnitude. Among the many flavors of analog computing, Oscillatory Neural Networks (ONNs) received special attention4. This is due to the facts that (1) ONNs are realizable by very simple circuits-either by emerging devices or CMOS devices (2) phasesand frequencies enable a rich and robust4 representation of information (3) biological systems seem to use oscillators to process information5-likely for a reason. Despite the significant current research efforts and the large literature, most ONNs seem to rely on some version of a Hebbian rule to define attractor states6 , so that the network converges to a stationary phase pattern, which is a the result of the computation. The Hebbian rule is used to calculate the value of physical couplings that define the circuit function. The reliance on the Hebbian rule turns ONNs to a sub-class of classical Hopfield networks-which are not very powerful by today’s standards. While there are a few ONN implementations not relying on basic Hebbian rules (notably7) it is likely that current ONNs does not fully exploit the potential of the hardware-due to the lack of a more powerful method to design the interconnections. In this paper we show that a state of art machine learning method, when applied to a SPICE-like model of the circuit, significantly enhances the computational power of ONNs. Our studied system is an ONN made of resistively coupled ring-oscillators8 ,9-its circuit topology is described in 1. Next, in Section 2 we develop the differential equations describing the circuit and show how a machine learning algorithm can be applied to design the circuit parameters. In Section 3, we exemplify the machine-learning framework for the design of an auto-associative memory and compare it to a standard Hebbian-rule based device. Section 3.3 furthers this concept by the design of a multi-layered network, which is a two-layer classifier and achieves superb performance c

FAQs

How does the method use machine learning to design ONNs?

It constructs a SPICE-like circuit model in a machine-learning-enabled simulator and applies Backpropagation through time (BPTT) to determine the coupling resistances between ring oscillators.

What tasks do the machine-learning-designed ONNs demonstrate?

They demonstrate associative memories and multi-layer ONN classifiers, including comparisons between multi-layer networks and single-layer ones.

Why is the proposed approach different from Hebbian learning?

The work argues that Hebbian rules lead ONNs to converge to stationary phase patterns and limit computational power, while machine learning enhances performance and simplifies the circuit topology.

Aug 2, 2026

Upload by Anda