Autonomous learning by simple dynamical systems with a discrete-time formulation

TL;DR

This paper introduces a discrete-time formulation for autonomous learning applicable to systems with intermittent error evaluation, demonstrated on neural networks and phase oscillators, expanding learning applicability in finite-time processing systems.

Contribution

It presents a novel discrete-time autonomous learning scheme suitable for systems where error measurement is not continuous, broadening the scope of autonomous learning methods.

Findings

Applicable to neural networks and oscillators

Handles systems with finite-time information processing

Demonstrates effectiveness on classification and synchronization tasks

Abstract

We present a discrete-time formulation for the autonomous learning conjecture. The main feature of this formulation is the possibility to apply the autonomous learning scheme to systems in which the errors with respect to target functions are not well-defined for all times. This restriction for the evaluation of functionality is a typical feature in systems that need a finite time interval to process a unit piece of information. We illustrate its application on an artificial neural network with feed-forward architecture for classification and a phase oscillator system with synchronization properties. The main characteristics of the discrete-time formulation are shown by constructing these systems with predefined functions.