Can One Design a Series of Brains for Neuromorphic Computing to solve complex inverse problems

TL;DR

This paper discusses designing neuromorphic systems with memristive circuits to effectively solve complex inverse problems, emphasizing training algorithms for optimal memristive configurations.

Contribution

It introduces a novel approach to train memristive circuits within neuromorphic computing to address ill-posed inverse problems based on finite element methods.

Findings

Proposed a method to train memristive circuits for inverse problems

Demonstrated the potential of neuromorphic systems in complex problem solving

Analyzed the dynamics of memristive circuits in this context

Abstract

In this position paper, we present a discussion on neuromorphic computing and especially the learning/training algorithm to design a series of brains with different memristive values to solve complex ill-posed inverse problems based on a Finite Element(FE) method. First, the neuromorphic computing is addressed and we focus on a type of memristive circuit computing that falls into the scope of neuromorphic computing. Secondly based on reference [1] in which the complex dynamics of the complex memristive circuit was studied, we design a method and an approach to train the memristive circuit so that the memristive values are optimally obtained.