Neural Networks from Biological to Artificial and Vice Versa

TL;DR

This paper explores the impact of dead neurons on artificial neural network performance, investigates training influences, and draws analogies between neural network optimizers and brain learning processes, with implications for neurological health.

Contribution

It specifically examines how dead neurons affect ANNs and explores the analogy between the Adam optimizer and brain learning, linking artificial and biological neural systems.

Findings

Dead neurons influence neighboring neurons and overall network performance.

Training algorithms and activation functions impact the effects of dead neurons.

Analogies between Adam optimizer and brain learning processes are discussed.

Abstract

In this paper, we examine how deep learning can be utilized to investigate neural health and the difficulties in interpreting neurological analyses within algorithmic models. The key contribution of this paper is the investigation of the impact of a dead neuron on the performance of artificial neural networks (ANNs). Therefore, we conduct several tests using different training algorithms and activation functions to identify the precise influence of the training process on neighboring neurons and the overall performance of the ANN in such cases. The aim is to assess the potential application of the findings in the biological domain, the expected results may have significant implications for the development of effective treatment strategies for neurological disorders. Successive training phases that incorporate visual and acoustic data derived from past social and familial experiences could be suggested to achieve this goal. Finally, we explore the conceptual analogy between the Adam optimizer and the learning process of the brain by delving into the specifics of both systems while acknowledging their fundamental differences.