Activation Functions in Artificial Neural Networks: A Systematic Overview

TL;DR

This paper provides a comprehensive and current overview of various activation functions used in neural networks, highlighting their properties and significance in deep learning.

Contribution

It offers an analytic synthesis of both traditional and recent activation functions, clarifying their roles and differences in neural network performance.

Findings

Summarizes key properties of popular activation functions

Highlights the proliferation of new activation functions in deep learning

Serves as a resource for researchers and practitioners

Abstract

Activation functions shape the outputs of artificial neurons and, therefore, are integral parts of neural networks in general and deep learning in particular. Some activation functions, such as logistic and relu, have been used for many decades. But with deep learning becoming a mainstream research topic, new activation functions have mushroomed, leading to confusion in both theory and practice. This paper provides an analytic yet up-to-date overview of popular activation functions and their properties, which makes it a timely resource for anyone who studies or applies neural networks.