Discovering Parametric Activation Functions

TL;DR

This paper introduces an automated method combining evolutionary search and gradient descent to discover and optimize activation functions, leading to consistent performance improvements across various neural network architectures on image classification tasks.

Contribution

It presents a novel automated approach for discovering and customizing activation functions, enhancing deep learning performance beyond traditional functions like ReLU.

Findings

Consistently improves accuracy over ReLU on CIFAR datasets

Discovers both general and architecture-specific activation functions

Effective across multiple neural network architectures

Abstract

Recent studies have shown that the choice of activation function can significantly affect the performance of deep learning networks. However, the benefits of novel activation functions have been inconsistent and task dependent, and therefore the rectified linear unit (ReLU) is still the most commonly used. This paper proposes a technique for customizing activation functions automatically, resulting in reliable improvements in performance. Evolutionary search is used to discover the general form of the function, and gradient descent to optimize its parameters for different parts of the network and over the learning process. Experiments with four different neural network architectures on the CIFAR-10 and CIFAR-100 image classification datasets show that this approach is effective. It discovers both general activation functions and specialized functions for different architectures, consistently improving accuracy over ReLU and other activation functions by significant margins. The approach can therefore be used as an automated optimization step in applying deep learning to new tasks.