Empirical study of the modulus as activation function in computer vision applications

TL;DR

This paper introduces the modulus activation function for computer vision, demonstrating it improves generalization, avoids vanishing gradients, and is suitable for hardware applications, outperforming traditional nonlinearities.

Contribution

It proposes a new non-monotonic activation function, the modulus, and empirically shows its advantages in accuracy and gradient behavior in vision tasks.

Findings

Up to 15% accuracy increase on CIFAR100

Eliminates vanishing gradient and dying neurons problems

Suitable for TinyML and hardware implementations

Abstract

In this work we propose a new non-monotonic activation function: the modulus. The majority of the reported research on nonlinearities is focused on monotonic functions. We empirically demonstrate how by using the modulus activation function on computer vision tasks the models generalize better than with other nonlinearities - up to a 15% accuracy increase in CIFAR100 and 4% in CIFAR10, relative to the best of the benchmark activations tested. With the proposed activation function the vanishing gradient and dying neurons problems disappear, because the derivative of the activation function is always 1 or -1. The simplicity of the proposed function and its derivative make this solution specially suitable for TinyML and hardware applications.