TanhExp: A Smooth Activation Function with High Convergence Speed for Lightweight Neural Networks

TL;DR

The paper introduces TanhExp, a novel smooth activation function that significantly enhances the convergence speed and accuracy of lightweight neural networks in image classification tasks, without increasing network size.

Contribution

Proposes TanhExp, a new activation function that improves performance and robustness of lightweight neural networks efficiently.

Findings

TanhExp outperforms existing activation functions in convergence speed.

TanhExp improves accuracy of lightweight networks on various datasets.

TanhExp remains stable under noisy conditions and dataset variations.

Abstract

Lightweight or mobile neural networks used for real-time computer vision tasks contain fewer parameters than normal networks, which lead to a constrained performance. In this work, we proposed a novel activation function named Tanh Exponential Activation Function (TanhExp) which can improve the performance for these networks on image classification task significantly. The definition of TanhExp is f(x) = xtanh(e^x). We demonstrate the simplicity, efficiency, and robustness of TanhExp on various datasets and network models and TanhExp outperforms its counterparts in both convergence speed and accuracy. Its behaviour also remains stable even with noise added and dataset altered. We show that without increasing the size of the network, the capacity of lightweight neural networks can be enhanced by TanhExp with only a few training epochs and no extra parameters added.