L*ReLU: Piece-wise Linear Activation Functions for Deep Fine-grained Visual Categorization

TL;DR

This paper introduces L*ReLU, a piece-wise linear activation function designed to improve fine-grained visual categorization by better modeling subtle feature details, outperforming traditional ReLUs across multiple datasets.

Contribution

The paper proposes L*ReLU, a novel activation function that enhances neural network performance in fine-grained visual tasks by capturing subtle feature variations.

Findings

L*ReLU achieves superior accuracy on seven benchmark datasets.

Different tasks benefit from different activation functions.

L*ReLU maintains computational efficiency similar to ReLUs.

Abstract

Deep neural networks paved the way for significant improvements in image visual categorization during the last years. However, even though the tasks are highly varying, differing in complexity and difficulty, existing solutions mostly build on the same architectural decisions. This also applies to the selection of activation functions (AFs), where most approaches build on Rectified Linear Units (ReLUs). In this paper, however, we show that the choice of a proper AF has a significant impact on the classification accuracy, in particular, if fine, subtle details are of relevance. Therefore, we propose to model the degree of absence and the presence of features via the AF by using piece-wise linear functions, which we refer to as L*ReLU. In this way, we can ensure the required properties, while still inheriting the benefits in terms of computational efficiency from ReLUs. We demonstrate our approach for the task of Fine-grained Visual Categorization (FGVC), running experiments on seven different benchmark datasets. The results do not only demonstrate superior results but also that for different tasks, having different characteristics, different AFs are selected.