The Resurrection of the ReLU

TL;DR

This paper introduces SUGAR, a surrogate gradient method for ReLU that enhances its performance and sparsity, effectively reviving the classical activation function across various deep learning architectures.

Contribution

The paper proposes a novel surrogate gradient technique for ReLU, improving its generalization, sparsity, and effectiveness in modern deep learning models.

Findings

SUGAR improves generalization in CNN architectures like VGG-16 and ResNet-18.

SUGAR enhances sparsity and resurrects dead ReLUs effectively.

Replacing GELU with SUGAR yields competitive or better results in advanced models.

Abstract

Modeling sophisticated activation functions within deep learning architectures has evolved into a distinct research direction. Functions such as GELU, SELU, and SiLU offer smooth gradients and improved convergence properties, making them popular choices in state-of-the-art models. Despite this trend, the classical ReLU remains appealing due to its simplicity, inherent sparsity, and other advantageous topological characteristics. However, ReLU units are prone to becoming irreversibly inactive - a phenomenon known as the dying ReLU problem - which limits their overall effectiveness. In this work, we introduce surrogate gradient learning for ReLU (SUGAR) as a novel, plug-and-play regularizer for deep architectures. SUGAR preserves the standard ReLU function during the forward pass but replaces its derivative in the backward pass with a smooth surrogate that avoids zeroing out gradients. We demonstrate that SUGAR, when paired with a well-chosen surrogate function, substantially enhances generalization performance over convolutional network architectures such as VGG-16 and ResNet-18, providing sparser activations while effectively resurrecting dead ReLUs. Moreover, we show that even in modern architectures like Conv2NeXt and Swin Transformer - which typically employ GELU - substituting these with SUGAR yields competitive and even slightly superior performance. These findings challenge the prevailing notion that advanced activation functions are necessary for optimal performance. Instead, they suggest that the conventional ReLU, particularly with appropriate gradient handling, can serve as a strong, versatile revived classic across a broad range of deep learning vision models.