AFN: Adaptive Fusion Normalization via an Encoder-Decoder Framework

TL;DR

This paper introduces Adaptive Fusion Normalization (AFN), a new normalization method that combines existing techniques to improve domain generalization and image classification performance.

Contribution

The paper proposes AFN, a novel normalization function that adaptively fuses multiple normalization techniques within an encoder-decoder framework.

Findings

AFN outperforms previous normalization methods in domain generalization.

AFN achieves superior results in image classification tasks.

AFN effectively mitigates weaknesses of individual normalization functions.

Abstract

The success of deep learning is inseparable from normalization layers. Researchers have proposed various normalization functions, and each of them has both advantages and disadvantages. In response, efforts have been made to design a unified normalization function that combines all normalization procedures and mitigates their weaknesses. We also proposed a new normalization function called Adaptive Fusion Normalization. Through experiments, we demonstrate AFN outperforms the previous normalization techniques in domain generalization and image classification tasks.