Set Aggregation Network as a Trainable Pooling Layer

TL;DR

This paper introduces the Set Aggregation Network (SAN), a trainable pooling layer that can embed sets of features into fixed-size vectors, improving classification accuracy and reducing overfitting in neural networks.

Contribution

SAN provides a flexible, trainable pooling mechanism that preserves input information and enhances model performance compared to traditional pooling methods.

Findings

SAN improves classification accuracy.

SAN reduces overfitting and acts as a regularizer.

SAN can embed sets into vectors of arbitrary size.

Abstract

Global pooling, such as max- or sum-pooling, is one of the key ingredients in deep neural networks used for processing images, texts, graphs and other types of structured data. Based on the recent DeepSets architecture proposed by Zaheer et al. (NIPS 2017), we introduce a Set Aggregation Network (SAN) as an alternative global pooling layer. In contrast to typical pooling operators, SAN allows to embed a given set of features to a vector representation of arbitrary size. We show that by adjusting the size of embedding, SAN is capable of preserving the whole information from the input. In experiments, we demonstrate that replacing global pooling layer by SAN leads to the improvement of classification accuracy. Moreover, it is less prone to overfitting and can be used as a regularizer.