Multi-layer Representation Learning for Robust OOD Image Classification

TL;DR

This paper proposes a multi-layer feature extraction approach using Hypercolumns in ResNet-18 to improve robustness and accuracy in out-of-distribution image classification tasks.

Contribution

It introduces a novel application of Hypercolumns to enhance OOD classification performance in CNNs, demonstrating significant accuracy improvements.

Findings

Improved accuracy on NICO dataset with Hypercolumns

Multi-layer features aid in trustworthy OOD predictions

ResNet-18 performance increases with intermediate layer features

Abstract

Convolutional Neural Networks have become the norm in image classification. Nevertheless, their difficulty to maintain high accuracy across datasets has become apparent in the past few years. In order to utilize such models in real-world scenarios and applications, they must be able to provide trustworthy predictions on unseen data. In this paper, we argue that extracting features from a CNN's intermediate layers can assist in the model's final prediction. Specifically, we adapt the Hypercolumns method to a ResNet-18 and find a significant increase in the model's accuracy, when evaluating on the NICO dataset.