Beyond One-Hot-Encoding: Injecting Semantics to Drive Image Classifiers

TL;DR

This paper proposes a method to incorporate semantic and ontological information into image classifier training, moving beyond traditional one-hot labels to improve interpretability, robustness, and accuracy by using enriched loss functions.

Contribution

It introduces a generic approach to integrate semantic knowledge into loss functions for image classification, enhancing model interpretability and robustness.

Findings

Semantic-enriched loss improves interpretability.

Trade-offs between accuracy and mistake severity are analyzed.

Method enhances robustness against adversarial attacks.

Abstract

Images are loaded with semantic information that pertains to real-world ontologies: dog breeds share mammalian similarities, food pictures are often depicted in domestic environments, and so on. However, when training machine learning models for image classification, the relative similarities amongst object classes are commonly paired with one-hot-encoded labels. According to this logic, if an image is labelled as 'spoon', then 'tea-spoon' and 'shark' are equally wrong in terms of training loss. To overcome this limitation, we explore the integration of additional goals that reflect ontological and semantic knowledge, improving model interpretability and trustworthiness. We suggest a generic approach that allows to derive an additional loss term starting from any kind of semantic information about the classification label. First, we show how to apply our approach to ontologies and word embeddings, and discuss how the resulting information can drive a supervised learning process. Second, we use our semantically enriched loss to train image classifiers, and analyse the trade-offs between accuracy, mistake severity, and learned internal representations. Finally, we discuss how this approach can be further exploited in terms of explainability and adversarial robustness. Code repository: https://github.com/S1M0N38/semantic-encodings