ECLAD: Extracting Concepts with Local Aggregated Descriptors

TL;DR

ECLAD introduces a new method for extracting and localizing concepts in CNNs using pixel-wise aggregated descriptors, validated on synthetic and real datasets, improving over existing techniques.

Contribution

The paper presents a novel concept extraction and localization approach based on pixel-wise aggregation of CNN features, with a validation process using synthetic datasets.

Findings

Outperforms state-of-the-art concept extraction methods

Effective on both synthetic and real-world datasets

Reduces need for human annotation in validation

Abstract

Convolutional neural networks (CNNs) are increasingly being used in critical systems, where robustness and alignment are crucial. In this context, the field of explainable artificial intelligence has proposed the generation of high-level explanations of the prediction process of CNNs through concept extraction. While these methods can detect whether or not a concept is present in an image, they are unable to determine its location. What is more, a fair comparison of such approaches is difficult due to a lack of proper validation procedures. To address these issues, we propose a novel method for automatic concept extraction and localization based on representations obtained through pixel-wise aggregations of CNN activation maps. Further, we introduce a process for the validation of concept-extraction techniques based on synthetic datasets with pixel-wise annotations of their main components, reducing the need for human intervention. Extensive experimentation on both synthetic and real-world datasets demonstrates that our method outperforms state-of-the-art alternatives.