Aligning Characteristic Descriptors with Images for Human-Expert-like Explainability

TL;DR

A novel approach is proposed that utilizes characteristic descriptors to explain model decisions by identifying their presence in images, thereby generating expert-like explanations and incorporates a concept bottleneck layer within the model architecture, which calculates the similarity between image and descriptor encodings to deliver inherent and faithful explanations.

Abstract

In mission-critical domains such as law enforcement and medical diagnosis, the ability to explain and interpret the outputs of deep learning models is crucial for ensuring user trust and supporting informed decision-making. Despite advancements in explainability, existing methods often fall short in providing explanations that mirror the depth and clarity of those given by human experts. Such expert-level explanations are essential for the dependable application of deep learning models in law enforcement and medical contexts. Additionally, we recognize that most explanations in real-world scenarios are communicated primarily through natural language. Addressing these needs, we propose a novel approach that utilizes characteristic descriptors to explain model decisions by identifying their presence in images, thereby generating expert-like explanations. Our method incorporates a concept bottleneck layer within the model architecture, which calculates the similarity between image and descriptor encodings to deliver inherent and faithful explanations. Through experiments in face recognition and chest X-ray diagnosis, we demonstrate that our approach offers a significant contrast over existing techniques, which are often limited to the use of saliency maps. We believe our approach represents a significant step toward making deep learning systems more accountable, transparent, and trustworthy in the critical domains of face recognition and medical diagnosis.

Figures (9)

• Figure 1: Illustration of our proposed explainable system for face recognition and chest x-ray diagnosis.
• Figure 2: Proposed architecture produces explanations and embeddings for tasks like classification or face recognition. It can work with or without concept supervision. If concept labels are available, they are used for back-propagating the loss; otherwise, only the task loss is back-propagated.
• Figure 3: Illustrative examples of some of the a) facial characteristic descriptors as recommended in the FISWG Facial Image Comparison Guide. b) characteristic descriptors for chest x-ray diagnosis as defined by radiologists.
• Figure 4: Examples of explanations provided by our model for its decisions in face recognition and chest x-ray diagnosis.
• Figure 5: Proposed Explainable Face Verification Pipeline.