Q-SENN: Quantized Self-Explaining Neural Networks

TL;DR

Q-SENN introduces a quantized, interpretable neural network that maintains high accuracy, improves explainability with binary class-feature relations, and aligns features with human concepts without extra supervision.

Contribution

The paper presents Q-SENN, a novel self-explaining neural network that enhances interpretability and applicability to complex datasets while preserving model accuracy.

Findings

Q-SENN outperforms previous models in interpretability and accuracy metrics.

It describes class-feature relationships as positive, negative, or neutral.

Features can be aligned with human concepts without additional supervision.

Abstract

Explanations in Computer Vision are often desired, but most Deep Neural Networks can only provide saliency maps with questionable faithfulness. Self-Explaining Neural Networks (SENN) extract interpretable concepts with fidelity, diversity, and grounding to combine them linearly for decision-making. While they can explain what was recognized, initial realizations lack accuracy and general applicability. We propose the Quantized-Self-Explaining Neural Network Q-SENN. Q-SENN satisfies or exceeds the desiderata of SENN while being applicable to more complex datasets and maintaining most or all of the accuracy of an uninterpretable baseline model, out-performing previous work in all considered metrics. Q-SENN describes the relationship between every class and feature as either positive, negative or neutral instead of an arbitrary number of possible relations, enforcing more binary human-friendly features. Since every class is assigned just 5 interpretable features on average, Q-SENN shows convincing local and global interpretability. Additionally, we propose a feature alignment method, capable of aligning learned features with human language-based concepts without additional supervision. Thus, what is learned can be more easily verbalized. The code is published: https://github.com/ThomasNorr/Q-SENN