LAP: An Attention-Based Module for Concept Based Self-Interpretation and Knowledge Injection in Convolutional Neural Networks

TL;DR

This paper introduces LAP, an attention-based pooling layer that enhances CNN interpretability and allows knowledge injection without performance loss, applicable to trained models, verified on ImageNet.

Contribution

Proposes a pluggable attention-based pooling layer called LAP that provides self-interpretability and knowledge injection in CNNs without degrading accuracy.

Findings

LAP improves model interpretability over traditional explainers.

LAP can be integrated into existing CNNs, including trained models.

LAP maintains or improves classification performance.

Abstract

Despite the state-of-the-art performance of deep convolutional neural networks, they are susceptible to bias and malfunction in unseen situations. Moreover, the complex computation behind their reasoning is not human-understandable to develop trust. External explainer methods have tried to interpret network decisions in a human-understandable way, but they are accused of fallacies due to their assumptions and simplifications. On the other side, the inherent self-interpretability of models, while being more robust to the mentioned fallacies, cannot be applied to the already trained models. In this work, we propose a new attention-based pooling layer, called Local Attention Pooling (LAP), that accomplishes self-interpretability and the possibility for knowledge injection without performance loss. The module is easily pluggable into any convolutional neural network, even the already trained ones. We have defined a weakly supervised training scheme to learn the distinguishing features in decision-making without depending on experts' annotations. We verified our claims by evaluating several LAP-extended models on two datasets, including ImageNet. The proposed framework offers more valid human-understandable and faithful-to-the-model interpretations than the commonly used white-box explainer methods.