Efficient Decompositional Rule Extraction for Deep Neural Networks
This addresses the need for efficient and scalable interpretability tools for DNNs in domains like healthcare and physics, though it is incremental as it builds on existing decompositional rule extraction methods.
The paper tackles the problem of extracting rule-based models from deep neural networks (DNNs) to improve interpretability and debuggability, introducing ECLAIRE, a polynomial-time algorithm that scales to large DNNs and datasets, and shows it extracts more accurate and comprehensible rule sets with orders of magnitude less computational resources than state-of-the-art methods.
In recent years, there has been significant work on increasing both interpretability and debuggability of a Deep Neural Network (DNN) by extracting a rule-based model that approximates its decision boundary. Nevertheless, current DNN rule extraction methods that consider a DNN's latent space when extracting rules, known as decompositional algorithms, are either restricted to single-layer DNNs or intractable as the size of the DNN or data grows. In this paper, we address these limitations by introducing ECLAIRE, a novel polynomial-time rule extraction algorithm capable of scaling to both large DNN architectures and large training datasets. We evaluate ECLAIRE on a wide variety of tasks, ranging from breast cancer prognosis to particle detection, and show that it consistently extracts more accurate and comprehensible rule sets than the current state-of-the-art methods while using orders of magnitude less computational resources. We make all of our methods available, including a rule set visualisation interface, through the open-source REMIX library (https://github.com/mateoespinosa/remix).