Understanding Spending Behavior: Recurrent Neural Network Explanation and Interpretation

TL;DR

This paper enhances the explainability of RNN-based micro-segmentation in finance by extracting symbolic explanations and interpreting the dynamics of the model's state space, facilitating better understanding and trust.

Contribution

It introduces a method to interpret RNN temporal features through linear regression and dynamical systems, addressing the opacity of deep models in sensitive financial applications.

Findings

Linear regression accurately reconstructs RNN features.

Regression coefficients reveal hidden data relationships.

Attractors in the state space can be identified and labeled.

Abstract

Micro-segmentation of customers in the finance sector is a non-trivial task and has been an atypical omission from recent scientific literature. Where traditional segmentation classifies customers based on coarse features such as demographics, micro-segmentation depicts more nuanced differences between individuals, bringing forth several advantages including the potential for improved personalization in financial services. AI and representation learning offer a unique opportunity to solve the problem of micro-segmentation. Although ubiquitous in many industries, the proliferation of AI in sensitive industries such as finance has become contingent on the explainability of deep models. We had previously solved the micro-segmentation problem by extracting temporal features from the state space of a recurrent neural network (RNN). However, due to the inherent opacity of RNNs our solution lacked an explanation. In this study, we address this issue by extracting a symbolic explanation for our model and providing an interpretation of our temporal features. For the explanation, we use a linear regression model to reconstruct the features in the state space with high fidelity. We show that our linear regression coefficients have not only learned the rules used to recreate the features, but have also learned the relationships that were not directly evident in the raw data. Finally, we propose a novel method to interpret the dynamics of the state space by using the principles of inverse regression and dynamical systems to locate and label a set of attractors.