Background Knowledge Injection for Interpretable Sequence Classification

TL;DR

This paper presents a novel sequence classification method that combines linear classifiers with background knowledge embeddings, enhancing interpretability without sacrificing accuracy, demonstrated on activity recognition and amino acid sequences.

Contribution

Introduces a new sequence learning algorithm integrating background knowledge embeddings with linear classifiers for improved interpretability.

Findings

Maintains predictive accuracy comparable to existing methods.

Produces more interpretable models through knowledge-augmented features.

Proposes a new interpretability measure based on symbol embeddings.

Abstract

Sequence classification is the supervised learning task of building models that predict class labels of unseen sequences of symbols. Although accuracy is paramount, in certain scenarios interpretability is a must. Unfortunately, such trade-off is often hard to achieve since we lack human-independent interpretability metrics. We introduce a novel sequence learning algorithm, that combines (i) linear classifiers - which are known to strike a good balance between predictive power and interpretability, and (ii) background knowledge embeddings. We extend the classic subsequence feature space with groups of symbols which are generated by background knowledge injected via word or graph embeddings, and use this new feature space to learn a linear classifier. We also present a new measure to evaluate the interpretability of a set of symbolic features based on the symbol embeddings. Experiments on human activity recognition from wearables and amino acid sequence classification show that our classification approach preserves predictive power, while delivering more interpretable models.