Language Model Meets Prototypes: Towards Interpretable Text Classification Models through Prototypical Networks
Ximing Wen
TL;DR
This work addresses interpretability in transformer-based NLP by integrating prototype networks with LM encoders. It first develops a Transformer-prototype model for contextual sarcasm detection that uses semantic prototypes and an incongruity loss to capture explicit versus implicit sentiment. It then introduces a white-box Graph Attention prototype network for text classification, achieving competitive accuracy with interpretable prototype-based explanations and visualizations. Finally, it outlines extending the framework to document classification via graph neural networks and contrastive learning to strengthen interpretability without sacrificing performance.
Abstract
Pretrained transformer-based Language Models (LMs) are well-known for their ability to achieve significant improvement on NLP tasks, but their black-box nature, which leads to a lack of interpretability, has been a major concern. My dissertation focuses on developing intrinsically interpretable models when using LMs as encoders while maintaining their superior performance via prototypical networks. I initiated my research by investigating enhancements in performance for interpretable models of sarcasm detection. My proposed approach focuses on capturing sentiment incongruity to enhance accuracy while offering instance-based explanations for the classification decisions. Later, I developed a novel white-box multi-head graph attention-based prototype network designed to explain the decisions of text classification models without sacrificing the accuracy of the original black-box LMs. In addition, I am working on extending the attention-based prototype network with contrastive learning to redesign an interpretable graph neural network, aiming to enhance both the interpretability and performance of the model in document classification.