Explaining Neural Network Predictions on Sentence Pairs via Learning Word-Group Masks
arXiv:2104.04488
Abstract
Explaining neural network models is important for increasing their trustworthiness in real-world applications. Most existing methods generate post-hoc explanations for neural network models by identifying individual feature attributions or detecting interactions between adjacent features. However, for models with text pairs as inputs (e.g., paraphrase identification), existing methods are not sufficient to capture feature interactions between two texts and their simple extension of computing all word-pair interactions between two texts is computationally inefficient. In this work, we propose the Group Mask (GMASK) method to implicitly detect word correlations by grouping correlated words from the input text pair together and measure their contribution to the corresponding NLP tasks as a whole. The proposed method is evaluated with two different model architectures (decomposable attention model and BERT) across four datasets, including natural language inference and paraphrase identification tasks. Experiments show the effectiveness of GMASK in providing faithful explanations to these models.
arXiv commentNAACL-HLT 2021
References in corpus (16)
- Categorical Reparameterization with Gumbel-Softmax
- Axiomatic Attribution for Deep Networks
- The information bottleneck method
- Reasoning about Entailment with Neural Attention
- Consistent Individualized Feature Attribution for Tree Ensembles
- Attention is not Explanation
- Understanding Neural Networks through Representation Erasure
- e-SNLI: Natural Language Inference with Natural Language Explanations
- Beyond Word Importance: Contextual Decomposition to Extract Interactions from LSTMs
- Bilateral Multi-Perspective Matching for Natural Language Sentences
- Hierarchical interpretations for neural network predictions
- A Decomposable Attention Model for Natural Language Inference
- Interpretation of Prediction Models Using the Input Gradient
- Towards Hierarchical Importance Attribution: Explaining Compositional Semantics for Neural Sequence Models
- Can I trust you more? Model-Agnostic Hierarchical Explanations
- Interpreting Recurrent and Attention-Based Neural Models: a Case Study on Natural Language Inference