How does this interaction affect me? Interpretable attribution for feature interactions

Abstract

Machine learning transparency calls for interpretable explanations of how inputs relate to predictions. Feature attribution is a way to analyze the impact of features on predictions. Feature interactions are the contextual dependence between features that jointly impact predictions. There are a number of methods that extract feature interactions in prediction models; however, the methods that assign attributions to interactions are either uninterpretable, model-specific, or non-axiomatic. We propose an interaction attribution and detection framework called Archipelago which addresses these problems and is also scalable in real-world settings. Our experiments on standard annotation labels indicate our approach provides significantly more interpretable explanations than comparable methods, which is important for analyzing the impact of interactions on predictions. We also provide accompanying visualizations of our approach that give new insights into deep neural networks.

References in corpus (13)

• HuggingFace's Transformers: State-of-the-art Natural Language Processing
• Axiomatic Attribution for Deep Networks
• Learning Important Features Through Propagating Activation Differences
• Can AI help in screening Viral and COVID-19 pneumonia?
• Predictive learning via rule ensembles
• COVID-19 Image Data Collection
• Consistent Individualized Feature Attribution for Tree Ensembles
• Understanding Neural Networks through Representation Erasure
• Explaining Explanations: Axiomatic Feature Interactions for Deep Networks
• Towards Hierarchical Importance Attribution: Explaining Compositional Semantics for Neural Sequence Models
• Feature Interaction Interpretability: A Case for Explaining Ad-Recommendation Systems via Neural Interaction Detection
• Can I trust you more? Model-Agnostic Hierarchical Explanations
• The Shapley Taylor Interaction Index

Cited by in corpus (6)

• Beyond TreeSHAP: Efficient Computation of Any-Order Shapley Interactions for Tree Ensembles
• PredDiff: Explanations and Interactions from Conditional Expectations
• Algorithm Fairness in AI for Medicine and Healthcare
• Logic Traps in Evaluating Attribution Scores
• Interpretable Artificial Intelligence through the Lens of Feature Interaction
• Fine-grained Interpretation and Causation Analysis in Deep NLP Models