Bandits for Learning to Explain from Explanations

2021/02/07 by Freya Behrens, Behrens, Freya, Stefano Teso +3
Computer Science · Decision Sciences · #Advanced Bandit Algorithms Research #Explainable Artificial Intelligence (XAI) #FOS: Computer and information sciences #Machine Learning (cs.LG) #Machine Learning (stat.ML) #Machine Learning and Algorithms

paper · pdf · doi:10.48550/arxiv.2102.03815

openalex publication_date 2021/02/07 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28

Abstract

We introduce Explearn, an online algorithm that learns to jointly output predictions and explanations for those predictions. Explearn leverages Gaussian Processes (GP)-based contextual bandits. This brings two key benefits. First, GPs naturally capture different kinds of explanations and enable the system designer to control how explanations generalize across the space by virtue of choosing a suitable kernel. Second, Explearn builds on recent results in contextual bandits which guarantee convergence with high probability. Our initial experiments hint at the promise of the approach.

Citations

• A simpler condition for consistency of a kernel independence test
• Understanding Black-box Predictions via Influence Functions
• Interpreting Blackbox Models via Model Extraction
• Local Rule-Based Explanations of Black Box Decision Systems
• Towards Robust Interpretability with Self-Explaining Neural Networks
• A Survey of Methods for Explaining Black Box Models
• Machine Guides, Human Supervises: Interactive Learning with Global\n Explanations

Related

• Contextual Explanation Networks
• Explaining Explanations: An Overview of Interpretability of Machine\n Learning
• GLEAMS: Bridging the Gap Between Local and Global Explanations
• Parallel Sentence-Level Explanation Generation for Real-World Low-Resource Scenarios
• Global Explanations of Neural Networks: Mapping the Landscape of Predictions