Explainability Fact Sheets: A Framework for Systematic Assessment of Explainable Approaches

Abstract

Explanations in Machine Learning come in many forms, but a consensus regarding their desired properties is yet to emerge. In this paper we introduce a taxonomy and a set of descriptors that can be used to characterise and systematically assess explainable systems along five key dimensions: functional, operational, usability, safety and validation. In order to design a comprehensive and representative taxonomy and associated descriptors we surveyed the eXplainable Artificial Intelligence literature, extracting the criteria and desiderata that other authors have proposed or implicitly used in their research. The survey includes papers introducing new explainability algorithms to see what criteria are used to guide their development and how these algorithms are evaluated, as well as papers proposing such criteria from both computer science and social science perspectives. This novel framework allows to systematically compare and contrast explainability approaches, not just to better understand their capabilities but also to identify discrepancies between their theoretical qualities and properties of their implementations. We developed an operationalisation of the framework in the form of Explainability Fact Sheets, which enable researchers and practitioners alike to quickly grasp capabilities and limitations of a particular explainable method. When used as a Work Sheet, our taxonomy can guide the development of new explainability approaches by aiding in their critical evaluation along the five proposed dimensions.

Conference on Fairness, Accountability, and Transparency (FAT* '20), January 27-30, 2020, Barcelona, Spain

References in corpus (13)

• Towards A Rigorous Science of Interpretable Machine Learning
• Model Cards for Model Reporting
• Understanding Black-box Predictions via Influence Functions
• Model-Agnostic Interpretability of Machine Learning
• FACE: Feasible and Actionable Counterfactual Explanations
• Interpretable Predictions of Tree-based Ensembles via Actionable Feature Tweaking
• The Bayesian Case Model: A Generative Approach for Case-Based Reasoning and Prototype Classification
• A Nutritional Label for Rankings
• "Why Should You Trust My Explanation?" Understanding Uncertainty in LIME Explanations
• Using Visual Analytics to Interpret Predictive Machine Learning Models
• Defining Locality for Surrogates in Post-hoc Interpretablity
• Interpretable & Explorable Approximations of Black Box Models
• Towards a Characterization of Explainable Systems

Cited by in corpus (32)

• What Do We Want From Explainable Artificial Intelligence (XAI)? -- A Stakeholder Perspective on XAI and a Conceptual Model Guiding Interdisciplinary XAI Research
• From Anecdotal Evidence to Quantitative Evaluation Methods: A Systematic Review on Evaluating Explainable AI
• One Explanation Does Not Fit All: The Promise of Interactive Explanations for Machine Learning Transparency
• Interactive Model Cards: A Human-Centered Approach to Model Documentation
• Charting the Sociotechnical Gap in Explainable AI: A Framework to Address the Gap in XAI
• Data Minimization for GDPR Compliance in Machine Learning Models
• Towards Unifying Feature Attribution and Counterfactual Explanations: Different Means to the Same End
• Explainable Machine Learning for Public Policy: Use Cases, Gaps, and Research Directions
• A Critical Survey on Fairness Benefits of Explainable AI
• Directive Explanations for Monitoring the Risk of Diabetes Onset: Introducing Directive Data-Centric Explanations and Combinations to Support What-If Explorations
• A Framework and Benchmarking Study for Counterfactual Generating Methods on Tabular Data
• A domain-specific language for describing machine learning datasets
• Explainability Auditing for Intelligent Systems: A Rationale for Multi-Disciplinary Perspectives
• Interpretable Representations in Explainable AI: From Theory to Practice
• What Does Evaluation of Explainable Artificial Intelligence Actually Tell Us? A Case for Compositional and Contextual Validation of XAI Building Blocks
• FAT Forensics: A Python Toolbox for Algorithmic Fairness, Accountability and Transparency
• Towards a Comprehensive Human-Centred Evaluation Framework for Explainable AI
• SoK: Modeling Explainability in Security Analytics for Interpretability, Trustworthiness, and Usability
• Show Me the Work: Fact-Checkers' Requirements for Explainable Automated Fact-Checking
• Machine Learning Data Practices through a Data Curation Lens: An Evaluation Framework
• Towards Balancing Preference and Performance through Adaptive Personalized Explainability
• Comprehension Is a Double-Edged Sword: Over-Interpreting Unspecified Information in Intelligible Machine Learning Explanations
• LIMEtree: Consistent and Faithful Surrogate Explanations of Multiple Classes
• Classification Metrics for Image Explanations: Towards Building Reliable XAI-Evaluations
• Model Cards Revisited: Bridging the Gap Between Theory and Practice for Ethical AI Requirements
• Enhancing Counterfactual Explanation Search with Diffusion Distance and Directional Coherence
• ContextualSHAP : Enhancing SHAP Explanations Through Contextual Language Generation
• Navigating Explanatory Multiverse Through Counterfactual Path Geometry
• Towards a multi-stakeholder value-based assessment framework for algorithmic systems
• Hidden Conflicts in Neural Networks and Their Implications for Explainability
• Complex Sequential Data Analysis: A Systematic Literature Review of Existing Algorithms
• Evaluation of AI Ethics Tools in Language Models: A Developers' Perspective Case Study