The explanation game
A formal framework for interpretable machine learning
Bibliographic Data
| ID | 10811925 |
|---|---|
| Authors | David S Watson (0000-0001-9632-2159, University of Oxford, corresponding author), Luciano Floridi (0000-0002-5444-2280, Turing Institute) |
| Year | 2021 |
| Volume | 198 |
| Issue | 10 |
| Pages | 9211-9242 |
| Publication date | 2021-10-01 |
| Peer Reviewed | Yes |
| Open Access | Yes |
| Type | ARTICLE |
| Venue | Synthese (JOURNAL) |
| Journal identifiers | ISSN: 0039-7857 • E-ISSN: 1573-0964 |
| Publisher | Springer Science and Business Media LLC (PUBLISHER) |
| DOI | 10.1007/s11229-020-02629-9 |
| OpenAlex | W3014632866 |
| Language | EN |
| Citations received | 15 |
| References cited | 87 |
We propose a formal framework for interpretable machine learning. Combining elements from statistical learning, causal interventionism, and decision theory, we design an idealised explanation game in which players collaborate to find the best explanation(s) for a given algorithmic prediction. Through an iterative procedure of questions and answers, the players establish a three-dimensional Pareto frontier that describes the optimal trade-offs between explanatory accuracy, simplicity, and relevance. Multiple rounds are played at different levels of abstraction, allowing the players to explore overlapping causal patterns of variable granularity and scope. We characterise the conditions under which such a game is almost surely guaranteed to converge on a (conditionally) optimal explanation surface in polynomial time, and highlight obstacles that will tend to prevent the players from advancing beyond certain explanatory thresholds. The game serves a descriptive and a normative function, establishing a conceptual space in which to analyse and compare existing proposals, as well as design new and improved solutions
Epistemology · Machine learning · Mathematical economics · Normative · Relevance (law) · Adversarial Robustness in Machine Learning · Artificial Intelligence · Bayesian Modeling and Causal Inference · Computer Science · Explainable Artificial Intelligence (XAI · Mathematics · Theoretical Computer Science
AI and Its New Winter
Philosophy of science at sea
The ethics of algorithms
Explicability of humanitarian AI
AI support for ethical decision-making around resuscitation
All you need is…. justification
Explainability, Public Reason, and Medical Artificial Intelligence
Tacit knowledge and a multi-method approach in Asset Management
Competing narratives in AI ethics
What is Interpretability
On the Philosophy of Unsupervised Learning
Trust and Trustworthiness in AI
Sources of Understanding in Supervised Machine Learning Models
The epistemological foundations of data science
Conceptual challenges for interpretable machine learning
Hunting Causes and Using Them
Causation, Prediction, and Search
The Philosophy of Information
Causality
Causation, Prediction, and Search
Predictive Policing
Idealization and the Aims of Science
A Unified Framework of Five Principles for AI in Society
Explanation in artificial intelligence
A general reinforcement learning algorithm that masters chess, shogi, and Go through self-play
Statistical Modeling
European Union Regulations on Algorithmic Decision Making and a “Right to Explanation”
Dermatologist-level classification of skin cancer with deep neural networks
No free lunch theorems for optimization
Explaining Explanations in AI
Discrimination in the Age of Algorithms
Causal diagrams for empirical research
Causation in biology
What is Justified Belief?
Stop explaining black box machine learning models for high stakes decisions and use interpretable models instead
The Method of Levels of Abstraction
Is Justified True Belief Knowledge?
Why a Right to Explanation of Automated Decision-Making Does Not Exist in the General Data Protection Regulation
Minds, brains, and programs
AI4People—An Ethical Framework for a Good AI Society
The Logic of Scientific Discovery
The Black Box Society
Justification and Knowledge
A Value for n-Person Games
Goldman and His Critics
Transparency in Algorithmic and Human Decision-Making
Against Modularity, the Causal Markov Condition, and Any Link Between the Two
Crowdsourced science
Realism, rhetoric, and reliability
Semantic information and the network theory of account
On the Logical Unsolvability of the Gettier Problem
No understanding without explanation
On the Explanatory Depth and Pragmatic Value of Coarse-Grained, Probabilistic, Causal Explanations
Inaugurating Understanding or Repackaging Explanation
Explanatory Depth
Theory of Games and Economic Behavior
Interventionism and Causal Exclusion
Modularity and the Causal Markov Condition
High-Level Explanation and the Interventionist’s ‘Variables Problem’
Contrastive Explanation and the Demons of Determinism
Is Understanding A Species Of Knowledge
The scientific image
Automating Inequality
The ethics of algorithms
| Unique citing works | 15 |
|---|---|
| Citations per year | 2,5 |
| Citation span | 2020 - 2026 (7) |
| Citation velocity | current |
| Highly cited | No |
| Citation types | Neutral: 12 |