Research map / Learning theory and algorithms / Prediction and decision losses
Abduction-Based Explanations for Machine Learning Models
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Use a MILP oracle instead of an SMT oracle | Abduction-based explanation procedure · UCI and PennML datasets | explanation computation time | Mixed |
| Compute cardinality-minimal rather than subset-minimal explanations | Abduction-based explanation procedure · UCI and PennML datasets | explanation size: 21.6% to 85.3%; average 52.6% (cardinality-minimal), versus 28.5% to 86.7%; mean 60.5% (subset-minimal) | Helped |
| Compute cardinality-minimal rather than subset-minimal explanations | Abduction-based explanation procedure · UCI and PennML datasets | computation time and completion: 1800 seconds | Hurt |
| Use MILP rather than SMT for MNIST explanation queries | Subset-minimal explanation procedure · MNIST digits | successful explanations within 1 hour: 1 hour | Helped |
| Use heuristic pixel ordering to prioritize removing pixels far from the image center | Subset-minimal explanations for MNIST images · MNIST digits | human interpretability | Helped |
| Use the neural-network explanation approach instead of BDD-based Bayesian-network explanations | Voting Records sample explanation · Voting Records sample | explanation size: 9; 3 to 5 | Helped |
Closest papers by meaning
- An Improved Algorithm for Learning to Perform Exception-Tolerant Abduction (AAAI 2017 · similarity 0.89)
- Axiomatic Aggregations of Abductive Explanations (AAAI 2024 · similarity 0.85)
- Delivering Inflated Explanations (AAAI 2024 · similarity 0.85)
- Analysis for Abductive Learning and Neural-Symbolic Reasoning Shortcuts (ICML 2024 · similarity 0.84)
- Decomposing Constraint Networks for Calculating c-Representations (AAAI 2024 · similarity 0.84)
- Auditing Local Explanations is Hard (NeurIPS 2024 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.