Logic-Enabled Explanations for Machine Learning Models

April 23, 2019

Joao Marques Silva


Logic-Enabled Explanations for Machine Learning Models

Time:   10:45am
Location:   Meeting room 202 (Hill View), level 2

The practical successes of Machine Learning (ML) in different settings motivates the ability of computing small explanations for predictions made. Small explanations are generally accepted as easier for human decision makers to understand. Existing work on computing explanations is based on heuristic approaches, providing no guarantees of quality, in terms of how close such solutions are from cardinality- or subset-minimal explanations.

This talk describes a novel constraint-agnostic approach for computing explanations for any Machine Learning (ML) model. The proposed solution exploits abductive reasoning, and imposes the requirement that the ML model be represented as sets of constraints using some target constraint reasoning system for which the decision problem can be answered with some oracle. The talk also compares the new logic-enabled approach for computing explanations with existing heuristic approaches on well-known datasets.