ShapDBM: Exploring Decision Boundary Maps in Shapley Space
2026-03-23 • Human-Computer Interaction
Human-Computer InteractionMachine Learning
AI summaryⓘ
The authors studied a way to make visual maps that show how a machine learning program divides different categories. Usually, these maps can be messy because the data is complex and needs to be shrunk down from many dimensions. They suggested changing the data into a special form called Shapley space before shrinking it. This made the maps clearer and easier to understand than the usual methods. Their new approach keeps or improves quality while making decision areas look simpler.
Decision Boundary MapsDimensionality ReductionShapley SpaceMachine Learning ClassificationData VisualizationHigh Dimensional DataFeature TransformationShapley Values
Authors
Luke Watkin, Daniel Archambault, Alex Telea
Abstract
Decision Boundary Maps (DBMs) are an effective tool for visualising machine learning classification boundaries. Yet, DBM quality strongly depends on the dimensionality reduction (DR) technique and high dimensional space used for the data points. For complex ML datasets, DR can create many mixed classes which, in turn, yield DBMs that are hard to use. We propose a new technique to compute DBMs by transforming data space into Shapley space and computing DR on it. Compared to standard DBMs computed directly from data, our maps have similar or higher quality metric values and visibly more compact, easier to explore, decision zones.