Prototype selection for interpretable classification

Abstract

Prototype methods seek a minimal subset of samples that can serve as a distillation or condensed view of a data set. As the size of modern data sets grows, being able to present a domain specialist with a short list of "representative" samples chosen from the data set is of increasing interpretative value. While much recent statistical research has been focused on producing sparse-in-the-variables methods, this paper aims at achieving sparsity in the samples. We discuss a method for selecting prototypes in the classification setting (in which the samples fall into known discrete categories). Our method of focus is derived from three basic properties that we believe a good prototype set should satisfy. This intuition is translated into a set cover optimization problem, which we solve approximately using standard approaches. While prototype selection is usually viewed as purely a means toward building an efficient classifier, in this paper we emphasize the inherent value of having a set of prototypical elements. That said, by using the nearest-neighbor rule on the set of prototypes, we can of course discuss our method as a classifier as well.

Published in at http://dx.doi.org/10.1214/11-AOAS495 the Annals of Applied Statistics (http://www.imstat.org/aoas/) by the Institute of Mathematical Statistics (http://www.imstat.org). arXiv admin note: text overlap with arXiv:0908.2284

Cited by in corpus (43)

• A Survey on the Explainability of Supervised Machine Learning
• Questioning the AI: Informing Design Practices for Explainable AI User Experiences
• This Looks Like That: Deep Learning for Interpretable Image Recognition
• Designing Interpretable ML System to Enhance Trust in Healthcare: A Systematic Review to Proposed Responsible Clinician-AI-Collaboration Framework
• Explainable Artificial Intelligence: a Systematic Review
• Opening the Black-Box: A Systematic Review on Explainable AI in Remote Sensing
• On Interpretability of Artificial Neural Networks: A Survey
• This Looks Like That, Because ... Explaining Prototypes for Interpretable Image Recognition
• Tree Space Prototypes: Another Look at Making Tree Ensembles Interpretable
• Considerations When Learning Additive Explanations for Black-Box Models
• A Survey on Neural Network Interpretability
• Methods and Models for Interpretable Linear Classification
• Supersparse Linear Integer Models for Interpretable Classification
• 10,000+ Times Accelerated Robust Subset Selection (ARSS)
• Distribution Density, Tails, and Outliers in Machine Learning: Metrics and Applications
• Evaluation of Similarity-based Explanations
• Explainable Active Learning (XAL): An Empirical Study of How Local Explanations Impact Annotator Experience
• Machine-learning correction to density-functional crystal structure optimization
• ProtoAttend: Attention-Based Prototypical Learning
• Melody: Generating and Visualizing Machine Learning Model Summary to Understand Data and Classifiers Together
• Interpretable Medical Image Classification using Prototype Learning and Privileged Information
• SLEEPER: interpretable Sleep staging via Prototypes from Expert Rules
• RelatIF: Identifying Explanatory Training Examples via Relative Influence
• Comparative Document Summarisation via Classification
• Optimal 1-NN Prototypes for Pathological Geometries
• DIVINE: Diverse Influential Training Points for Data Visualization and Model Refinement
• Learning a sparse database for patch-based medical image segmentation
• Safety design concepts for statistical machine learning components toward accordance with functional safety standards
• TREX: Tree-Ensemble Representer-Point Explanations
• Robust and Scalable Column/Row Sampling from Corrupted Big Data
• Less is more: Selecting informative and diverse subsets with balancing constraints
• (Decision and regression) tree ensemble based kernels for regression and classification
• Unifying Model Explainability and Robustness via Machine-Checkable Concepts
• Explainability-aided Domain Generalization for Image Classification
• A Tale Of Two Long Tails
• Explaining Predictions of Deep Neural Classifier via Activation Analysis
• Model-Agnostic Explanations using Minimal Forcing Subsets
• ProtoShotXAI: Using Prototypical Few-Shot Architecture for Explainable AI
• Inductive Granger Causal Modeling for Multivariate Time Series
• Exact and Consistent Interpretation of Piecewise Linear Models Hidden behind APIs: A Closed Form Solution
• A Framework for an Assessment of the Kernel-target Alignment in Tree Ensemble Kernel Learning
• Interpretable Multiple-Kernel Prototype Learning for Discriminative Representation and Feature Selection
• Enumeration of Distinct Support Vectors for Interactive Decision Making