Interpreting Black Box Predictions using Fisher Kernels

TL;DR

This paper introduces a novel method using Fisher kernels and Sequential Bayesian Quadrature to interpret black box models by identifying training examples most responsible for specific predictions, applicable to various interpretability tasks.

Contribution

It presents a new approach that combines Fisher kernels with SBQ for principled, scalable interpretation of model predictions, extending influence functions and providing theoretical guarantees.

Findings

Effective in identifying influential training examples

Applicable to data cleaning and correction tasks

Provides theoretical convergence bounds for the method

Abstract

Research in both machine learning and psychology suggests that salient examples can help humans to interpret learning models. To this end, we take a novel look at black box interpretation of test predictions in terms of training examples. Our goal is to ask `which training examples are most responsible for a given set of predictions'? To answer this question, we make use of Fisher kernels as the defining feature embedding of each data point, combined with Sequential Bayesian Quadrature (SBQ) for efficient selection of examples. In contrast to prior work, our method is able to seamlessly handle any sized subset of test predictions in a principled way. We theoretically analyze our approach, providing novel convergence bounds for SBQ over discrete candidate atoms. Our approach recovers the application of influence functions for interpretability as a special case yielding novel insights from this connection. We also present applications of the proposed approach to three use cases: cleaning training data, fixing mislabeled examples and data summarization.