A Generic Approach for Reproducible Model Distillation

TL;DR

This paper introduces a universal method for stable model distillation that ensures reproducibility across various interpretable models by using a statistical testing framework based on the central limit theorem.

Contribution

It develops a generic, statistically grounded approach for stable model distillation applicable to multiple model types, improving reproducibility and reliability.

Findings

Successfully applied to decision trees, rule lists, and symbolic regression.

Demonstrated effectiveness on Mammographic Mass and Breast Cancer datasets.

Provided theoretical analysis and publicly available code.

Abstract

Model distillation has been a popular method for producing interpretable machine learning. It uses an interpretable "student" model to mimic the predictions made by the black box "teacher" model. However, when the student model is sensitive to the variability of the data sets used for training even when keeping the teacher fixed, the corresponded interpretation is not reliable. Existing strategies stabilize model distillation by checking whether a large enough corpus of pseudo-data is generated to reliably reproduce student models, but methods to do so have so far been developed for a specific student model. In this paper, we develop a generic approach for stable model distillation based on central limit theorem for the average loss. We start with a collection of candidate student models and search for candidates that reasonably agree with the teacher. Then we construct a multiple testing framework to select a corpus size such that the consistent student model would be selected under different pseudo samples. We demonstrate the application of our proposed approach on three commonly used intelligible models: decision trees, falling rule lists and symbolic regression. Finally, we conduct simulation experiments on Mammographic Mass and Breast Cancer datasets and illustrate the testing procedure throughout a theoretical analysis with Markov process. The code is publicly available at https://github.com/yunzhe-zhou/GenericDistillation.