Towards Reliable Testing of Machine Unlearning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Mazhar, Anna, Galhotra, Sainyam
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917417643409408
author Mazhar, Anna
Galhotra, Sainyam
author_facet Mazhar, Anna
Galhotra, Sainyam
contents Machine learning components are now central to AI-infused software systems, from recommendations and code assistants to clinical decision support. As regulations and governance frameworks increasingly require deleting sensitive data from deployed models, machine unlearning is emerging as a practical alternative to full retraining. However, unlearning introduces a software quality-assurance challenge: under realistic deployment constraints and imperfect oracles, how can we test that a model no longer relies on targeted information? This paper frames unlearning testing as a first-class software engineering problem. We argue that practical unlearning tests must provide (i) thorough coverage over proxy and mediated influence pathways, (ii) debuggable diagnostics that localize where leakage persists, (iii) cost-effective regression-style execution under query budgets, and (iv) black-box applicability for API-deployed models. We outline a causal, pathway-centric perspective, causal fuzzing, that generates budgeted interventions to estimate residual direct and indirect effects and produce actionable "leakage reports". Proof-of-concept results illustrate that standard attribution checks can miss residual influence due to proxy pathways, cancellation effects, and subgroup masking, motivating causal testing as a promising direction for unlearning testing.
format Preprint
id arxiv_https___arxiv_org_abs_2604_16536
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Towards Reliable Testing of Machine Unlearning
Mazhar, Anna
Galhotra, Sainyam
Machine Learning
Artificial Intelligence
Machine learning components are now central to AI-infused software systems, from recommendations and code assistants to clinical decision support. As regulations and governance frameworks increasingly require deleting sensitive data from deployed models, machine unlearning is emerging as a practical alternative to full retraining. However, unlearning introduces a software quality-assurance challenge: under realistic deployment constraints and imperfect oracles, how can we test that a model no longer relies on targeted information? This paper frames unlearning testing as a first-class software engineering problem. We argue that practical unlearning tests must provide (i) thorough coverage over proxy and mediated influence pathways, (ii) debuggable diagnostics that localize where leakage persists, (iii) cost-effective regression-style execution under query budgets, and (iv) black-box applicability for API-deployed models. We outline a causal, pathway-centric perspective, causal fuzzing, that generates budgeted interventions to estimate residual direct and indirect effects and produce actionable "leakage reports". Proof-of-concept results illustrate that standard attribution checks can miss residual influence due to proxy pathways, cancellation effects, and subgroup masking, motivating causal testing as a promising direction for unlearning testing.
title Towards Reliable Testing of Machine Unlearning
topic Machine Learning
Artificial Intelligence
url https://arxiv.org/abs/2604.16536