Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Machine Unlearning in the Era of Quantum Machine Learning: An Empirical Study

Carla Crivoi, Radu Tudor Ionescu

TL;DR

The paper provides the first empirical study of machine unlearning in hybrid quantum-classical neural networks, adapting a broad suite of MU methods to variational quantum circuits and introducing two quantum-tailored strategies (LCA and ADV-UNIFORM). Through Iris, MNIST, and Fashion-MNIST experiments, it shows that shallow VQCs naturally limit memorization, while deeper hybrids require structured interventions to balance utility, forgetting quality, and similarity to retraining, with EU-k1, LCA, and Certified Unlearning emerging as robust baselines. The work highlights the need for quantum-aware unlearning algorithms and theoretical guarantees, and suggests directions for hardware-backed evaluations and quantum-native forgetting objectives. Overall, it establishes baseline empirical insights into quantum machine unlearning and informs future development of robust, privacy-preserving quantum learning systems.

Abstract

We present the first comprehensive empirical study of machine unlearning (MU) in hybrid quantum-classical neural networks. While MU has been extensively explored in classical deep learning, its behavior within variational quantum circuits (VQCs) and quantum-augmented architectures remains largely unexplored. First, we adapt a broad suite of unlearning methods to quantum settings, including gradient-based, distillation-based, regularization-based and certified techniques. Second, we introduce two new unlearning strategies tailored to hybrid models. Experiments across Iris, MNIST, and Fashion-MNIST, under both subset removal and full-class deletion, reveal that quantum models can support effective unlearning, but outcomes depend strongly on circuit depth, entanglement structure, and task complexity. Shallow VQCs display high intrinsic stability with minimal memorization, whereas deeper hybrid models exhibit stronger trade-offs between utility, forgetting strength, and alignment with retrain oracle. We find that certain methods, e.g. EU-k, LCA, and Certified Unlearning, consistently provide the best balance across metrics. These findings establish baseline empirical insights into quantum machine unlearning and highlight the need for quantum-aware algorithms and theoretical guarantees, as quantum machine learning systems continue to expand in scale and capability. We publicly release our code at: https://github.com/CrivoiCarla/HQML.

Machine Unlearning in the Era of Quantum Machine Learning: An Empirical Study

TL;DR

The paper provides the first empirical study of machine unlearning in hybrid quantum-classical neural networks, adapting a broad suite of MU methods to variational quantum circuits and introducing two quantum-tailored strategies (LCA and ADV-UNIFORM). Through Iris, MNIST, and Fashion-MNIST experiments, it shows that shallow VQCs naturally limit memorization, while deeper hybrids require structured interventions to balance utility, forgetting quality, and similarity to retraining, with EU-k1, LCA, and Certified Unlearning emerging as robust baselines. The work highlights the need for quantum-aware unlearning algorithms and theoretical guarantees, and suggests directions for hardware-backed evaluations and quantum-native forgetting objectives. Overall, it establishes baseline empirical insights into quantum machine unlearning and informs future development of robust, privacy-preserving quantum learning systems.

Abstract

We present the first comprehensive empirical study of machine unlearning (MU) in hybrid quantum-classical neural networks. While MU has been extensively explored in classical deep learning, its behavior within variational quantum circuits (VQCs) and quantum-augmented architectures remains largely unexplored. First, we adapt a broad suite of unlearning methods to quantum settings, including gradient-based, distillation-based, regularization-based and certified techniques. Second, we introduce two new unlearning strategies tailored to hybrid models. Experiments across Iris, MNIST, and Fashion-MNIST, under both subset removal and full-class deletion, reveal that quantum models can support effective unlearning, but outcomes depend strongly on circuit depth, entanglement structure, and task complexity. Shallow VQCs display high intrinsic stability with minimal memorization, whereas deeper hybrid models exhibit stronger trade-offs between utility, forgetting strength, and alignment with retrain oracle. We find that certain methods, e.g. EU-k, LCA, and Certified Unlearning, consistently provide the best balance across metrics. These findings establish baseline empirical insights into quantum machine unlearning and highlight the need for quantum-aware algorithms and theoretical guarantees, as quantum machine learning systems continue to expand in scale and capability. We publicly release our code at: https://github.com/CrivoiCarla/HQML.
Paper Structure (11 sections, 6 equations, 1 figure, 6 tables)

This paper contains 11 sections, 6 equations, 1 figure, 6 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Problem Formulation
  4. Proposed Methods
  5. Experimental Setup
  6. Results and Analysis
  7. Results on Iris Dataset
  8. MNIST Dataset
  9. Fashion-MNIST Dataset
  10. Discussion
  11. Conclusion and Future Work

Figures (1)

  • Figure 1: Generic machine unlearning pipeline for hybrid quantum-classical neural models. In the first stage, the model is trained on a dataset, e.g. Fashion-MNIST. In the second stage, some samples / classes (forget set) have to be deleted via unlearning. In the evaluation stage, the model has to preserve its performance level on samples / classes (retain set) that should not have been deleted. Best viewed in color.