遺忘比較器：一個用於機器遺忘方法比較評估的視覺分析系統

Unlearning Comparator: A Visual Analytics System for Comparative Evaluation of Machine Unlearning Methods

August 18, 2025

作者: Jaeung Lee, Suhyeon Yu, Yurim Jang, Simon S. Woo, Jaemin Jo

cs.AI

摘要

機器遺忘（Machine Unlearning, MU）旨在從已訓練的模型中移除特定訓練數據，使這些被移除的數據不再影響模型的行為，從而履行數據隱私法規下的「被遺忘權」義務。然而，我們觀察到，在這個迅速興起的領域中，研究人員在分析和理解不同MU方法的行為時面臨挑戰，尤其是在MU的三個基本原則：準確性、效率和隱私方面。因此，他們往往依賴於聚合指標和臨時評估，這使得準確評估方法之間的權衡變得困難。為填補這一空白，我們引入了一個視覺分析系統——遺忘比較器（Unlearning Comparator），旨在促進MU方法的系統性評估。我們的系統支持評估過程中的兩項重要任務：模型比較和攻擊模擬。首先，它允許用戶在類別、實例和層次上比較兩個模型的行為，例如由某種方法生成的模型與重新訓練的基準模型，以更好地理解遺忘操作後的變化。其次，我們的系統模擬成員推斷攻擊（Membership Inference Attacks, MIAs）來評估方法的隱私性，其中攻擊者試圖確定特定數據樣本是否屬於原始訓練集。我們通過一個案例研究來評估我們的系統，視覺化分析主流的MU方法，並展示它不僅幫助用戶理解模型行為，還能提供改進MU方法的洞察。

English

Machine Unlearning (MU) aims to remove target training data from a trained model so that the removed data no longer influences the model's behavior, fulfilling "right to be forgotten" obligations under data privacy laws. Yet, we observe that researchers in this rapidly emerging field face challenges in analyzing and understanding the behavior of different MU methods, especially in terms of three fundamental principles in MU: accuracy, efficiency, and privacy. Consequently, they often rely on aggregate metrics and ad-hoc evaluations, making it difficult to accurately assess the trade-offs between methods. To fill this gap, we introduce a visual analytics system, Unlearning Comparator, designed to facilitate the systematic evaluation of MU methods. Our system supports two important tasks in the evaluation process: model comparison and attack simulation. First, it allows the user to compare the behaviors of two models, such as a model generated by a certain method and a retrained baseline, at class-, instance-, and layer-levels to better understand the changes made after unlearning. Second, our system simulates membership inference attacks (MIAs) to evaluate the privacy of a method, where an attacker attempts to determine whether specific data samples were part of the original training set. We evaluate our system through a case study visually analyzing prominent MU methods and demonstrate that it helps the user not only understand model behaviors but also gain insights that can inform the improvement of MU methods.