RESTOR：机器遗忘中的知识恢复
RESTOR: Knowledge Recovery in Machine Unlearning
AI总结：
针对机器遗忘评估难题，提出RESTOR框架，通过同时评估遗忘目标知识与恢复原始知识状态，揭示现有算法仅重遗忘不重恢复等缺陷，并发现定位遗忘目标可提升性能。
摘要：
在网页规模语料库上训练的大型语言模型可能会记住包含错误信息、受版权保护的材料或私人或敏感信息的不良数据。最近，人们提出了几种机器遗忘算法，以消除这些数据点对训练模型的影响——即近似一个从一开始就从未在这些数据点上训练过的模型。然而，评估遗忘算法的有效性仍然是一个悬而未决的挑战。先前的工作依赖于启发式方法——例如验证模型不再能够重现针对移除的特定信息，同时保持对无关测试数据的准确性。这些方法不足以完全捕捉逆转数据点对训练模型影响的效果。在这项工作中，我们提出了用于机器遗忘评估的RESTOR框架，该框架通过评估模型遗忘这些数据点中引入的知识的能力，同时恢复模型若从未遇到这些数据点时应有的知识状态，来评估遗忘算法针对目标数据擦除的能力。RESTOR有助于揭示关于流行遗忘算法及其运作机制的若干新见解——例如，识别出某些算法仅强调遗忘但不恢复知识，以及定位遗忘目标可以增强遗忘性能。
英文摘要：
Large language models trained on web-scale corpora can memorize undesirable data containing misinformation, copyrighted material, or private or sensitive information. Recently, several machine unlearning algorithms have been proposed to eliminate the effect of such datapoints from trained models -- that is, to approximate a model that had never been trained on these datapoints in the first place. However, evaluating the effectiveness of unlearning algorithms remains an open challenge. Previous work has relied on heuristics -- such as verifying that the model can no longer reproduce the specific information targeted for removal while maintaining accuracy on unrelated test data. These approaches inadequately capture the complete effect of reversing the influence of datapoints on a trained model. In this work, we propose the RESTOR framework for machine unlearning evaluation, which assesses the ability of unlearning algorithms for targeted data erasure, by evaluating the ability of models to forget the knowledge introduced in these datapoints, while simultaneously recovering the model's knowledge state had it never encountered these datapoints. RESTOR helps uncover several novel insights about popular unlearning algorithms, and the mechanisms through which they operate -- for instance, identifying that some algorithms merely emphasize forgetting but not recovering knowledge, and that localizing unlearning targets can enhance unlearning performance.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
