Reference-Specific Unlearning Metrics Can Hide the Truth: A Reality Check

• URL: http://arxiv.org/abs/2510.12981v1
• Date: Tue, 14 Oct 2025 20:50:30 GMT
• Title: Reference-Specific Unlearning Metrics Can Hide the Truth: A Reality Check
• Authors: Sungjun Cho, Dasol Hwang, Frederic Sala, Sangheum Hwang, Kyunghyun Cho, Sungmin Cha,
• Abstract summary: We propose Functional Alignment for Distributional Equivalence (FADE), a novel metric that measures distributional similarity between unlearned and reference models.<n>We show that FADE captures functional alignment across the entire output distribution, providing a principled assessment of genuine unlearning.<n>These findings expose fundamental gaps in current evaluation practices and demonstrate that FADE provides a more robust foundation for developing and assessing truly effective unlearning methods.
• Score: 60.77691669644931
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Current unlearning metrics for generative models evaluate success based on reference responses or classifier outputs rather than assessing the core objective: whether the unlearned model behaves indistinguishably from a model that never saw the unwanted data. This reference-specific approach creates systematic blind spots, allowing models to appear successful while retaining unwanted knowledge accessible through alternative prompts or attacks. We address these limitations by proposing Functional Alignment for Distributional Equivalence (FADE), a novel metric that measures distributional similarity between unlearned and reference models by comparing bidirectional likelihood assignments over generated samples. Unlike existing approaches that rely on predetermined references, FADE captures functional alignment across the entire output distribution, providing a principled assessment of genuine unlearning. Our experiments on the TOFU benchmark for LLM unlearning and the UnlearnCanvas benchmark for text-to-image diffusion model unlearning reveal that methods achieving near-optimal scores on traditional metrics fail to achieve distributional equivalence, with many becoming more distant from the gold standard than before unlearning. These findings expose fundamental gaps in current evaluation practices and demonstrate that FADE provides a more robust foundation for developing and assessing truly effective unlearning methods.

Related papers

• Toward Reliable Machine Unlearning: Theory, Algorithms, and Evaluation [1.7767466724342065]
  We introduce Adrial Machine UNlearning (AMUN), which surpasses prior state-of-the-art methods for image classification based on SOTA MIA scores.<n>We show that existing methods fail in replicating a retrained model's behavior by introducing a nearest-neighbor membership inference attack (MIA-NN)<n>We then propose a fine-tuning objective that mitigates this leakage by approximating, for forget-class inputs, the distribution over remaining classes that a model retrained from scratch would produce.
  arXiv  Detail & Related papers  (2025-12-07T20:57:25Z)
• A Probabilistic Perspective on Unlearning and Alignment for Large Language Models [48.96686419141881]
  We introduce the first formal probabilistic evaluation framework for Large Language Models (LLMs)<n> Namely, we propose novel metrics with high probability guarantees concerning the output distribution of a model.<n>Our metrics are application-independent and allow practitioners to make more reliable estimates about model capabilities before deployment.
  arXiv  Detail & Related papers  (2024-10-04T15:44:23Z)
• Unconditional Truthfulness: Learning Unconditional Uncertainty of Large Language Models [104.55763564037831]
  We train a regression model that leverages attention maps, probabilities on the current generation step, and recurrently computed uncertainty scores from previously generated tokens.<n>Our evaluation shows that the proposed method is highly effective for selective generation, achieving substantial improvements over rivaling unsupervised and supervised approaches.
  arXiv  Detail & Related papers  (2024-08-20T09:42:26Z)
• DECIDER: Leveraging Foundation Model Priors for Improved Model Failure Detection and Explanation [18.77296551727931]
  We propose DECIDER, a novel approach that leverages priors from large language models (LLMs) and vision-language models (VLMs) to detect failures in image models. DECIDER consistently achieves state-of-the-art failure detection performance, significantly outperforming baselines in terms of the overall Matthews correlation coefficient.
  arXiv  Detail & Related papers  (2024-08-01T07:08:11Z)
• Towards Effective Evaluations and Comparisons for LLM Unlearning Methods [97.2995389188179]
  This paper seeks to refine the evaluation of machine unlearning for large language models.<n>It addresses two key challenges -- the robustness of evaluation metrics and the trade-offs between competing goals.
  arXiv  Detail & Related papers  (2024-06-13T14:41:00Z)
• Cycles of Thought: Measuring LLM Confidence through Stable Explanations [53.15438489398938]
  Large language models (LLMs) can reach and even surpass human-level accuracy on a variety of benchmarks, but their overconfidence in incorrect responses is still a well-documented failure mode. We propose a framework for measuring an LLM's uncertainty with respect to the distribution of generated explanations for an answer.
  arXiv  Detail & Related papers  (2024-06-05T16:35:30Z)
• Model Will Tell: Training Membership Inference for Diffusion Models [15.16244745642374]
  Training Membership Inference (TMI) task aims to determine whether a specific sample has been used in the training process of a target model. In this paper, we explore a novel perspective for the TMI task by leveraging the intrinsic generative priors within the diffusion model.
  arXiv  Detail & Related papers  (2024-03-13T12:52:37Z)
• Robust Fine-Tuning of Vision-Language Models for Domain Generalization [6.7181844004432385]
  Foundation models have impressive zero-shot inference capabilities and robustness under distribution shifts. We present a new recipe for few-shot fine-tuning of the popular vision-language foundation model CLIP. Our experimentation demonstrates that, while zero-shot CLIP fails to match performance of trained vision models on more complex benchmarks, few-shot CLIP fine-tuning outperforms its vision-only counterparts.
  arXiv  Detail & Related papers  (2023-11-03T20:50:40Z)
• A Call to Reflect on Evaluation Practices for Failure Detection in Image Classification [0.491574468325115]
  We present a large-scale empirical study for the first time enabling benchmarking confidence scoring functions. The revelation of a simple softmax response baseline as the overall best performing method underlines the drastic shortcomings of current evaluation.
  arXiv  Detail & Related papers  (2022-11-28T12:25:27Z)
• Learning the Truth From Only One Side of the Story [58.65439277460011]
  We focus on generalized linear models and show that without adjusting for this sampling bias, the model may converge suboptimally or even fail to converge to the optimal solution. We propose an adaptive approach that comes with theoretical guarantees and show that it outperforms several existing methods empirically.
  arXiv  Detail & Related papers  (2020-06-08T18:20:28Z)
• Meta-Learned Confidence for Few-shot Learning [60.6086305523402]
  A popular transductive inference technique for few-shot metric-based approaches, is to update the prototype of each class with the mean of the most confident query examples. We propose to meta-learn the confidence for each query sample, to assign optimal weights to unlabeled queries. We validate our few-shot learning model with meta-learned confidence on four benchmark datasets.
  arXiv  Detail & Related papers  (2020-02-27T10:22:17Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.