arXiv：cs.AI· Andrada Gobeaja, Ionut Hodoroaga, Marius Leordeanu, Elena Burceanu·· 1 天前
idSCD：用语义相关描述符识别模型训练数据集
idSCD: Identifying Training Datasets through Semantic Correlation Descriptors
AI 导读
研究者提出白盒语义指纹方法 idSCD，基于语义相关描述符（SCD）刻画模型学到的语义相关结构，从而判断某数据集是否属于模型的训练数据混合。在自然语言推理、情感分类和医疗文本分类三组实验中，基于该分数的分类器平均表现最优且标准差最低，超过黑盒基线 RMIA、Attack-P、LiRA 及白盒基线 SIF。当数据集组呈现明显语义特性时，ROC-AUC 的最大相对提升超过 60%。
正文
Abstract:Can a dataset be recognized from the spurious correlations it induces during training? We argue that datasets leave dataset-specific traces in a model's learned semantic correlation structure: incidental regularities that are predictive within a dataset, but not causal for the underlying task, can be internalized during training. We use this insight to study dataset-level membership inference, moving beyond existing methods that rely on behavioral or distributional evidence such as confidence scores, losses, margins, generated samples, or query responses. We introduce a white-box semantic fingerprinting approach based on semantic correlation descriptors (SCDs), which capture the semantic correlation structure learned by a model and make it comparable across dataset mixtures. In a controlled leave-one-dataset-out diagnostic, SCDs recover dataset-specific changes and perfectly separate matching from non-matching dataset pairs. We then propose a practical SCD-based membership score that tests whether a target dataset is part of a model's training mixture using only the model's SCD and the target dataset's standalone SCD, without requiring leave-one-dataset-out models. Across three diverse experimental settings, with dataset groups for natural language inference, emotion classification, and medical text classification, we test both the advantages and limitations of SCD-based membership inference with different degrees of semantic separation and keyword support between dataset splits. On average, the classifier based on this score achieves the highest performance and the lowest std, outperforming black-box baselines RMIA, Attack-P, and LiRA, as well as the white-box SIF baseline. These results show that dataset membership can be traced through internal semantic correlations, with the largest relative gain exceeding 60% in ROC-AUC when dataset groups expose distinct semantic particularities.
| Comments: | 20 pages, 6 figures |
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2605.30462 [cs.LG] |
| (or arXiv:2605.30462v2 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2605.30462 arXiv-issued DOI via DataCite |
Submission history
From: Andrada Gobeaja [view email]
[v1]
Thu, 28 May 2026 18:38:15 UTC (362 KB)
[v2]
Mon, 12 Oct 2026 11:50:21 UTC (381 KB)
来源：arXiv：cs.AI · arxiv.org