arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2502.03323cs.CLcs.AIcs.LG
收藏

利用合成数据生成的分布外检测

Out-of-Distribution Detection using Synthetic Data Generation

IBM Research(IBM研究院);MIT-IBM Watson AI Lab(麻省理工学院-IBM沃森人工智能实验室)

Momin Abbas, Muneeza Azmat, Raya Horesh, Mikhail Yurochkin

AI总结:

针对OOD数据难以获取的问题,提出利用LLM生成合成OOD代理数据的检测方法,在多任务多数据集上显著降低假阳性率且保持分布内任务高精度。

摘要:

区分分布内与分布外(OOD)输入对分类系统的可靠部署至关重要。然而,OOD数据通常难以获取或收集困难,给准确的OOD检测带来了重大挑战。在本研究中,我们提出了一种利用大语言模型(LLMs)生成能力创建高质量合成OOD代理数据的方法,消除了对任何外部OOD数据源的依赖。我们在毒性检测、情感分类等经典文本分类任务,以及LLM开发与部署中的分类任务(如训练用于RLHF的奖励模型、检测对齐失效的生成内容)上研究了该方法的有效性。在9个InD-OOD数据集对和不同模型规模上开展的大量实验表明,我们的方法在保持分布内任务高精度的同时,大幅降低了假阳性率(部分场景下实现了完美的零假阳性),性能显著优于基线方法。

英文摘要:

Distinguishing in- and out-of-distribution (OOD) inputs is crucial for reliable deployment of classification systems. However, OOD data is typically unavailable or difficult to collect, posing a significant challenge for accurate OOD detection. In this work, we present a method that harnesses the generative capabilities of Large Language Models (LLMs) to create high-quality synthetic OOD proxies, eliminating the dependency on any external OOD data source. We study the efficacy of our method on classical text classification tasks such as toxicity detection and sentiment classification as well as classification tasks arising in LLM development and deployment, such as training a reward model for RLHF and detecting misaligned generations. Extensive experiments on nine InD-OOD dataset pairs and various model sizes show that our approach dramatically lowers false positive rates (achieving a perfect zero in some cases) while maintaining high accuracy on in-distribution tasks, outperforming baseline methods by a significant margin.

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。

↑