用大模型生成假异常数据，让分类系统更准地识别异常输入。
Out-of-Distribution Detection using Synthetic Data Generation
- 用大模型自动生成高质量异常样本，无需真实异常数据
- 在多个任务中将误报率降至零，且不影响正常分类准确率
- 特别适合大模型训练和部署中的安全检测场景
区分分布内（InD）与分布外（OOD）输入对可靠部署分类系统至关重要。然而，真实OOD数据通常难以获取或根本不存在，给准确的OOD检测带来重大挑战。本文提出一种方法，利用大语言模型（LLM）的生成能力，创建高质量的合成OOD代理样本，从而摆脱对外部OOD数据源的依赖。我们在经典文本分类任务（如毒性检测、情感分析）以及大模型开发与部署中的分类任务（如强化学习人类反馈的奖励模型训练、生成内容不一致性检测）上评估该方法的有效性。在九组InD-OOD数据集对及多种模型规模下进行的大量实验表明，该方法显著降低了误报率（部分情况下实现零误报），同时保持了对分布内任务的高准确率，相比基线方法有明显提升。
原文摘要 · Abstract (English)
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.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。