用扩散模型生成更逼真的异常数据，提升模型对未知输入的识别能力
Non-Linear Outlier Synthesis for Out-of-Distribution Detection
- 在扩散模型嵌入空间直接生成异常样本，避免多模型拼接缺陷
- 在ImageNet100和CIFAR100上达到新最优性能，检测准确率显著提升
- 适合关注鲁棒性与异常检测的机器学习研究者使用
监督分类器在面对意外输入时可靠性严重受限，促使学术界广泛关注分布外（OOD）检测。近期基于合成异常样本训练的OOD检测器，特别是利用大扩散模型生成的异常样本，已展现出定义稳健决策边界的潜力。本文提出NCIS方法，在扩散模型的嵌入空间中直接生成合成异常样本，而非依赖多个分离模型；同时引入条件体积保持网络，对类别条件流形进行建模，以更精准刻画训练数据分布。实验表明，这些改进在标准ImageNet100和CIFAR100基准上取得了新的最先进结果，并揭示了数据预处理及其他关键设计选择的重要性。代码已公开于https://github.com/LarsDoorenbos/NCIS。
原文摘要 · Abstract (English)
The reliability of supervised classifiers is severely hampered by their limitations in dealing with unexpected inputs, leading to great interest in out-of-distribution (OOD) detection. Recently, OOD detectors trained on synthetic outliers, especially those generated by large diffusion models, have shown promising results in defining robust OOD decision boundaries. Building on this progress, we present NCIS, which enhances the quality of synthetic outliers by operating directly in the diffusion's model embedding space rather than combining disjoint models as in previous work and by modeling class-conditional manifolds with a conditional volume-preserving network for more expressive characterization of the training distribution. We demonstrate that these improvements yield new state-of-the-art OOD detection results on standard ImageNet100 and CIFAR100 benchmarks and provide insights into the importance of data pre-processing and other key design choices. We make our code available at \url{https://github.com/LarsDoorenbos/NCIS}.
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。