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 2401.08689cs.CVcs.LG
收藏

NODI:利用扩散噪声进行分布外检测

NODI: Out-Of-Distribution Detection with Noise from Diffusion

Chinese University of HongKong(香港中文大学)

Jingqiu Zhou, Aojun Zhou, Hongsheng Li

AI总结:

本文提出 NODI,将扩散模型整合全训练集信息的噪声向量用于 OOD 检测,并推导闭式噪声分数,在 OpenOOD 上显著提升性能且对多种图像编码器具有鲁棒性。

摘要:

分布外(OOD)检测是安全部署机器学习模型的关键组成部分。该问题已得到广泛研究,文献中发展出了大量方法。该问题通过计算 OOD 分数来解决,然而以往方法在计算 OOD 分数时对分布内数据集的利用有限。例如,OOD 分数仅利用一小部分分布内数据的信息来计算。此外,这些方法使用神经图像编码器对图像进行编码,而这些方法相对于不同训练方式和架构的图像编码器的鲁棒性很少得到检验。在本工作中,我们将扩散过程引入 OOD 任务。扩散模型将整个训练集的信息整合到预测的噪声向量中。此外,我们推导了噪声向量(稳定点)的闭式解。随后,噪声向量被转换为我们的 OOD 分数;我们在 OOD 基准 OpenOOD 上测试了深度模型预测的噪声向量和闭式噪声向量。在所有类型的图像编码器上,我们的方法均优于以往的 OOD 方法(表 main)。在基于 MAE 的图像编码器上,性能提升达到 3.5%。此外,我们通过应用不同类型的图像编码器研究了 OOD 方法的鲁棒性。当图像编码器从 ResNet 切换到 Vision Transformers 时,一些 OOD 方法无法很好地泛化,而我们的方法在所有图像编码器上都表现出良好的鲁棒性。

英文摘要:

Out-of-distribution (OOD) detection is a crucial part of deploying machine learning models safely. It has been extensively studied with a plethora of methods developed in the literature. This problem is tackled with an OOD score computation, however, previous methods compute the OOD scores with limited usage of the in-distribution dataset. For instance, the OOD scores are computed with information from a small portion of the in-distribution data. Furthermore, these methods encode images with a neural image encoder. The robustness of these methods is rarely checked with respect to image encoders of different training methods and architectures. In this work, we introduce the diffusion process into the OOD task. The diffusion model integrates information on the whole training set into the predicted noise vectors. What's more, we deduce a closed-form solution for the noise vector (stable point). Then the noise vector is converted into our OOD score, we test both the deep model predicted noise vector and the closed-form noise vector on the OOD benchmarks \cite{openood}. Our method outperforms previous OOD methods across all types of image encoders (Table. \ref{main}). A $3.5\%$ performance gain is achieved with the MAE-based image encoder. Moreover, we studied the robustness of OOD methods by applying different types of image encoders. Some OOD methods failed to generalize well when switching image encoders from ResNet to Vision Transformers, our method performs exhibits good robustness with all the image encoders.

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。

↑