arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2305.17797cs.CV
收藏

T2FNorm:用于OOD检测的极其简单的缩放训练时特征归一化

T2FNorm: Extremely Simple Scaled Train-time Feature Normalization for OOD Detection

NepAl Applied Mathematics and Informatics Institute (NAAMII)(尼泊尔应用数学与信息学研究所(NAAMII));University College London(伦敦大学学院);West Virginia University(西弗吉尼亚大学);University of Aberdeen(阿伯丁大学)

Sudarshan Regmi, Bibek Panthi, Sakar Dotel, Prashnna K. Gyawali, Danail Stoyanov, Binod Bhattarai

AI总结:

本文提出T2FNorm,在训练时将特征缩放归一化到超球空间、评分时使用原始空间,从而在不降低ID准确率的前提下增强多种事后OOD检测。

摘要:

神经网络因作为过度自信的预测器而广为人知,这给其在现实应用中的安全部署带来了重大挑战。尽管特征归一化已在深度学习文献中获得大量关注,但当前用于分布外(Out-of-Distribution,OOD)检测的训练时正则化方法尚未充分挖掘这一潜力。事实上,在神经网络中简单地加入特征归一化并不能保证OOD检测性能获得实质性提升。在本工作中,我们提出T2FNorm,这是一种在训练期间将特征变换到超球空间、同时在OOD评分时采用非变换空间的新颖方法。该方法在不损害分布内(in-distribution,ID)模型准确率的情况下,显著增强了OOD检测能力。我们的研究表明,所提出的技术大幅降低了所有样本的特征范数,并且对分布外样本的降低程度更大,从而解决了神经网络中普遍存在的过度自信问题。该方法还显著改进了多种事后OOD检测方法。

英文摘要:

Neural networks are notorious for being overconfident predictors, posing a significant challenge to their safe deployment in real-world applications. While feature normalization has garnered considerable attention within the deep learning literature, current train-time regularization methods for Out-of-Distribution(OOD) detection are yet to fully exploit this potential. Indeed, the naive incorporation of feature normalization within neural networks does not guarantee substantial improvement in OOD detection performance. In this work, we introduce T2FNorm, a novel approach to transforming features to hyperspherical space during training, while employing non-transformed space for OOD-scoring purposes. This method yields a surprising enhancement in OOD detection capabilities without compromising model accuracy in in-distribution(ID). Our investigation demonstrates that the proposed technique substantially diminishes the norm of the features of all samples, more so in the case of out-of-distribution samples, thereby addressing the prevalent concern of overconfidence in neural networks. The proposed method also significantly improves various post-hoc OOD detection methods.

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。

↑