%0 Journal Article %1 Taghikhah2024-rw %A Taghikhah, Masoud %A Kumar, Nishant %A Segvić, Sinisa %A Eslami, Abouzar %A Gumhold, Stefan %D 2024 %I Association for the Advancement of Artificial Intelligence (AAAI) %J Proc. Conf. AAAI Artif. Intell. %K Quantile-based detection likelihood maximum outlier training %N 19 %P 21610--21618 %T Quantile-based maximum likelihood training for outlier detection %V 38 %X Discriminative learning effectively predicts true object class for image classification. However, it often results in false positives for outliers, posing critical concerns in applications like autonomous driving and video surveillance systems. Previous attempts to address this challenge involved training image classifiers through contrastive learning using actual outlier data or synthesizing outliers for self-supervised learning. Furthermore, unsupervised generative modeling of inliers in pixel space has shown limited success for outlier detection. In this work, we introduce a quantile-based maximum likelihood objective for learning the inlier distribution to improve the outlier separation during inference. Our approach fits a normalizing flow to pre-trained discriminative features and detects the outliers according to the evaluated log-likelihood. The experimental evaluation demonstrates the effectiveness of our method as it surpasses the performance of the state-of-the-art unsupervised methods for outlier detection. The results are also competitive compared with a recent self-supervised approach for outlier detection. Our work allows to reduce dependency on well-sampled negative training data, which is especially important for domains like medical diagnostics or remote sensing.

@article{Taghikhah2024-rw, abstract = {Discriminative learning effectively predicts true object class for image classification. However, it often results in false positives for outliers, posing critical concerns in applications like autonomous driving and video surveillance systems. Previous attempts to address this challenge involved training image classifiers through contrastive learning using actual outlier data or synthesizing outliers for self-supervised learning. Furthermore, unsupervised generative modeling of inliers in pixel space has shown limited success for outlier detection. In this work, we introduce a quantile-based maximum likelihood objective for learning the inlier distribution to improve the outlier separation during inference. Our approach fits a normalizing flow to pre-trained discriminative features and detects the outliers according to the evaluated log-likelihood. The experimental evaluation demonstrates the effectiveness of our method as it surpasses the performance of the state-of-the-art unsupervised methods for outlier detection. The results are also competitive compared with a recent self-supervised approach for outlier detection. Our work allows to reduce dependency on well-sampled negative training data, which is especially important for domains like medical diagnostics or remote sensing.}, added-at = {2024-11-12T15:19:09.000+0100}, author = {Taghikhah, Masoud and Kumar, Nishant and {\v S}egvi{\'c}, Sini{\v s}a and Eslami, Abouzar and Gumhold, Stefan}, biburl = {https://puma.scadsai.uni-leipzig.de/bibtex/2b91a64429688f73c27db0eaadc5414db/scadsfct}, interhash = {7c786336ddfb0cb1bb45ed6498b87571}, intrahash = {b91a64429688f73c27db0eaadc5414db}, journal = {Proc. Conf. AAAI Artif. Intell.}, keywords = {Quantile-based detection likelihood maximum outlier training}, month = mar, number = 19, pages = {21610--21618}, publisher = {Association for the Advancement of Artificial Intelligence (AAAI)}, timestamp = {2024-11-12T15:19:09.000+0100}, title = {Quantile-based maximum likelihood training for outlier detection}, volume = 38, year = 2024 }