Training OOD Detectors in their Natural Habitats

Katz-Samuels, Julian; Nakhleh, Julia; Nowak, Robert; Li, Yixuan

doi:10.48550/arxiv.2202.03299

Cited by 1 publication

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…GBND obtains good OOD performance by using better and better OOD signals in the process of selecting OOD instances over and over again. [22] proposes a framework for constrained optimization on unlabeled wild data mixed by ID and OOD classes and applies ALM (Augmented Lagrangian Method) on a deep neural network to solve this constrained optimization problem. Although these methods using unlabeled data are closer to real-world situations than those using auxiliary OOD datasets and get more OOD signals than those generating virtual instances, these methods are still based on a strong assumption that ID labeled data is sufficient.…”

Section: Ood Detection With Unlabeled Datamentioning

confidence: 99%

Topological Structure Learning for Weakly-Supervised Out-of-Distribution Detection

He¹,

Li²,

Han³

et al. 2022

Preprint

View full text Add to dashboard Cite

Out-of-distribution (OOD) detection is the key to deploying models safely in the open world. For OOD detection, collecting sufficient in-distribution (ID) labeled data is usually more time-consuming and costly than unlabeled data. When ID labeled data is limited, the previous OOD detection methods are no longer superior due to their high dependence on the amount of ID labeled data. Based on limited ID labeled data and sufficient unlabeled data, we define a new setting called Weakly-Supervised Out-of-Distribution Detection (WSOOD). To solve the new problem, we propose an effective method called Topological Structure Learning (TSL). Firstly, TSL uses a contrastive learning method to build the initial topological structure space for ID and OOD data. Secondly, TSL mines effective topological connections in the initial topological space. Finally, based on limited ID labeled data and mined topological connections, TSL reconstructs the topological structure in a new topological space to increase the separability of ID and OOD instances. Extensive studies on several representative datasets show that TSL remarkably outperforms the state-of-the-art, verifying the validity and robustness of our method in the new setting of WSOOD.

show abstract

Section: Ood Detection With Unlabeled Datamentioning

confidence: 99%