REVIEW 6 cited by
How to Exploit Hyperspherical Embeddings for Out-of-Distribution Detection?
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
Out-of-distribution (OOD) detection is a critical task for reliable machine learning. Recent advances in representation learning give rise to distance-based OOD detection, where testing samples are detected as OOD if they are relatively far away from the centroids or prototypes of in-distribution (ID) classes. However, prior methods directly take off-the-shelf contrastive losses that suffice for classifying ID samples, but are not optimally designed when test inputs contain OOD samples. In this work, we propose CIDER, a novel representation learning framework that exploits hyperspherical embeddings for OOD detection. CIDER jointly optimizes two losses to promote strong ID-OOD separability: a dispersion loss that promotes large angular distances among different class prototypes, and a compactness loss that encourages samples to be close to their class prototypes. We analyze and establish the unexplored relationship between OOD detection performance and the embedding properties in the hyperspherical space, and demonstrate the importance of dispersion and compactness. CIDER establishes superior performance, outperforming the latest rival by 19.36% in FPR95. Code is available at https://github.com/deeplearning-wisc/cider.
Forward citations
Cited by 6 Pith papers
-
VAE with Hyperspherical Coordinates: Improving Anomaly Detection from Hypervolume-Compressed Latent Space
VAEs using hyperspherical latent coordinates compress the hypervolume to achieve superior unconditional and conditional anomaly detection on complex image datasets.
-
ArcAD: Anomaly-Rectified Calibration for Cold-Start Supervised Anomaly Detection
ArcAD is a plug-and-play push-pull calibration method that projects limited normal samples onto a hypersphere for compact clustering while synthesizing pseudo-anomalies and using real anomalies to refine the decision ...
-
ArcAD: Anomaly-Rectified Calibration for Cold-Start Supervised Anomaly Detection
Under cold-start scarcity, ArcAD's Sinkhorn-balanced hyperspherical clustering plus anomaly-guided repulsion lifts reconstruction-based anomaly detection, with the clearest gains (+3.7 to +11.2 I-AUROC) on large multi...
-
ConjNorm: Tractable Density Estimation for Out-of-Distribution Detection
ConjNorm reframes OOD detection score design as optimizing norm p in an exponential family density model via a Bregman divergence theorem, with a tractable Monte Carlo estimator, claiming SOTA gains on CIFAR-100 and I...
-
TINS: Test-time ID-prototype-separated Negative Semantics Learning for OOD Detection
TINS improves OOD detection by learning negative semantics at test time with ID-prototype separation, cutting average FPR95 from 14.04% to 6.72% on the Four-OOD benchmark with ImageNet-1K.
-
VAE with Hyperspherical Coordinates: Improving Anomaly Detection from Hypervolume-Compressed Latent Space
Compressing a VAE's latent vectors onto a small 'island' on a hypersphere improves kNN-based anomaly detection on several image benchmarks.
Discussion (0). Sign in to comment.