AdBench
PulseAugur coverage of AdBench — every cluster mentioning AdBench across labs, papers, and developer communities, ranked by signal.
2 day(s) with sentiment data
-
RIFT: New deterministic anomaly detection method introduced
Researchers have introduced RIFT (Relative Isolation From Trees), a novel deterministic method for anomaly detection. This technique generates a minimum spanning tree and scores points based on the apparent sizes of tre…
-
New regularization technique tackles anomaly detection collapse
Researchers have identified a phenomenon called "convergence collapse" in certain anomaly detection models. This occurs when models trained to minimize squared-error loss become less effective at detecting anomalies as …
-
New methods adapt prior-data fitted networks for tabular anomaly detection
Researchers have developed new methods to adapt prior-data fitted networks (PFNs) for anomaly detection in tabular data. The study, which began by using frozen TabPFN features, found that scoring samples by their distan…
-
New DIFFINT method offers interpretable anomaly detection
Researchers have developed a new anomaly detection method called DIFFINT, which uses a differentiable interval bottleneck within an autoencoder. This approach allows the model to identify anomalies in numerical data whi…
-
New DIFFINT autoencoder offers interpretable anomaly detection
Researchers have developed DIFFINT, a novel autoencoder designed for interpretable anomaly detection in numerical data. Unlike traditional black-box models, DIFFINT structures its latent bottleneck as a set of soft, axi…
-
New theory explains inlier-memorization effect in outlier detection · arXiv paper
Researchers have developed a theoretical framework to explain the inlier-memorization (IM) effect, a phenomenon where deep learning models learn normal data patterns before anomalous ones. By studying a simple autoencod…
-
Diffusion models advance anomaly detection for diverse data types
Researchers are exploring the use of masked diffusion models for anomaly detection across various data types, including tabular, text, and integrated circuit (IC) measurements. These models learn to identify deviations …
-
MacrOData benchmark suite offers thousands of datasets for tabular outlier detection
Researchers have introduced MacrOData, a new benchmark suite designed to improve the evaluation of outlier detection methods for tabular data. This suite significantly expands upon existing benchmarks like AdBench by in…