Biomedical, image, and embedding data¶
These examples show robust-distance scoring on feature vectors extracted from signals, images, or embeddings.
Biomedical signal windows
Detect abnormal signal windows from correlated time and frequency features.
Image-feature anomaly
Find unusual feature vectors in image or representation spaces.
Text embedding outliers
Screen embedding vectors for off-topic or shifted observations.
How to use this topic¶
Start with the first card if you want the simplest demonstration. Then move to the more specialized page when the data shape matches your problem. Every page includes captured output, plots, interpretation notes, and a command to reproduce the result.