DRÆM - A discriminatively trained reconstruction embedding for surface anomaly detection.
Vitjan ZavrtanikMatej KristanDanijel SkocajPublished in: ICCV (2021)
Keyphrases
- anomaly detection
- discriminatively trained
- three dimensional
- object detection
- intrusion detection
- discriminative learning
- network intrusion detection
- detecting anomalies
- network traffic
- discriminative training
- sequence classification
- anomalous behavior
- intrusion detection system
- network anomaly detection
- detect anomalies
- d objects
- negative selection algorithm
- generative model
- high resolution
- one class support vector machines
- unsupervised learning
- multi class
- pairwise
- data sets
- surface reconstruction
- pattern classification
- text classification
- natural language processing
- computer vision