COMPUTER IMPLEMENTED METHOD AND DEVICE FOR ANOMALY DETECTION
A device and a computer implemented method of anomaly detection, characterized by processing a digital representation (302) of a signal or image with a wavelet decomposition to generate a first plurality of decomposed representations (304), processing a first decomposed representation (304a) of the...
Saved in:
Main Authors | , |
---|---|
Format | Patent |
Language | English French German |
Published |
28.07.2021
|
Subjects | |
Online Access | Get full text |
Cover
Loading…
Summary: | A device and a computer implemented method of anomaly detection, characterized by processing a digital representation (302) of a signal or image with a wavelet decomposition to generate a first plurality of decomposed representations (304), processing a first decomposed representation (304a) of the first plurality of decomposed representations (304) with a wavelet decomposition to generate a second plurality of decomposed representations (306), processing a first decomposed representation (306a) of the second plurality of decomposed representations (306) with a wavelet decomposition to generate a third plurality of decomposed representations (308), processing a first decomposed representation (308a) of the third plurality of decomposed representations (308) with a first generative model (310) to determine a first likelihood (pθ1), processing at least one second decomposed representation (308b, 308c, 308d) of the third plurality of decomposed representations (308) with a second generative model (312) to determine a second likelihood (pθ2), processing (214) at least one second decomposed representation (306b, 306c, 306d) of the second plurality of decomposed representations (306) with a third generative model (314) to determine a third likelihood (pθ3), processing at least one second decomposed representation (304b, 304c, 304d) of the first plurality of decomposed representations (304) with a fourth generative model (316) to determine a fourth likelihood (pθ4), detecting an anomaly when at least one of the first likelihood (pθ1), the second likelihood (pθ2), the third likelihood (pθ3) and the fourth likelihood (pθ4) meets a criterium for anomaly detection. |
---|---|
Bibliography: | Application Number: EP20200153553 |