Computer implemented method and device for anomaly detection

A device for anomaly detection, the device configured to:

• process a digital representation of a signal or image with a wavelet decomposition to generate a first plurality of decomposed representations;
• process a first decomposed representation of the first plurality of decomposed representations with a wavelet decomposition to generate a second plurality of decomposed representations;
• process a first decomposed representation of the second plurality of decomposed representations with a wavelet decomposition to generate a third plurality of decomposed representations;
• process a first decomposed representation of the third plurality of decomposed representations with a first generative model to determine a first likelihood;
• process at least one second decomposed representation of the third plurality of decomposed representations with a second generative model to determine a second likelihood;
• process at least one second decomposed representation of the second plurality of decomposed representations with a third generative model to determine a third likelihood;
• process at least one second decomposed representation of the first plurality of decomposed representations with a fourth generative model to determine a fourth likelihood;
• and
• detect an anomaly when at least one of:
• the first likelihood and/or the second likelihood and/or the third likelihood and/or the fourth likelihood, meets a criterium for anomaly detection.

A non-transitory computer readable storage medium on which is stored a computer program for anomaly detection, the computer program, when executed by a computer, causing the computer to perform the following steps:

• processing a digital representation of a signal or image with a wavelet decomposition to generate a first plurality of decomposed representations;
• processing a first decomposed representation of the first plurality of decomposed representations with a wavelet decomposition to generate a second plurality of decomposed representations;
• processing a first decomposed representation of the second plurality of decomposed representations with a wavelet decomposition to generate a third plurality of decomposed representations;
• processing a first decomposed representation of the third plurality of decomposed representations with a first generative model to determine a first likelihood;
• processing at least one second decomposed representation of the third plurality of decomposed representations with a second generative model to determine a second likelihood;
• processing at least one second decomposed representation of the second plurality of decomposed representations with a third generative model to determine a third likelihood;
• processing at least one second decomposed representation of the first plurality of decomposed representations with a fourth generative model to determine a fourth likelihood;
• and
• detecting an anomaly when at least one of:
• the first likelihood and/or the second likelihood and/or the third likelihood and/or the fourth likelihood, meets a criterium for anomaly detection.