18978437. MULTI-SCALE NEURAL NETWORK FOR ANOMALY DETECTION (INTEL CORPORATION)
MULTI-SCALE NEURAL NETWORK FOR ANOMALY DETECTION
Organization Name
Inventor(s)
Anthony Daniel Rhodes of Portland OR US
Celal Savur of Hillsboro OR US
Bhagyashree Desai of Brooklyn NY US
Richard Beckwith of Portland OR US
Giuseppe Raffa of Portland OR US
MULTI-SCALE NEURAL NETWORK FOR ANOMALY DETECTION
This abstract first appeared for US patent application 18978437 titled 'MULTI-SCALE NEURAL NETWORK FOR ANOMALY DETECTION
Original Abstract Submitted
A neural network model for anomaly detection may include convolutional blocks with different spatial scales. The model may be trained with training data, which may be normal data that lacks anomaly. The convolutional blocks may generate embedding features having different spatial scales. A distance between each embedding feature and a corresponding model embedding may be determined. The distances for the embedding features may be accumulated for determining a loss of the model. The model may be trained based on the loss. An accuracy of the trained model may be tested with testing data that has verified anomaly. One or more convolutional blocks may be selected from all the convolutional blocks in the model, e.g., based on the spatial scales of the convolutional blocks and the spatial scale of data on which anomaly detection is to be performed. The selected convolutional block(s) may be used to detect anomaly in the data.