US 12,033,048 B1
Anomaly detection using feedback
Laurent Callot, Berlin (DE); Jasmeet Chhabra, Sammamish, WA (US); Lifan Chen, Mercer Island, WA (US); Ming Chen, Seattle, WA (US); Tim Januschowski, Berlin (DE); Andrey Kan, Seattle, WA (US); Luyang Kong, Seattle, WA (US); Baris Kurt, Berlin (DE); Pramuditha Perera, Secausus, NJ (US); Mostafa Rahmani, Santa Clara, CA (US); and Parminder Bhatia, Seattle, WA (US)
Assigned to Amazon Technologies, Inc., Seattle, WA (US)
Filed by Amazon Technologies, Inc., Seattle, WA (US)
Filed on Nov. 30, 2020, as Appl. No. 17/107,820.
Int. Cl. H04L 29/06 (2006.01); G06F 18/214 (2023.01); G06N 20/20 (2019.01)
CPC G06N 20/20 (2019.01) [G06F 18/214 (2023.01)] 20 Claims
OG exemplary drawing
 
1. A computer-implemented method comprising:
receiving a request to develop an anomaly detection system having at least one scoring model;
training the anomaly detection system on training data as indicated in the request;
receiving data to detect potential anomalies using the anomaly detection system;
processing the data using the anomaly detection system to score the data to determine when the data is potentially anomalous based on a threshold;
requesting feedback on at least one determined potential anomaly;
receiving feedback on the least one determined potential anomaly;
determining the feedback is positive;
determining an anomaly detection rate is below a detection rate threshold; and
adjusting, based on the feedback and based on the determining the anomaly detection rate is below the detection rate threshold, the threshold used to determine when the data is potentially anomalous.