Hello,
I am playing with anomaly detection using a network with a Spatial Pooler and Temporal Memory on an example of a slightly noisy square wave that just stops at a point in time. I am wondering if like an SVM I should expect to see an anomaly raised where the square wave should restart but fails to do so. In playing with parameters, I have not managed to achieve detection of the signal stopping. Other types of detection seem to work.
Any suggestions would be appreciated.
Thanks,
Jason