Anomaly Detection algorithm takes more than 20 hours



I don’t know if that is the right code in HTM Studio to change or not. The point of HTM Studio is to allow people like you to find out if HTM will produce valuable anomalies for your specific temporal scalar data. Now that you know that information, it is up to you to build something that fits your needs, either with NuPIC, another existing HTM implementation, or even create your own HTM.


The NAB code has a SPATIAL_TOLERANCE parameter that might take care of this already:

I don’t think HTM Studio has this built in, but I’m not sure.