…like boost strength, boost period, learning rate and like that.
You can find them by running a swarm optimisation.
http://nupic.docs.numenta.org/stable/guides/swarming/running.html
1 Like
But aren’t there hyperparameter sets already computed on usual datasets publicly available?
We have example codebases, each with sample model parameters. We have a standard set of parameters for anomaly detection on single input scalar data.
1 Like