Comparing NuPIC to other ML techniques

Richard Crowder edited this page Dec 23, 2015 · 2 revisions
Clone this wiki locally

There have been a lot of calls for comparisons of NuPIC's CLA to other AI/ML approaches (SVM,deep learning,..) be it in the form of reviewed academic publications, benchmarks on standart datasets etc.

There's a NPC for that (TODO link)

And here's first fruit:

Subutai's CLA-ARIMA (auto-regressive integrated moving average)

data: real world, time-series prediction (hotgym dataset) TODO: howto paste a file :P http://lists.numenta.org/pipermail/nupic_lists.numenta.org/2013-September/001411.html

Evaluating Real-time Anomaly Detection Algorithms - the Numenta Anomaly Benchmark

October 2015 - http://arxiv.org/abs/1510.03336 (https://github.com/numenta/NAB)

Continuous online sequence learning with an unsupervised neural network model

December 2015 - http://arxiv.org/abs/1512.05463