I’d love to use sci-kit-learn or sklearn or even just a python package to use the HTM to produce future predictions. What’s the best implementation to use?
1 Like
I think one big issue is that HTM addresses “intelligence” as a problem of continuous on-line learning and prediction of a data stream.
frame1 → next_frame_guess → frame 2 → (corrective learning +) next_frame_guess → frame 3 …
While sklearn has a quite different (statistical?) metaphor. Input data set and a corresponding output data set for training, and another pair of input-output data set for testing.
Closer to HTM would be RNN (LSTM, GRU) models (without the online learning capability), or even (remotely related) reinforcement learning but I don’t think any of these are represented in sklearn.
1 Like