Exploring the "Repeating Inputs" problem

[Paul_Lamb]

@marty1885

cells_.growSynapse(last_active, apply_learning, initial_permanence_); //Make new connections to all active cells that’s not connected

The difference is probably here (this is exactly how I also first implemented the TM algorithm). In the official algorithm, there are a couple of differences:

1. Only grow connections to previously active winner cells (not to ones which were not winners)
2. If a segment has “max new synapse count” active synapses when its post-synaptic cell becomes predicted active, then do not grow any additional synapses (to winners or otherwise)

@mrcslws explained the reasoning behind this:

@mrcslws

it won’t grow new connections to the previous winner cells, because it already has SYNAPSE_SAMPLE_SIZE (a.k.a. “maxNewSynapseCount”) active synapses.

If this logic used “number of synapses to previous winner cells” rather than “number of synapses to previous active cells”, then it would have the alternate behavior that you’re expecting. But that would have other bad effects: if the TM learns sequences “ABCD” and “XBCY”, it would assign the same SDR for both occurrences of C, and then it would always predict a union of D and Y afterward, regardless of whether it had seen “ABC” or “XBC”.