Re: TemporalMemory and Network API

2014-11-13 Thread David Ragazzi
> David, the predictions are put in `bottomUpOut`. `topDownOut` would be for reconstruction I believe and apparently wasn't implemented for the new TM. The TPShim allows the region for the old temporal memory class to be used with the new class (they have slightly different interfaces). Thank you,

Re: TemporalMemory and Network API

2014-11-12 Thread Scott Purdy
We will probably create a new region for the new TM at some point but it just hasn't been a priority so far. On Wed, Nov 12, 2014 at 4:32 PM, Scott Purdy wrote: > David, the predictions are put in `bottomUpOut`. `topDownOut` would be for > reconstruction I believe and apparently wasn't implement

Re: TemporalMemory and Network API

2014-11-12 Thread Scott Purdy
David, the predictions are put in `bottomUpOut`. `topDownOut` would be for reconstruction I believe and apparently wasn't implemented for the new TM. The TPShim allows the region for the old temporal memory class to be used with the new class (they have slightly different interfaces). On Wed, Nov

TemporalMemory and Network API

2014-11-12 Thread David Ragazzi
Hi Chetan and others, I would like know if Temporal Memory is fully integrated to Network API.. I have tried getting predicted values from temporal node but when I set TopDownMode = True, API return this: 'TPShim' object has no attribute 'topDownCompute' The same with anomalyMode = True I have