Hi Azat, The CEPT SDR's are produced from "greyscale" images using a process similar to the Spatial Pooler. The greyscale images are a 2D representation of 16K semantic attributes derived for each word. Francisco from CEPT has been involved in helping to prepare for the NLP parts of the hackathon, and some of the tasks will be using their data and API.
Regards, Fergal Byrne On Mon, Oct 21, 2013 at 4:06 PM, Azat <[email protected]> wrote: > Hi Matt, > > Definitely great real-life data to try and I found some datasets online > to work with before the Hackathon (not sure but I don't recall any home > automation at 1st one ?) - again multichannel and spanning months. Just > hoping somebody else tried with different parameters and results. This > reminds me of Nest thermostat (and now smoke alarms), but I'm sure they > don't use CLAs :-) > > Do you guys think a Grok-like region (1 layer, 2048x32) is enough for > home automation? > > As for card tx-ns - some places still use rule-based engines and/or > human monitoring, and maybe the field is ripe for the CLA's application. > > Azat > Btw: one of the next topics at the hackathon is NLP by which something > similar to CEPT's work is meant ? Anyone have info on how they get their > SDRs for words ? > > _______________________________________________ > nupic mailing list > [email protected] > http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org > -- Fergal Byrne <http://www.examsupport.ie>Brenter IT [email protected] +353 83 4214179 Formerly of Adnet [email protected] http://www.adnet.ie
_______________________________________________ nupic mailing list [email protected] http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org
