Hi Aseem,

That sounds amazing! I had the same feeling but was too lazy to test the
idea, so well done. I can't see anything there at the URL, so perhaps you
need to change the access permissions (it's listed as CLA under your
username level).

I have a suspicion adding my TP hack to the SP might help with the
prediction part, so perhaps we can collaborate on that. See my Friday email
about that.

Looking forward to discussing this (once I recover from jet lag and a very
intense and marvellous week spent on NuPIC).


On Fri, Nov 8, 2013 at 2:13 PM, Aseem Hegshetye <[email protected]>wrote:

> Hi,
> I registered for last hackathon remotely but couldnot finish my project.
> Now its ready.
> I rebuild CLA and wanted to test if reconstruction is possible without
> look up tables.
> IT IS !!
>
> I am uploading the code on github.
> https://github.com/aseem-hegshetye/CLA
>
> I dont know if I have uploaded it correctly.
> Prediction part has something wrong in it.. It does not predict well but
> after learning is over if i use the current memory matrix to make
> prediction through connections from columns to the input bits, it works.
>
> Later on it gets stuck within 2 possibilities.
>
> thanks
> Aseem Hegshetye
>
> _______________________________________________
> 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

Reply via email to