Maybe a dumb question, but when we're talking about "W", are you talking about the letter W or the idea of W or the sound W or what? Aseem, what would be the actual raw input to the system you're imagining? Are you tackling text parsing? --------- Matt Taylor OS Community Flag-Bearer Numenta
On Mon, Mar 10, 2014 at 3:22 AM, Marek Otahal <[email protected]> wrote: > > On Mon, Mar 10, 2014 at 11:08 AM, Aseem Hegshetye <[email protected]> > wrote: >> >> >> every letter has its own representation. should the letter 'W' have same >> representation in both words because it comes after 'O' or it should have >> different representation because 'O' that precedes it is actually preceded >> by two different letters 'H' and 'C' ? > > the context spans longer (dependent on calling TP.reset() and an "aging > variable") sequences, so the representations will differ. > >> >> Should 'W' have same cells firing in both words? [ columns for 'W' will be >> same] > > ok, now im not sure if i understand. The representation of letter W by > encoder will always be the same, The "end-result" representation (output of > CLA after feeding the sequence) will be different for the two cases of 'W', > see above. > > cheers, > > _______________________________________________ > nupic mailing list > [email protected] > http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org > _______________________________________________ nupic mailing list [email protected] http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org
