Hi Chetan,
On Fri, Nov 15, 2013 at 4:57 AM, Chetan Surpur <[email protected]> wrote: > Mark, > > As for your hot plate example, I don't think it's the fact that the pain > comes 1 second or 1 hour later that makes the difference, it's that 1 > second's worth of inputs to the brain vs 1 hour's worth of inputs to the > brain has been processed before the pain is processed. > Yes, that's exactly what I was saying. From the texts above (I hope) it's clear I assume a certain # of (brain) ticks per second. And the stream always runs. So, more accurate to say would be, there;s 1000 ticks between the touch and pain, contrary to 10^6, say. > > One hypothesis is that the brain doesn't need to keep track of time > internally, it just has a feel for how much relative time has passed due to > how much input it's had to process and learn from since a particular event. > Now, I don't know if this is the case, but suffice to say that in NuPIC, we > can use time input as just another datum if we need, but if we just want to > measure the relative passing of time, we can just let the number of records > processed take care of correlating inputs that are close together in "time". > This is what I was trying to say is sufficient. Cheers, M. -- Marek Otahal :o)
_______________________________________________ nupic mailing list [email protected] http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org
