Hihi Matt, Thanks for your answer and references.
However, I'm confused about some items (probably due to my poor understanding). Thus will appreciate further references/help so that I will not be forever ignorant. 1. Based on https://github.com/numenta/nupic/wiki/Natural-Language-Processing, quoted under Design header, "However, without a hierarchy, it will not be able to formulate a deep understanding of input text, because it is limited to one small region of the brain within it's model.". My understanding is probably wrong and will appreciate any assistance to help correct my mistakes. HTM is based on hierarchical (eg higher function will have bigger picture of the world but each level will know the stuff for it's own level), Temporal (timed based) and memory (self explanatory) and also prediction. Does the above statement "without a hierarchy" means that currently nobody had created the hierarchy but nuPIC allow us to form our own hierarchy? Or does it mean in this version, there is no hierarchy support (which I find it strange as hierarchy is supposed to be a core component of HTM) but in future version, it will be implemented? 2. In the On Intelligence book located at http://papers.harvie.cz/unsorted/Jeff%20Hawkins%20-%20On%20Intelligence.pdf, are the 6 layers column (eg page 105, figure 11) corresponding to the 6 layers of neocortex for human? (Thus for dolphin, there shall only be 3 layers) 3. In https://www.youtube.com/watch?v=H3lPxKhi1I0 (CLA as implemented in NuPic), the term column in nuPIC, does it refer to whole column from layer 1 to 6 or only the column in few of the layers (eg 2-3)? In the clip at time 6.25 - 7.15, I understand that each cell in the column have different significance. Is there a limit to the number of relations to the columns? Eg We have 1000 cells in a column, my understanding is that each cell is a relation to the column (eg milk), thus one cell might refer to mum (in relation to milk), another cow, another cat etc. Does it means that there can only be 1000 relations to milk? Thanks Ng Kock Leong
