traversing a large graph

[Ed Pell]

Continuing the question of traversing a large graph. Many are talking about the importance of context. When I cook pasta I can put much of my memory content on standby. I do not need to consider knowledge of physics, math, sports, cars, people, friends, etc.... I lower the load by at least two orders of magnitude, maybe three.

In neurons the distal portion of the dendrite can form coincidence detector out of groups of say 40 synapses. If five triggers not much signal sent to the soma. With 20 triggering large signal sent (none linear exponential raise with number). This can provide a context priming.

With a graph we could mark each node with a sparse coded vector and sip distant nodes on a first pass. If we have time and are "deep thinking" we can consider more distant nodes.

Just an idea.

Ed    

[Linas Vepstas]

On Mon, Aug 22, 2016 at 7:53 PM, Ed Pell <edp...@gmail.com> wrote:

With a graph we could mark each node with a sparse coded vector and sip distant nodes on a first pass. If we have time and are "deep thinking" we can consider more distant nodes.

The opencog graphs are already represented as edges and vertexes, not vectors,  so they are already sparse.  The concept of distance that you refer to is called "attention", and "attention allocation" helps focus on those atoms that are important for achieving the current task.  (Every atom has an attention value associated with it).

--linas 

For more options, visit https://groups.google.com/d/optout.