On 4/16/19 6:52 AM, Linas Vepstas wrote:
What about PLN? Well, today's PLN, built on the pattern matcher, will run thousands of CPU cycles and then do a small handful of float-point ops. The good news: almost all PLN pattern matches are really quite very small (that's good- it should be fast) and of course all the current PLN demos fit into 300MB. I currently have no clue how to apply PLN to large datasets.
I should correct that, most forward chainer queries are small because the forward chainer is in fact an iterative chainer, processes each rule iteratively.
However most backward chainer queries are big because the entire inference tree is queried (well, only the leaves but that's still a lot). However each query is very similar, for that reason, it is likely some general enough form of cashing/memoizing could speed that up.
Another form of memoization that could speed up the URE is unification, fortunately I know how to do it because the unification code is mostly functional. For the pattern matcher I don't have much clue, maybe Andre Senna's Pattern Index can help along the way.
Are there other ways to think of PLN that do less searching and more multiplying? Can you swap inner and outer loops somehow? Nil might like to ponder this.
Accurately calculating the second order probabilities can take a lot of multiplying, and is very amenable to paralellization. As for swapping the inner and outer loops, maybe using macro rules is a way to do that, it's likely that in practice there's gonna be a number of macro rules that are gonna be used a ton of times.
Nil -- You received this message because you are subscribed to the Google Groups "opencog" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at https://groups.google.com/group/opencog. To view this discussion on the web visit https://groups.google.com/d/msgid/opencog/d8605d93-df5a-6de9-0078-80e46196b8f3%40gmail.com. For more options, visit https://groups.google.com/d/optout.
