On Fri, Apr 25, 2014 at 10:10 PM, Anastasios Tsiolakidis <[email protected]>wrote:
> On Fri, Apr 25, 2014 at 1:40 PM, YKY (Yan King Yin, 甄景贤) > <[email protected]>wrote: > >> As a general rule, the more expressive a logic, the slower the inference >> engine. So this is a trade-off situation > > > Thanks, we are getting somewhere. So, in your opinion, there really has > been no "inference progress" > Only minor progress -- all inference algorithms are combinatorial in nature, and they all exhibit exponential worst case or worse (eg undecidable). I'm hoping to find some techniques to translate the combinatorial search to a continuous setting... and one would have to write in "logic assembly language" to have any chance > of a logic software system performing acceptably, compared to some other > statistical or whatever optimized alternative number-crunching (putting > words in your mouth here, lol, basically thinking of e.g. playing chess > with Prolog vs a minimax engine). Is there really no compiler that can > translate expressive logics into passable simple ones? > Translating to propositional logic (ie propositionalization) is not the only inference technique, in fact it is not the most popular (but we're not sure if it is a dead end or not). The fastest algorithms seem to be specific to each type of logic. For example there are fast propositional solvers (eg the DPLL algorithm and its variants), and then there are fast first-order logic provers (the fastest currently is Vampire based on the first-order resolution algorithm + heuristics). To quote a simple example, I'd rather remove castling from a list of > possible moves, rather than keep evaluating castled(t_5) all the time, > obviously. Am I asking for too much? > It is obviously possible with first-order logic, eg, you can easily represent the problem in Prolog in many ways you want. For propositional logic it may be a bit tricky -- perhaps you have to generate different propositions for each iteration and call the solver. So the set of propositions is not static -- the set is constantly being generated. That's what I mean by "propositionalization". ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657 Powered by Listbox: http://www.listbox.com
