AS-MOSES is a work in progress, and should only be used at this point by folks who want to help develop it...
Development is being led by Kasim in our Addis Ababa office, guided by Nil in Bulgaria (one of our long time OpenCog code and theory gurus...) For practical bio-AI work right now, you should use the good old standalone MOSES code... ben On Sun, Mar 3, 2019 at 5:12 AM Linas Vepstas <[email protected]> wrote: > > > > On Sat, Mar 2, 2019 at 2:21 PM Abu Naser <[email protected]> wrote: >> >> Hi Linas, >> >> It is probably easier for your to change for me. > > OK. > >> >> >> Thanks for letting me know the performance limitation of as-moses. > > > There might not be one. I'm just not sure. OpenCog is all about "how do you > represent knowledge?" and "how do you move it around frictionlessly between > different processing systems?" A meta-goal is to avoid brittle stove-pipe > constructions, AKA spaghetti-code of data moving from here to there. > > Moses does a very specific kind of machine learning. The way it represents > data is "almost like" how the atomspace does. The AtomSpace provides a > foundation for reasoning and inference and other kinds of knowledge > processing; (e.g. adding appropriate neural-net support is a > work-in-progress). So far, we don't have seamless, friction-free access > between different algorithms, and the different kinds of data people use. But > we're making small steps. In the meanwhile, you are welcome to create > stove-pipe constructions that are just good enough for what you need to do... > > -- Linas >> >> >> With regards. >> >> Abu >> >> On Sat, 2 Mar 2019 at 18:32, Linas Vepstas <[email protected]> wrote: >>> >>> Hi Abu, >>> >>> The changes you propose are completely reasonable. >>> >>> If you really want to participate in the formal coding process, you should >>> create a github pull request with these changes in them. .. or I could make >>> them for you. Let me know. >>> >>> A word of caution about as-moses: it may have some unintended performance >>> profile. The original moses got very optimized to run fast; but the >>> atomspace is an in RAM graph database which means it has indexes which >>> means index insertion has a real cost. Exactly how the two get stapled >>> together is an open work item. >>> >>> ----- >>> BTW -- if Nil or Kasim or other as-moses people are listening -- perhaps it >>> makes more sense to have make combo-trees be Value-trees during the >>> computation -- Values are NOT indexed, they're meant to be fast, provide >>> fast access. Then after computation is done, the best trees, the results >>> can be copied into the atomspace. >>> >>> Another possibility might be to have either a tree-Value, (i.e. a new kind >>> of Value that is a combo-tree) or something that converts LinkValues into >>> combo trees (or vice-versa). All of this might alleviate issues related >>> to performance and mapping & etc. I don't know where you guys have these >>> design discussions, but if any of this sounds appealing, let me know; I can >>> clarify details. >>> >>> --linas >>> >>> On Sat, Mar 2, 2019 at 10:56 AM Abu Naser <[email protected]> wrote: >>>> >>>> Thank you Ben for your suggestions. I have installed AS-MOSES (the version >>>> works on atomspace). However, I had to make following changes during >>>> installation: >>>> >>>> 3) In bscores.cc, I had to insert #include <set> >>>> 2) in build_knobs.cc, at line 237, I had to change distance(from, to) to >>>> std::distance(from, to) >>>> 1) In table.cc, at line 259, I had to change return >>>> distance(labels.begin(), pos) to return std::distance(labels.begin(), pos) >>>> >>>> Hope these changes will not raise other issues down the line. >>>> >>>> Thank you again Ben, I have now something to play around with. >>>> >>>> With regards, >>>> >>>> Abu >>>> >>>> >>>> >>>> On Fri, 1 Mar 2019 at 23:59, Ben Goertzel <[email protected]> wrote: >>>>> >>>>> The MOSES-for-supervised-learning stuff is a bit specialized, and >>>>> doesn't use the main OpenCog codebase, MOSES is a standalone system >>>>> >>>>> However Nil wrote code for importing learned MOSES models into the >>>>> Atomspace, and once you've done that you can play with PLN inference >>>>> or other tools for combining MOSES models w/ other knowledge >>>>> >>>>> A great project for a brave soul right now would be >>>>> >>>>> -- run MOSES on some categorial SNP or gene expression data to get a >>>>> big diverse model ensemble >>>>> >>>>> -- load these models into Atomspace >>>>> >>>>> -- use Nil's new Pattern Miner to find surprising patterns among the >>>>> MOSES models >>>>> >>>>> Then one could bring PLN into the story to integrate, say, Gene >>>>> Ontology into the surprisingness assessment... >>>>> >>>>> ben >>>>> >>>>> On Sat, Mar 2, 2019 at 7:41 AM Abu Naser <[email protected]> wrote: >>>>> > >>>>> > Hi Ben, >>>>> > >>>>> > Thank your for responding my queries. I will try to find the papers and >>>>> > guessing you are one of the authors of the papers. It would be very >>>>> > useful for me if I could get Bioinfromatics related toy examples to >>>>> > play around, which will help me to get to know the opencog code. Hope >>>>> > Mike will point me to one of those example code. >>>>> > >>>>> > With kind regards, >>>>> > >>>>> > Abu >>>>> > >>>>> > On Fri, 1 Mar 2019 at 23:14, Ben Goertzel <[email protected]> wrote: >>>>> >> >>>>> >> There are two pieces to what we've been doing w/ OpenCog for >>>>> >> bioinformatics >>>>> >> >>>>> >> 1) supervised classification of genomics datasets (e.g. SNPs, gene >>>>> >> expression) using MOSES >>>>> >> >>>>> >> 2) integration of various ontologies into the Atomspace, import of >>>>> >> MOSES classification models into the Atomspace, and use of PLN >>>>> >> reasoning (currently on a pretty ad hoc, human guided basis) to do >>>>> >> inference based on this combined data >>>>> >> >>>>> >> There are publications on both of these things, though I don't have >>>>> >> time to dig up the URL right now, and Mike can point you to the >>>>> >> code... >>>>> >> >>>>> >> ben >>>>> >> >>>>> >> On Sat, Mar 2, 2019 at 5:51 AM Linas Vepstas <[email protected]> >>>>> >> wrote: >>>>> >> > >>>>> >> > Hi, >>>>> >> > >>>>> >> > Opencog has many parts to it.There is an agi-bio part for >>>>> >> > bioinformatics but it seems slim, and I think the main part of that >>>>> >> > system is somewhere else. I'm hoping Mike Duncan can respond in more >>>>> >> > detail. >>>>> >> > >>>>> >> > --linas >>>>> >> > >>>>> >> > On Fri, Mar 1, 2019 at 1:15 PM Abu Naser <[email protected]> wrote: >>>>> >> >> >>>>> >> >> Hello Everyone, >>>>> >> >> >>>>> >> >> I have just installed opencog on Ubuntu. I am a biologist and wish >>>>> >> >> to use opencog for solving biological problems. I have been >>>>> >> >> wondering whether there is any tutorial related to Bioinformatics. >>>>> >> >> Please let me know. >>>>> >> >> >>>>> >> >> With best regards, >>>>> >> >> >>>>> >> >> Abu >>>>> >> >> >>>>> >> >> -- >>>>> >> >> 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/b91a8dda-25db-48ea-80ed-776fed30dea6%40googlegroups.com. >>>>> >> >> For more options, visit https://groups.google.com/d/optout. >>>>> >> > >>>>> >> > >>>>> >> > >>>>> >> > -- >>>>> >> > cassette tapes - analog TV - film cameras - you >>>>> >> > >>>>> >> > -- >>>>> >> > 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/CAHrUA34wuXH1z%3DiDwE9_HX7%2BN04SPyTHjzSp8hKPR6%2Bgih1rXA%40mail.gmail.com. >>>>> >> > For more options, visit https://groups.google.com/d/optout. >>>>> >> >>>>> >> >>>>> >> >>>>> >> -- >>>>> >> Ben Goertzel, PhD >>>>> >> http://goertzel.org >>>>> >> >>>>> >> "Listen: This world is the lunatic's sphere, / Don't always agree >>>>> >> it's real. / Even with my feet upon it / And the postman knowing my >>>>> >> door / My address is somewhere else." -- Hafiz >>>>> >> >>>>> >> -- >>>>> >> 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/CACYTDBe4hX_MK2x2wXRmZ6ZiVqGO5Q48FC%3DJoyPYWAUYOv3VBQ%40mail.gmail.com. >>>>> >> For more options, visit https://groups.google.com/d/optout. >>>>> > >>>>> > -- >>>>> > 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/CAMw3wdhhdkdNMGLGPxf0aRZOG9VkaU05sNdc87%2B7TOYYiZi0tg%40mail.gmail.com. >>>>> > For more options, visit https://groups.google.com/d/optout. >>>>> >>>>> >>>>> >>>>> -- >>>>> Ben Goertzel, PhD >>>>> http://goertzel.org >>>>> >>>>> "Listen: This world is the lunatic's sphere, / Don't always agree >>>>> it's real. / Even with my feet upon it / And the postman knowing my >>>>> door / My address is somewhere else." -- Hafiz >>>>> >>>>> -- >>>>> 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/CACYTDBf4tv8-sJ9FV%3DFbGwb9ny3GbZ%3DYRZWSS%2BbpKCjOYnBKpw%40mail.gmail.com. >>>>> For more options, visit https://groups.google.com/d/optout. >>>> >>>> -- >>>> 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/CAMw3wdgRL0UtUxbMY2%2BfmYH0aNO5CoY67dRyUVdb4NLd%3D7vo-Q%40mail.gmail.com. >>>> For more options, visit https://groups.google.com/d/optout. >>> >>> >>> >>> -- >>> cassette tapes - analog TV - film cameras - you >>> >>> -- >>> 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/CAHrUA35L-f0_VkcPfXzTK%2BhwVtVKzD78NCo1MnCH2ep%3Dhc%3DodQ%40mail.gmail.com. >>> For more options, visit https://groups.google.com/d/optout. >> >> -- >> 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/CAMw3wdiYO8dQoc9uLgy4x37aBonf8%2BCU7ypOS5%3DkcV6S2UNkxQ%40mail.gmail.com. >> For more options, visit https://groups.google.com/d/optout. > > > > -- > cassette tapes - analog TV - film cameras - you > > -- > 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/CAHrUA346w_c0w1F_c_jUONu3zj-MPa5VJq1tozz8Dkt0NRp0RQ%40mail.gmail.com. > For more options, visit https://groups.google.com/d/optout. -- Ben Goertzel, PhD http://goertzel.org "Listen: This world is the lunatic's sphere, / Don't always agree it's real. / Even with my feet upon it / And the postman knowing my door / My address is somewhere else." -- Hafiz -- 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/CACYTDBcCxPJLF5ky3UF-OcRvZidrYPC4Jxyb%3Dz_QkEA6bxWWTA%40mail.gmail.com. For more options, visit https://groups.google.com/d/optout.
