When you say "canonical paper" are you thinking a peer reviewed paper or an updated white paper? Is it more important to be comprehensive (white paper) or published in peer reviewed journals? Or are you thinking something else? Jeff
-----Original Message----- From: nupic [mailto:[email protected]] On Behalf Of Matthew Taylor Sent: Tuesday, January 14, 2014 1:12 PM To: NuPIC general mailing list. Subject: Re: [nupic-discuss] Hong Kong meeting outcome Outstanding! Thanks so much Stewart, Rik, and Francisco! I just love hearing about things like this. I like your suggestions as well. --------- Matt Taylor OS Community Flag-Bearer Numenta On Tue, Jan 14, 2014 at 12:12 PM, Stewart Mackenzie <[email protected]> wrote: > Just a follow up. > > This meetup was one of the best nights we've had at our hackerspace. > > Rik gave a wonderful talk encompassing the ideas of nupic and you'll never guess who emailed me 2 hours before the talk. Francisco Webber happened to be in town. > The night was intense to say the least. > > Walk away points: > > NuPIC needs: > * a canonical paper detailing the current implementation and theory of NuPIC, something that gives a language for developers to talk to each other. Something that puts a very clear goal into the minds of all participants. Something up to date, in black and white, we can point to and say 'that' on page 6, for example. > * a framework to determine change efficacy to cortical learning algorithms. > > This gives us the community a strong rudder to direct a growing community. If we don't do this, I fear lead thinkers might be disenchanted and disengage from the community. We absolutely have to keep these guys engaged, focused and growing. Quick feedback via testing framework enables this, quick effecnive changes to algos is achieved by deeply understanding theory delivered in the (white)paper. It also allows implementers to be able to communicate with the thinkers. > > Other community adoption strategies such as nupic-core, API documentation (the list Matt extracted from the community) are important. But without the above two over arching guiding criteria, we, as a community will start to develop bad scientific habits. This inturn causes frustration in the lead thinkers. Matt's job becomes harder, developers go on their own mission and become content to just hack and have fun. The project stagnates, prediction quality doesn't improve. > > Might I make a proposition? May I literally lock the lead thinkers in a room for a week to nail this document down? Might you guys and gals get on a plane somewhere, not just numenta folk but core community members (this _firmly_ includes Francisco) and have a high bandwidth discussion nailing it. This document can only be formed by you guys which _also_ includes ways of testing NuPIC. Once we have this in place, we can start to learn the language and then forge a testing framework and this'll certainly help with the nupic-core. > > Thank you to Francisco for that wonderful talk, you really opened my eyes to quite a few important things. I certainly hope to see Rik and Francisco again! > > Any other nupic'ers traveling through town are welcome to stay with me. Don't be shy to ask! > > > Kind regards > Stewart > > -- > Please excuse my typos and brevity > > _______________________________________________ > nupic mailing list > [email protected] > http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org _______________________________________________ nupic mailing list [email protected] http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org _______________________________________________ nupic mailing list [email protected] http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org
