Hello NuPIC, We have been wondering at Numenta whether it is possible to be even more open about our algorithm research. Here is a proposal, and I'd like to know what you think.
We could, in an effort to be more transparent with ongoing prototypes and experiments with algorithm implementations, publish this work in a public repository at https://github.com/numenta/nupic.research. This would be temporary experimental code, which poses some challenges. This would mean: - it can change without warning or explanation - it will change quickly as experiments are discarded and recreated - it might not change at all for a while - it could just be plain wrong or buggy for periods of time - code will not be production-quality - comments and questions about this code may be ignored - Numenta is under no obligation to properly document or explain this codebase or follow any understandable process - repository will be read-only We want to be as transparent as possible, but we also want to move fast with these experiments so the finalized algorithms can be included into NuPIC as soon as they are ready. Would a public repository like this benefit the NuPIC community? Or create confusion about what's coming next? --------- Matt Taylor OS Community Flag-Bearer Numenta PS: A copy of this message was send to nupic-discuss and nupic-theory. You may safely respond to either message to send back to the respective list. I'll be monitoring both threads, so either one is fine.
