Interesting question and, from my perspective, much larger than one may
think.
We are so used to the concept of money-driven research that we forget
about research 'per se'.
I think that Numenta's main goal isn't profit but understanding and
eventually recreating the product of the massive, inimaginable, amount
of brute force engineering (i.e. evolution) that nature has already
invested in for us: the human brain.
I wonder how Jeff and all of Numenta remains so open - and I'm
searching for a "context" to hold this process in, so that I have a
totally inclusive way of looking at this choice to be open sourced and
transparent. Please help me understand this choice and how Numenta
views this?
Their white paper (along with the cortical's white paper on semantic
folding) is enlightening (in my opinion even for neurologists and
physiologists) about the deep understanding of the learning methods
behind the neocortex. That being said, could you copyright an algorithm
resembling the conversion from glucose to atp? Probably you could
copyright some software/code, but once the theory is out (with some
published article) everybody could recreate such an algo - because it is
made by nature, just discovered by you.
I think then that in such a case open source would be the 'only'
doable[1] way.
In the case of NuPIC, being opensource, you get lots of free beta
testers, suggestions, coders.. than can speed up your goal and enrich
your team.
I get concerned that the technology will somehow be "compromised" or
"tainted" or have its momentum siphoned off somehow by people taking
bits and pieces and those "pieces" somehow becoming very popular?
If those pieces performs better in real life scenarios (although I
understand that giving a score like in classic ML fields is pretty
difficult for HTM)... why not (for non-commercial products)? Of course,
it would be a must to acknowledge Numenta for their code (as, in any
case, is expected under AGPL license) and/or for their papers and
researches (as expected, generally, in science). In any case, that
wouldn't interfere what Numenta's goal: research.
My two cents.
Cheers,
Raf
[1] : You could of course create your own closed implementation for
paying clients - but very probably you'd have lots of open source
communities as competitors.
On 31/01/2016 10:35, cogmission (David Ray) wrote:
Hi,
Am I just a small person for being concerned and worried that someone
may grab portions of the technology and integrate it with classic ML
techniques and take credit for HTM theory under some renamed hybrid
without acknowledging origins in Numenta?
Or the possibility that people will just take it over and possibly
steer it in an "impure" (by "impure" I mean take it in a direction
that is not aligned with Numenta's projected trajectory for
enhancement) - leading to "impure" development (development not a
product of heavily considered correspondence with the biology), and
therefore misguiding its future?
I get concerned that the technology will somehow be "compromised" or
"tainted" or have its momentum siphoned off somehow by people taking
bits and pieces and those "pieces" somehow becoming very popular?
I wonder how Jeff and all of Numenta remains so open - and I'm
searching for a "context" to hold this process in, so that I have a
totally inclusive way of looking at this choice to be open sourced and
transparent. Please help me understand this choice and how Numenta
views this?
Additionally, I feel that this is an important human social experiment
and an avenue for human social growth too. I have never (to my
knowledge) seen a company be this transparent with its process and
products. For that reason, I feel this project is important on so many
levels, and there is so much to learn above and beyond the obvious
pursuit of creating man made intelligence through reverse engineering
the neocortex.
In short:
I wonder how Numenta sees this process and this wonderful experiment
unfolding? Secondly, I wonder how I should "think" about this process
so I hold it in the right context so that my actions represent this
community in the "right" way (the way that will nurture its progress
to its fullest extent)?
Cheers,
David
--
/With kind regards,/
David Ray
Java Solutions Architect
*Cortical.io <http://cortical.io/>*
Sponsor of: HTM.java <https://github.com/numenta/htm.java>
[email protected] <mailto:[email protected]>
http://cortical.io <http://cortical.io/>
--
Raf
www.madraf.com/algotrading
reply to: [email protected]
skype: algotrading_madraf
