"Advances in Modeling Neocortex and its Impact on Machine Intelligence" by Jeff Hawkins (SmithGroup)
Ғылым және технология
Smith Group Lecture by Jeff Hawkins presented at the Beckman Institute for Advanced Science and Technology at the University of Illinois at Urbana-Champaign on November 12, 2010. Introductions by professors Arthur Kramer and Stephen Levinson. Hawkins is the co-founder of Numenta, founder and director of the Redwood Neuroscience Institute, and the founder of Palm and Handspring.
Пікірлер: 29
He is a great speaker. He is passionate about the things he is working on. "On Intelligence" is my favorite book.
Awesome. I love Jeff's lectures.
24:00 3-4 Layer feedforward system prediction in neocortex theory
58:00 time coding representation intrinsic in imaging along one axis (q allostatic shift cases neural algorithm equ) [JR]
30:14 MN Neuron Model (linear sum proximal feedfwd / threshold distal rtal)
@keysle Maybe sound isn't interpeted as 1d by the brain. According to wikipedia (Hearing sense article) the nerve pulses sent to the brain contains more types of info (ex location) than we usualy have with a digital recorded sound. I think all of our senses works as sequences of 2d input for the brain.
44:00 DDx theory [JR] learning rules from bio out put data insp 46:00
14:00 connectivity theory
9:00 premises & assumptions
I like the way you think.
1:10:00 negotiation of runaway propagation in predictive seq (excitoto tox & differentiation biasing to chose correct novel env model changes from rep input) (burst firing in innovation pyramidl examplr)
40:00 Predictive machines, resting potential & sideways inhibition
11:50 Intelligence cycle (cognitive)
and your my inspiration.
"but it lacks unity. Singularity will always revolt." Can you explain in more detail what you mean by this?
It's way too long for a comment but i can suggest a list of material that only leads to further lists of even more material. Start with recreation of action potential. History of known religion. Egyptian history and theory. You can dabble in Tree of Life and global network theory. You can dabble in kundalini theory or merkaba's with some basis on action potential. You can also get a lot of weird ideas in unconscious living energy or individuality in moderation. All mass was energy. I can go on
Debating how I should respond to this... I came up with two options... 1) Holy Crap, Jeff?!?! ha-ha. 2) Hey, I just met you and this is crazy...
50:30 network/t fraud detection & predictive search hx profiling (Private sector developed as Gen Hayden said)
32:30 HTM Regions
and I love you.
I'll call you maybe.
It seems that they developed the theory...what is left? :S
C-C-C-Combo Brea..... wait...
Not trying to. If anything I would say my last post sounds preachy if not apocalyptic. Still, we could spend days going through the useless knowledge in my head that serves me so little purpose other than beer talk. The real quest I see is in things like neuroscience or trying to compare far too many religions coming out of one another. That quick social wit to pin left or right brain on some confidence or pity. Knowledge that gets you laid. This shit, farts in the wind til science catches up.
So if life becomes a machine and rather than trying to create AI, you create the capacity to generate predictive algorithms with a changeable preface that can update ability based on whatever it perceives to need new predictive algorithms for? I like it on a biological level, but it lacks unity. Singularity will always revolt.
You try to respect 500 characters with 5 million years of facade truths. The fact you insult it means this is a logical reinforcement to an emotion I have no knowledge of. I'm not even going to attempt to figure out your defenses. I have come to my ideas in my own right. You come to your own conclusions. However I stress ideas and not knowledge. The whole human limitation will never allow complete chaos. In time, everything you know will be a burning mass of energy. No more or less life.
If you can't summarize what you mean in 500 characters, then it is highly likely to be a load of bollocks.