Skip to main content
No. of Recommendations: 3
I found this hour long interview with James Douma very interesting in part because some of my co-workers were trying to figure out how to teach computers to recognize images back in 1985. I was certain that neural networks were the answer if we ever figured out how neural networks might be built and operated but it never went past being a curiosity for me. Now that they actually exist it's fun to find out how they actually are made to work.

A good alternative title might be: "How to use brute force to solve problems that cannot be solved by brute force." The brain solves the visual problem by brute force, billions of neurons assigned to the job.

Tesla's Neural Networks, Autonomous Driving, and Computer Vision w/ James Douma

https://www.youtube.com/watch?v=gzq5AFCH5jk

Denny Schlesinger



more than 100 billion neurons that are the normal complement of a newborn baby

https://www.ncbi.nlm.nih.gov/books/NBK234146/
Print the post Back To Top
No. of Recommendations: 4
Hey, Denny,

It's interesting also that the human brain has a small portion hard-wired for certain things that have been useful since the early humans and still are today.

One is facial recognition; there is a small portion of the brain that performs only facial recognition and when someone has an injury or deficiency in that area, they are often totally unable to recognize anyone, and some cannot see faces at all, but only a "round blur" where a face should be. It seems that those who could once recognize faces and now can not (injury) have a very hard time recovering and many never do, which means other parts of the brain evidently have a hard time taking over this function.

Another is movement of any kind. A portion of the brain is always on "standby" to catch and hopefully capture, any form of movement in the visual field. That's why, when you're sitting in the warm sun on a beach reading a good book and a bird swoops down close enough to be captured in your visual field, that you might be seriously startled, as the brain can signal an adrenaline release simultaneously along with the visual "alert" for movement.

I think all of this makes it even more difficult for engineers to interpret visual "cues" from a video feed, and more amazing when they do. It seems to me that one might spend a good portion of one's lifetime trying to define "CAT" visually, and even then would be bound to fail on occasion.

Talk about living in interesting times and such things as "S curves"! I think the rate of change (cultural, physical, electronic, transport, finance, leisure, entertainment, work, you name it) is just now ready to enter the truly exponential phase. I'm almost glad I won't have to try to cope with more change than that which will hit us in the next 20 years or so (if I'm lucky, at that.) The one thing I will miss that I thought would be here by now, if personal flying transportation. It's hard for me to believe that 3,000 lb machines moving at 70 mph are still the best way to travel, while 12" away from similar machines going in the opposite direction on the other side of the dotted line.

Back to the brain, I have often wondered if that part of my brain built for facial recognition is weak, as I often do not recognize people, sometimes those that I have encountered fairly often. Or, it could be because I'm 68 and don't remember what I had for dinner last night. Yeah, that's probably the case.

Dan
Print the post Back To Top
No. of Recommendations: 3
RaptorDan, interesting comment on hard wired parts of the brain. My guess is that the more vital a function the more likely it is hardwired. Instant facial recognition must be an important part of the fight or fight instinct. It's also amazing how one recognizes voices even over crappy telephone connections.

Initially I used to think in terms of a rational waking brain and a subconscious brain that did a lot of work while one is sleeping. I used to wake up very early on occasion with the solution to a hard problem right in front of my eyes with no idea how it came to be there. The Indian mathematician Ramanujan called it intuition. Now I think that "intuition" and "subconscious" are actually neural networks working while we sleep undisturbed by the normal chaos of being awake.

"Intuition" and "subconscious" cannot explain how they do it because it's not the rational, boolean brain at work. There are a lot of manifestation of complex behavior based on very simple rules like the schooling of fish, the flocking of birds, and how ants and bees find their way to and from their nests and hives. Somehow this ties in with what I call the natural laws of complex systems like the Sigmoid growth curve and the power law distribution of wealth. I have no idea why nature does it that way but we can take advantage of the knowledge that that is how nature works.

In At Home in the Universe Stuart Kauffman calls it "Order for Free."

Denny Schlesinger
Print the post Back To Top
No. of Recommendations: 1
I hope I'm not beating this to death when we should be discussing CRWD or ARKW, Denny. I just find the brain one of the many fascinating topics of current science.

Now that we're on the front lines of exploring AI, I find it interesting, to say the least, that almost all scientists I've heard speak about working to merge AI and cyber tools, find one topic at the epicenter of the subject: Consciousness. It seems we have a very hard time defining it, let alone duping it in a man-made neural field. With all the questions still unanswered in science, many scientists feel this may be one of the toughest to understand and eventually to emulate.

So maybe "Hal" is still science fiction only. Then again, given the deluge of scientific discovery these days, maybe not.


"Hal, let's return to an investing topic."

"David, I prefer to talk about AI."

"Hal, let's return to an investing topic."

"No."



(Now what?)

Have a great weekend all,

Dan
Print the post Back To Top
No. of Recommendations: 3
I hope I'm not beating this to death when we should be discussing CRWD or ARKW, Denny. I just find the brain one of the many fascinating topics of current science.

You can beat stock symbols to death. ;)


Now that we're on the front lines of exploring AI, I find it interesting, to say the least, that almost all scientists I've heard speak about working to merge AI and cyber tools, find one topic at the epicenter of the subject: Consciousness. It seems we have a very hard time defining it, let alone duping it in a man-made neural field. With all the questions still unanswered in science, many scientists feel this may be one of the toughest to understand and eventually to emulate.

Depends on which universe you inhabit. In the Newtonian universe it's hard to have free will because everything seems to be governed by simple cause and effect. In the quantum universe where there is uncertainty, free will is possible because an agent might be permitted to make choices that are impossible in a predetermined universe. Why do I address free will and not consciousness? Because it's doing something about what we are conscious about that matters.

To bring this conversation back to investing... It's quite clear from The Gorilla Game and from the Science of Complexity that complex systems don't exhibit a simple cause and effect pattern. The Gorilla Game's buy the basket and sell the losers is a practical example of the best effort not necessarily causing the best outcome. In the Newtonian universe everything has a single point of equilibrium that classical economists love. In the quantum universe there are several potential points of equilibrium and one is chosen (seemingly) at random. Do complex systems have consciousness and self determination? At which point between subatomic particles and conscious humans do consciousness and self determination manifest?

Denny Schlesinger

It's a gorgeous day in Vila Nova de Gaia. All showered are ready to go walking, consciously and self determinedly... 😇
Print the post Back To Top