ฝัง
- เผยแพร่เมื่อ 4 เม.ย. 2022
- Have you ever wondered what makes you intelligent? How are you able to see, hear, think, read, sing, solve problems, and perform any number of intelligent tasks?
Your brain learns a model of the world, and this model recreates the structure of everything you know. Everything you do and experience is based on this model. Intelligence is the ability to create this model of the world.
But how can a bunch of cells in your brain create a model of the world and everything in it? The Thousand Brains Theory provides an answer. Not only that, but it also provides a blueprint for how to build truly intelligent machines.
Visit numenta.com/ for more information.
Produced by Mind's Eye Creative Studio: www.mec.co.za/
- - - - -
Numenta is leading the new era of machine intelligence. Our deep experience in theoretical neuroscience research has led to tremendous discoveries on how the brain works. We have developed a framework called the Thousand Brains Theory of Intelligence that will be fundamental to advancing the state of artificial intelligence and machine learning. By applying this theory to existing deep learning systems, we are addressing today’s bottlenecks while enabling tomorrow’s applications. - วิทยาศาสตร์และเทคโนโลยี
Still missing Matt and his HTM School Videos 😓
I'm sorry i was the 70th like but it has to be liked and shared. Numenta is a great addition to understanding our brain and building intelligent machines. I had my eyes on it since a professor of mine introduced us to the book On Intelligence.
Likewise since reading On Intelligence i never missed anything (i know of) published by Jeff Hawkins and the brilliant minds of the Numenta team.
Yes, me the same. All starts from On Intelligence. Everything related to Jeff Hawkins perks my ears.
This is an excellent video . Thank you .
Very nice to see these ideas in animated form :)
Absolutely amazing video!
Nice explanation
Great video! Thank you :)
This was very interesting thank you
Nice nod to Wittgenstein with the "What is a game" and the idea of the language-game and how meaning arises out of the rules the meaning is embedded in
Very cool.
At some level it have some similarity to Marvin Minsky's amazing "The Society of Mind"
Great video! I’ll start working on creating a demo to understand this better and go deeper ;). Nupic code is updated with this new model?
Thanks! NuPIC has been in maintenance mode for several years so it is not updated with our new TBT algorithms. We're currently focused on building sensorimotor-based models of these cortical columns, and figuring out how knowledge is encoded in these models via reference frames. Stay tuned :-)
@@NumentaTheory Its pitty your dec code is not available for that. I'll try to base my work on the latest NUpic version and based on what you state in the papers and so on I'll do my best to generate something close ;). I'll stay tuned for any piece of info you issue on the matter. Attending the numenta open meetings will help I suppose also? Thanks anyway, if you need any assistance on the project I would be super hyped to contribute in any way possible ;).
Recently, interesting associations from research on associative learning in box jellyfish by Jan Bielecki, a neurobiologist at Kiel University in Germany. From Scientific American- "Bielecki and his colleagues looked for associative learning in small jellyfish that sport four eye structures called rhopalia that each contain six eyes and about 1,000 neurons, he says. (Each rhopalium takes turns acting as the jellyfish’s noncentralized nervous system.)" - Could the rhopalia be analogous to a reference frame as in the Thousand Brains theory?
Genius
What about the concept of self
Love it, but has some serious Jurassic Park vibes :)
Why the confidence that these machines will be able to think faster and sense more than humans? By virtue of having more cortical columns programmed into them, and a wider array of sensory inputs?
Our "hardware" is limited and evolution is slow. Unless we come up with smart bioengineering and genetic enhancements.
The machines will be faster as they operate at nanoseconds and biological life works in milliseconds.
Our primary senses are vision, touch and hearing but the types of sensors and embodiments used in intelligent machines are almost limitless. For example, we might create intelligent machines that exist inside individual cells and can directly sense and understand proteins, or robots capable of rescuing people trapped in caves with radar sensors so they can see in the dark.
And as you've mentioned, a machine that approaches or exceeds human intelligence will have a wide array of sensory inputs - just as we do. If you only had one finger for touching, you could still learn about the world but it would take much longer, and the actions you could perform would be limited, and vice versa.
@@NumentaTheory very nice, thank you!
@@AleksandarKamburov thank you! maybe when the machines exceed us, they can come up with the smart bioengineering and genetic enhancements that we need, in turn, to exceed them hehe
@@sunnyinvladivostok on this topic i recommend watching the movie Transcendence , but you need to read between the lines and also not take seriously the ending 😀
Wait, each column is the size of a grain of rice? Do you mean a grain of sand? A grain of rice seems awfully big. I know there are mini-columns and hyper-columns, but even a hyper-column that's the size of a grain of rice seems huge. Isn't that like 100,000 neurons? Do hyper-columns really have that many? Or is there another step up, mega-columns or something like that?
Neocortex is 2.5mm thick and one cortical column is about 1mm wide so it's close to rice grain or mini spaghetti strand size. Cortical column has about 100k neurons and 5k synapses per neuron for a total of 500 million synapses. 150 000 cortical columns in all of neocortex total. The mini-columns are smaller about 50 micrometers wide so there's few hundred mini-columns in a cortical column. :)
Scientist could prove this theory by optogenetically disabling and re-enabling single columns in animal experimental models?
hmmh i dont think there's a way to disable or enable one column genetically. Genes just code the structure of one cortical column and then that bluebrint is copied 150k times in the neocortex with some minor changes per area.
@@JaniMikaelOllenberg Optogenetically is possible to turn off or on genes in specific areas of the cortex with light. But to do so you must genetically modify the mouse and surgically implant fiber optics or other means of getting the electromagnetic radiation in the desired area.
you could then train the mouse to do recognition tasks and see how single columns do or don't contain specific models.
@@lucamatteobarbieri2493 yes. What i mean is that there's no way to isolate single cortical columns with that method because theyre all built with the same genes so making something light sensitive would affect all other columns too.
@@JaniMikaelOllenberg No. With optogenetics you can turn on and off neurons only where you shine the right frequency of light. The mouse must be genetically modified. Look it up.
For Coffee Cup's sake, could you please find another toy example? Please.
The Thousand Brains Theory of Understanding Coffee Cups 🧠☕👈👀💡 😋
Interesting idea. But until this can be mapped to and understood at the level of atoms and molecules, let’s not assume too much. Exactly how is a reference frame stored and expressed at the molecular level? What ions are involved, and in what quantity and in what arrangement?
Sooo 1 brain 2 cups 00:57
0:00 No, that's a cartoonish drawing of a brain.
As a longtime advocate of your work, I proposed legislation that forbids degrees in artificial intelligence without first having a doctorate in cognitive neuroscience.
Why?
We just better learn how to filer out hubris in these next generation "intelligent" machines, otherwise they will inevitably consider themselves gods.
Wrong read the book the big picture by sean carroll to understand the original purpose of life and emergence of intelligence, I have made a theory of cellular intelligence based on it
Based on prey predator model
Why is it wrong? How?