18 сағат бұрын
Пікірлер
@abhishekshakya6072 4 күн бұрын
It is probably dumb or anyone else have trouble understanding folding analogy at 12:15? Is he suggesting that the planes are superimposed with one top of other? Or is he suggesting that sum of figure a and figure b lead to figure c? Can anyone help me in understanding it?
@yannickzelle5796 4 күн бұрын
Thank you for your talk. I found it extremly interesting. I have some comment on your statement that simplicity is implied by utility : Differential Equations are very useful in describing our world, however they are at least in my mind not simple and to most people also not familiar. I would love to discuss about it !
@JefferyRoss-yr7ve 5 күн бұрын
Define these letters already
@Gunni1972 6 күн бұрын
Which A.I. came up with that Theory? They usually hide inside Brains.
@karmavil4034 7 күн бұрын
Best 25 minutes of my life.. Physician telling me a language model is just a chatbot refined.. I can do that. Let's go!
@adiankunda204 7 күн бұрын
I really want to contribute to physics like Prof. Elliot😅, dude is phenomenal.
@sinankupeli628 8 күн бұрын
one of the best suggestions of the algorithm. there is a phrase widely used in education circles nowadays: 'Learning how to learn' and it is often criticised as human babies are already born with the ability to learn. But in the case of machines I suspect that is the way to go. They lack the genetic encoding we embedded in our biological systems for so long. Maybe we should treat these early machine learning models as their DNA?
@NUSORCA 9 күн бұрын
The mental features discoursed of as the analytical are in themselves but little susceptible of analysis. --- Edgar Allan Poe To put it simply, it’s hard to understand why prodigies are the way they are
@non-inertialobserver946 9 күн бұрын
Great talk!!
@TechnologyRules 9 күн бұрын
Why does he have a giant stick, is he blind?
@codybarton2090 12 күн бұрын
Yea it’s great if ur trying to use a system In a system to garner traction
@theodorranebo2909 12 күн бұрын
Can change ringing be done with electric motors instead of bellringers?
@johnmorrison3465 12 күн бұрын
i find it interesting special relativity wasn't an example of needing to fit the data. oh wait, i'm not, because there was no underlying data
@Dosteyboi 14 күн бұрын
Along with being so intelligent, he is such a likable guy. In the 4 minutes I've heard him talk, I already like the guy
@orbatos 14 күн бұрын
I was pretty surprised to see this not actually purpose much of anything other than using tools to analyse patterns, three same tools that have been in use for decades. Is this a venture pitch? Throwing more processing at it helps, but doesn't "solve" anything on its own.
@MarkAhlquist 14 күн бұрын
Your cat doesn't teleport?
@dendrites 15 күн бұрын
Most likely a biological neural network
@kalla103 15 күн бұрын
yooo, i'm so glad i came across this! i've been thinking about how neural networks can teach us about our own thinking and pattern finding; i'm glad there is discussion about it
@kalla103 15 күн бұрын
okay it's not about what i initially thought, but whoa. this polymath approach sounds excellent. i feel it's similar to how people who study many different fields can be quicker to grasp a novel problem
@scaomath 15 күн бұрын
33:16 Mark my words. There won't be any foundational-level model can achieve 5-digit of accuracy like the finite difference does for PDE, which was popularized three hundred years ago by Euler. Using the model alone (without the help of non-blackbox outer algorithms or second-order optimizer), no matter you have 1000 billion params, or what, never. 1000 years later our AI overlords will still use finite difference (maybe the BDF table will be learned by blackbox).
@Sairfecht 15 күн бұрын
What happens if it’s trained on Schrödinger’s cat videos?
@mosarof_cp 15 күн бұрын
What makes him great? I don't found any of good quality between him!
@echoeversky 16 күн бұрын
Well more fun than AI churning out what are the most lethal substances. Modern nerve agent analog were in the outputs.
@mubarakbutt8080 16 күн бұрын
Very nice and interesting
@0MVR_0 16 күн бұрын
Folding is an analogue to reducing dimensional complexity.
@dougclendening5896 16 күн бұрын
So wisdom is better than raw intellect. Got it. The more we know, the more it enhances anything else we do and think about. Not really a mystery, but nice to see the theory and it being proven. ML has always felt wrong to me because of the "toddler" he describes. Now that ai can know everything that the entire human race knows, of course it will start solving problems we can't think to solve alone with our own minds and data within. Who built our ai?
@primenumberbuster404 16 күн бұрын
Damn wholesome mathematician.
@staticinteger 16 күн бұрын
Wow this is incredible and sort of confirms some thoughts I’ve had about neural networks and the compression of knowledge.
@thechessmaster9291 17 күн бұрын
Go f yourself and learn to explain things in an orderly manner . You are a disgrace to humanity ...
@drewkelly1955 18 күн бұрын
Why is everyone recommended this video?
@palimondo 17 күн бұрын
Mostly because of the folding analogy around 12:00 that helps build a good intuition about how chaining the layers of the neural network from perspective of a single neuron does produce a complex function that can learn to approximate complex relationships. The main topic of the presentation is fine, but more like an early sketch of an idea how we might extract new simplified scientific models from LLM - but the prompt method doesn’t seem to be scalable enough for practical application. The core insight seems to valid, though.
@jAvViRolDaN 18 күн бұрын
Interesting topic and well put presentation, however the speaker should practice a bit more its communication skills (not saying "I mean" or "like" every 2s) and stop reading the damn PowerPoint - I bet most viewers can read-. Again, it is a great topic and a good presentation. I will read his previous research after watching this.
@MrWolynski 19 күн бұрын
Planets are older and dead stars.
@Xsiondu 19 күн бұрын
Here's an April fools episode idea. Tales from the dark side episode where you discuss the extremely creative solutions you had to come up with while trying to do chase films on some UFO test object. Deliver it in a straight face with digressions on how it would be cool if / or how you are waiting to see the technology make it's way into the civilian market ect . For the photos have blurry images of birds or maybe say that the reason pictures are blurry is because UFOs are just blurry. Even when in the hangar
@mathematik1865 19 күн бұрын
Quote (16:40): State of the art for symbolic regression... 25 days later a paper was released where so called KAN's where used to do symbolic regression, and I am pretty sure that this will be the state of the art. I know it was used only on small datasets and has some other flaws, but this is not worth talking about since we will make it work. They also refrence Miles Cranmer.
@piotr780 17 күн бұрын
KANs does not scale well
@JustJOGGIN-qb9yo 20 күн бұрын
The AI field is hallucinating.
@RandomNooby 21 күн бұрын
Biological or silicon network?
@petercoool 21 күн бұрын
Wow; what a bridge from physics to number groups😊
@zef3k 21 күн бұрын
Great presentation! My main takeaway is that we need a more unified approach to neural network models. Interoperability is important and can substitute for or even supercede the quality increase of pre-training.
@bogpetre 22 күн бұрын
even at 720p the quality is too poor to make out important details from his figures. Makes this difficult to follow.
@TheMap1997 22 күн бұрын
She only got this award because Ukraine got invaded. If nothing happens to Ukraine, the award would go to an even more outstanding mathematician
@musicmarlene 23 күн бұрын
Loved this video. Now I know a little more about you, Barry, and who my son, William Stein, was talking about. You and Gretchen are amazing people.
@colmaniot6331 23 күн бұрын
Grate path to walk on .. wish luck to the lecturer and hiss fellow researches
@philopolymath 24 күн бұрын
I'll bet my Bitcoin it won't.
@5ty717 24 күн бұрын
Eric Lerner has spoken to H Alfven Wave and Plasma instabilities-stabilities along Magnetic field lines of the sun and stars in general for decades. It’s not an association that some will dare to make yet I think he deserves recognition for his huge contributions since VNadi and others.
@gillesdeleuze6083 24 күн бұрын
They could put the questions!
@ardaozkut1089 25 күн бұрын
Is the part in 12.40 just convolution or am I just dreaming?
@Daniel-qr6sx 25 күн бұрын
this was very interresting
@caduaraujo331 25 күн бұрын
unknowingly to him, this man was one my teachers in maths. God bless him.
@googleyoutubechannel8554 25 күн бұрын
This guy has obviously not spent enough time with LLMs or transformers, the ways in which they are 'dumb' and _don't_ make very obvious connections is far more astounding than the ones they do.... I wouldn't trust that NNs in the way we've formulated have any hope of making any type of interesting connection that one of 8 billion humans has not thought of already.
@seanmcdonough8815 24 күн бұрын
He's doesn't seem to be extolling those achievements. More like how a shitty NN can do water wiggles better than the special effects team making water for movies. (Just the most fundamental part of water, but if it does that with far less gas, that's an achievement. It's actually what the champion info compressor does [black holes]
@googleyoutubechannel8554 23 күн бұрын
@@seanmcdonough8815 He does initially, but I watched the rest. You are right. He has a point about simple-curve-fitting-NNs. I approve of ignoring Navier Stokes nonsense, rock on you self-water-wiggling algos, rock on...
@kamilziemian995 25 күн бұрын
What a great talk.
@pabloe1802 26 күн бұрын
show me the code