Deep Maths - machine learning and mathematics

[@michellepeterandrobin9181]

I just love at 41:36 when Marc basically says "Yeah well we spent a year working on the conjecture but it turns out the answer was given to us by the AI at the beginning"

[@bubbles_mc_rainbow]

I love the mix of confidence and curiosity they each displayed while sharing their portion of the paper. What a fun and exciting time they all seemed to have and this was so refreshing and engaging to listen to.

[@fitzounet14]

"if linear algebra can do it then surely you can do it" best sentence in the video !

[@catsaresocute650]

I don't think that's such a clear cut case.

[@markcolman4713]

As a maths Mattix graduate, more than half a lifetime ago, who is just beginning to study machine learning I found this video fascinating. Thank you.

[@jamieoglethorpe]

This raises some questions about neural networks: - we should expect them to justify their answers, e.g. give intelligible answers to Why probes. - can we extract heuristic information from them, e.g. what factors are important when Deep Zero assesses a game position? - can we convert a neural network to an expert system? - could the mathematician provide a hypothesis for the AI to test against the training data? This would be against the trained network or to pre-process the data during training. Taking the example of recognising the cat under a towel image, what is the essence of cattiness? My introspection is that the relationship between the nose, mouth, and eyes with the rest of the face are most important.

[@bubbles_mc_rainbow]

These questions kinda glitched my brain a little bit because of the thought put into them, especially the last one about what parameters actually *do* define catness. I would love to know as well!

[@reyansh5049]

The way of teaching all subjects in oxford university is just awesome and I love oxford from my heart

[@eliasbouhout1]

I've always been curious about applying machine learning to math problems, very cool to see the process, hopefully the methods used will be further refined in the near future

[@dereksisco4790]

Something I've been thinking about as well, kinda trying to play with it right now but keep hitting walls -_- lol

[@sophiac.700]

Some time stamps: 0:20 Alex Davies Machine learning and mathematics 11:35 Geordie Williamson Machine learning and the combinatorial invariance conjecture 21:21 Andras Juhasz Low dimensional topology and knot theory 30:21 Marc Lackenby Knot theory and machine learning 43:52 Panel

[@Nishantpatale9423]

Thanks

[@jasonkim1642]

Incredibly abstract yet informative!! Sincerely appreciate it!

[@davidwilkie9551]

"You just look at it" our very experienced, Oxford trained Math Professor told us, in the way intelligent people with the right aptitude for pictorial memory "communicate" (not) their ideas. Most of us are quickly weeded out if it doesn't happen easily. So: Looking at the world from Euler's POV is to see the Origin-all elements of e-Pi-i existence in line-of-sight superposition density-intensity, an orthogonal-normal picture-plane resonance of radial exponential and recursive i-reflection balance, ie Pi-bifurcation wave-particle axial-tangential => tangency, or unit circle type modular transverse trancendental log-antilog => crystal condensation sync-duration connectivity. Existence is a naturally occurring, inherent "Neural Network" which, interpreted at the ONE-Infinity unity-connection categorization is holographic consciousness, ..compounded in/of AM-FM interpenetrating Fluxion-Integral wave-packaging pure-math Actuality, to which this synthetic Neural Network as described, has the associated Conformal Field Condensation Correspondence, in functional e-Pi-i Principle, limited in con-vergent sequential Form @.dt 1-0-infinity probability i-reflection containment.

[@wayneqwele8847]

This is absolutely fascinating.

[@__________________________6910]

we need more free content on ml, ai, deep learning

[@mikaa2004]

Amazing!!! Thanks for providing it online.

[@ateljeayin]

So basically what he is saying on the 35:11 topic is that, the hyperbolic invariants DO have a signature of 4 dimenosional invariant by the aproximation of K/2 + some error.. so this function is fitting the data set with this error estimate and that this error with which is the aproximation fitted have been given to us by machine learning. What is the volume K and the injectivity K to the power of -3?

[@Self-Duality]

Extremely intriguing and stimulating…. closely considering the thoughts of these theorists 😎💭☘️💖

[@scientificpublishingreview2809]

41:23 I have submitted a paper in two journals, the reviews from both the journals responded saying I have used simple algebraic mathematics and theorems that are proposed by me and was rejected on basis of this but I see they were wrong and was a trend in journals to see sophisticated mathematics but with simplicity we can achieve bigger things like the formula achieved by Dr. Marc Lackenby.

[@asalamkamal6365]

I just started loving math and I want to start learning it how to start. and what to start with

[@yuanzhilee6405]

Pure maths isnt easy by all means

[@asalamkamal6365]

@@kaikyscot6968 you'r right

[@doctorbroker4690]

You can star with looking at flowers

[@asalamkamal6365]

Or sitting under an apple tree

[@asalamkamal6365]

@@kaikyscot6968 Don't be naive

[@elleguha9163]

What other attribution techniques did you try for establishing the conjecture for the Bruhat Graph?

[@Vyanjansharma]

Me also

[@SanKum7]

FYI, Bruhat means - Big/Huge/Enormus, from one of the Indian languages.

[@brazni]

This was fascinating and inspiring, great discussion. Thank you for making it available.

[@TheMISBlog]

Very Informative webinar, Thanks

[@MangDynasty]

42:20 - and are they gonna be blown away again when they realize the Cusp volume also matters?

[@BruinChang]

Will you invite Professor Nando De Freitas to give us talks in the future ?

[@alexander_stahl]

Very satisfying video. Thank you

[@guillaume6373]

this was awesome, thanks for the talk!

[@jatigre1]

All these brilliant minds combined cannot come to grasp the simple concept of rotating the Michelson Morley interferometer on the vertical axis.

[@aulalivre5481]

Thanks for sharing ✌

[@SwissOnZ]

Thank you.

[@JuanGabrielOyolaCardona]

Thanks for sharing 😃🇨🇴

[@anewbeginning277]

Are you from ClaMabia, Romania or Armenia?

[@bravosix5861]

thanks

[@z3h-c6y]

기대됩니다

[@ZainKhan-sm8gr]

나도 ㅋㅋ

[@oneday9star]

좋은 말씀^^

[@AIRPAX]

왜 이런게 뜨는거지?

[@gorizontout]

이게 다 우리의 미래이니끼

[@Swampy293]

nothing new if you are in the topic ;)

[@HernandezOmarCharlie]

Ok

[@googleservice5552]

Hi

[@seeker4430]

For the first time in history the British are doing for free

[@__________________________6910]

hummmmmmmm

[@SPVLaboratories]

@@__________________________6910 good name

[@__________________________6910]

@@SPVLaboratories noice

[@dologongpoloponobonotongpo235]

deep mafs

[@jose000]

🍪

[@TheGroundskeeper]

Here’s a riddle for you. How many smart people does it take to make a video unwatchable

[@notapplicable2515]

Haha, and here I thought the math guys had something to teach us ... ALL of this is just newton-raphson on log square loss ffs.

[@JohnSmith-bm6zg]

Boring