Uncovering the Mysteries of Category Theory!

Patreon: / mlst
Discord: / discord
Twitter: / mlstreettalk
The Math Wizards of Cambridge Computer Lab: Uncovering the Mysteries of Category Theory!
Welcome back everyone! A few weeks ago, I had an incredible day at the Cambridge computer laboratory, where I met with Dr. Petar Veličković (DeepMind), Andrea Deec, a PhD student at Mila, and Dr. Andrew Dudzik (DeepMind). We explored the fascinating world of mathematical structures and their applications in various fields.
In this video, you'll get a tour of the Cambridge University Computing Lab, and a glimpse of our conversation about category theory, a branch of mathematics that focuses on abstract structures, their relationships, and transformations. They discussed lattices, groupoids, and how these structures can be interesting from a categorical perspective.
We also touched on the concept of categorification, which involves uplifting problems to higher categorical levels, emphasizing the importance of understanding the problem at hand before attempting to categorify it. Petar and Andrew shared their preference for working with algebra first to gain a deeper understanding of the problem before attempting to generalize and lift it to higher categorical levels.
So, buckle up and get ready to dive into the world of category theory and mathematical structures with our brilliant guests. And don't forget to check out our other videos from this day, we released the video with a leetcode challenge a few weeks ago.
We will release the main show with Petar, Andrew and Andrea on algorithmic reasoning and their interview in the coming weeks.
Dr. Petar Veličković
petar-v.com/
Dr. Andrew Dudzik
/ andrew-dudzik-222789142
Andreea Deac
andreeadeac22....
Video by Dr. Tim Scarfe - Co-founder XRAI Glass xrai.glass

[@nonchai]

Just a wee thanks for your great channel - I particularly like that you ( rightly ) include a fair bit of relevant analytic philosophy, as so much of the issues in AI - now and upcoming - are in that terrain and not really answerable (I'd say ) via experiment or empirically. I reccomend Richard Browns YT channel for his series of interviews with many philosophers of mind. Maybe worth you interviewing a few philosophers too ? Its timely!

[@MachineLearningStreetTalk]

Thanks Daniel! I have a few philosophers booked in, who I suspect you will enjoy!

[@astroid-ws4py]

More diagrams are needed to explain thus subject. Remember Einstein's quote: "if you can't explain a subject to a 5 year old you do not really understand the subject", Frankly many subjects of Mathematics can be explained intuitively and clearly, Be it Complex Numbers, Epsilon-Delta definition of limit, Lebesgue integral and much more given suffcient amounts of drawings/animations.

[@pretzelboi64]

That Einstein quote is brain dead and straight up invalid. If it truly was said by Einstein, it was definitely in a moment of complete senility. There are things that are so technically complex that you cannot explain them in 5 minutes or to a 5-year-old and most definitely not to a 5-year-old in 5 minutes. People wouldn't go to college, have PhDs, or technical jobs otherwise. Any attempt at explaining those things that simply would result in you not actually understanding them at all. Good luck explaining something like the Levi-Civita connection or the covariant derivative to a 5-year-old in 5 minutes. Your examples are so easy it's laughable. Complex numbers are just glorified 2D vectors so that's not exactly a challenge but some things are not that simple. You can't just teleport to the leaves of the tree of knowledge without walking along the branches first.

[@AbdennacerAyeb]

We appreciate the efforts you are making to enlight us with such knowledge. Greeting from Tunisia. We follow you from all over the world.

[@vonehrenheim]

Thank you for this! We had the opportunity to have Petar present some of his work at EQT Motherbrain recently and he is really amazing. Very inspiring and relevant to our research! 🦄

[@angelxrj6707]

I am glad you keep track of this amazing research path towards AGI along with your interviews and random chats.

[@muhokutan4772]

Always great to hear Petar! When does the full version come to Spotify?

[@oncedidactic]

Awesome snippets, thanks for posting! Gives me vibes this will later be like a documentary on the historical AI inflection point

[@5pp000]

4:58 No, that's not punch cards, it's paper tape! The small holes don't carry information; they're just for the teeth of the drive wheel to fit into.

[@avi3681]

By non-cartesian product and non-cocartesian coproduct does he mean non-categorical product and non-categorical coproduct?

[@kamilziemian995]

Interesting discussion.

[@curtkennedy4170]

Trying to understand how Category Theory relates to Machine Learning. Any idea? Would be cool to see if it had any application.

[@Julie-do3oi]

To be categorical do you also have to be factorial to become rhetorical for a basic understanding of interest 2 = interesting hypothetically or would all of that just be allegorical borderline fictitious in retrospect subjectively speaking becoming enlightened to comprehensive awareness so you don't have to be apprehensive to being pensive generally speaking????

[@invictus327]

The only real question remains: is the door to the Mathematical Laboratory actually the door of the TARDIS?

[@vslaykovsky]

6:48 is the crown intentional? :)

[@InquilineKea]

What are alternatives to cocartesians?

[@neverhavingfun]

👁▲

[@rajuaditya1914]

Awesome stuff as always!