Professor Montanari, I attended your talk at Brown University and was wondering if you are interested in dynamical regimes for large two-layer neural networks in Reinforcement Learning? There is a serious theory practice gap in Deep RL.
Pretty cool results from Boston Dynamics using motion capture suit and trained using RL. The question is how well do the motions generalize to unseen behavior?
Pretty cool results from Boston Dynamics using motion capture suit and trained using RL. The question is how well do the motions generalize to unseen behavior?
instructions:
hanumakinds latest banger is probably the best thing to start your week with
www.youtube.com/watch?v=MbJ7...
Barto and Sutton receive the Turing award!
Amazing recognition for two kind, pioneering, and genuinely inspiring scientists. Congratulations!!
awards.acm.org/about/2024-t...
Seeing Andrew Barto at RLC last year was one of my favorite moments. Overjoyed to see him get the turing award (long overdue).
Reading this incredible paper today: The O.D.E. Method for Convergence of Stochastic Approximation and Reinforcement Learning. What a beautiful approach to proving convrgence of Q-learning.
In terms of absolute money, the poorer states are receiving ..
Read more at:
timesofindia.indiatimes.com/articleshow/...
ever wondered about fractional derivatives?
www.youtube.com/watch?v=2dwQ...
whats a "bella" ?
Interesting new take on intelligence: "Your cat is smarter than you are on certain things, and you're smarter than it on certain things". Intelligence is not a one-dimensional linear scale! www.forbes.com/sites/johnwe...
does anyone have good resources on the Hida
Malliavin derivative?
A man wearing surgical gloves gesturing with his hands with a screen in the background showing a skull with injuries on it
unpopular opinion: salunke was the real workhorse of CID, he carried daya and ACP
might just be teething issues that they eventually end up solving, text summarization is more or less "solved" ig
it is kind of remarkable how you can write down crazy jolty processes as an equation like this double well stochastic process
any CS departments doing personality hires next job season?
live picture of me looking going through my ICLR reviews
ever wondered why warmup the learning rate for deep learning? here's a cool paper by physicists arxiv.org/abs/2406.09405
so cute!!!
moo deng grad student
Andy Barto at #RLC
Magnolias, moon and mausoleum #DALLE
A cartoon sketch of a frog with its tongue sticking out
froggy
More Gentrification!! More destruction!!
All powerful Gentrifies!!
Destructive, unstoppable!
