Our paper: Decision-Point Guided Safe Policy Improvement
We show that a simple approach to learn safe RL policies can outperform most offline RL methods. (+theoretical guarantees!)

How? Just allow the state-actions that have been seen enough times! 🤯

arxiv.org/abs/2410.09361

Going to my first ever AISTATS in Thailand this year! 🎉🌴

Wow this is amazing! Thanks for sharing!

The call for papers for RLC is now up! Abstract deadline of 2/14, submission deadline of 2/21!
Please help us spread the word.
rl-conference.cc/callforpaper...

NEW: we have an exciting opportunity for a tenure-track professor at the #KempnerInstitute and the John A. Paulson School of Engineering and Applied Sciences (SEAS). Read the full description & apply today: academicpositions.harvard.edu/postings/14362
#ML #AI

I was today years old when I realized that when people use log-sum-exp instead of the softmax function in the soft-Bellman operator, they are probably going for the mellowmax function!

How are public comments a thing?? @iclr-conf.bsky.social

I can see why they can help get feedback from the crowds, but then why even have a double-blind process?

The notes are great! Thank you!

Would be cool to be included! (work on RL in healthcare)..