Thiago D. Simão

A huge thanks to my collaborators, Thiago D. Simao (@tdsimao) and Nils Jansen (@thesuperjansen), as well as my advisor, Ufuk Topcu, for his support during our collaboration. Check out our work: Paper: openreview.net/forum?id=f3QR9…) Code: github.com/cevahir-koprul…

PhD position RL. Are you eager to work on #AI safety, unleashing the potential of #RL? I'm #hiring a PhD student to work on #RL while exploring relevant areas of #ML to tackle challenges in partial observability and safety. @Data_AI_TUe @TUeindhoven jobs.tue.nl/en/vacancy/phd…

New Assistant Professor position in our group: jobs.tue.nl/en/vacancy/ass…

Our Database group (DB) is looking for a highly motivated university lecturer in Data Management. A great opportunity for candidates with a passion for teaching! jobs.tue.nl/en/vacancy/uni…

📢 The submission deadline is approaching! This is the 2nd CFP for #AAMAS2024: We invite you to submit your best work in the area of agents and multiagent systems—and to join us in Auckland next May! Abstract submission: 2 Oct 2023, full paper: 9 Oct 2023 aamas2024-conference.auckland.ac.nz/calls/call-for…

This week, we are presenting "Risk-aware Curriculum Generation for Heavy-tailed Task Distributions" at the Conference on Uncertainty in Artificial Intelligence (UAI) 2023! [1/5]

To learn more about our work, visit our poster at UAI 2023 (on Tuesday, August 1, 4:30 pm EDT) or read our paper at proceedings.mlr.press/v216/koprulu23……! Joint work with @tdsimao, @thesuperjansen, and Ufuk Topcu. [5/5]

A population density map of Brasil for the last day of the year. May the new year bring joy and adventure in equal measure! 🎇 And more rayverse updates. 😎 #rayshader adventures, an #rstats tale

Carmel Freeman is now on Spotify. open.spotify.com/artist/2ZQwdb7…

Did you ever ask yourself how to learn a model of a system that is changing? Or did you work with interval MDPs but had no clue where the intervals came from? Or did you wonder which models are robust against uncertainty? We also did. Check out our paper at @NeurIPSConf 22.

Nils Jansen (@thesuperjansen) will talk about the applicability of shields via tight integration with state-of-the-art deep RL. Come listen! 👏 5/9

Paper “Robust Anytime Learning of Markov Decision Processes” accepted at #NeurIPS2022. Big congratulations to my PhD student Marnix Suilen and my postdoc Thiago D. Simão. Great collaboration with Dave Parker! arxiv.org/abs/2205.15827 #MachineLearning #verification @NeurIPSConf

"Safety-constrained reinforcement learning with a distributional safety critic" by Qisong Yang, Thiago D. Simão, Simon H. Tindemans & Matthijs T. J. Spaan (link.springer.com/article/10.100…)

🔀Do you prefer to go the long, but easy way round, or the short but more challenging path? Thom Badings and @thesuperjansen explore how a #UAV could make this decision to ensure safe deliveries

the Bézier video is now out!! 🎉 I've spent the last month working on this - I hope you'll like my audiovisual exploration of these math noodles! 💖 youtube.com/watch?v=aVwxzD…

The PRL Workshop – "Bridging the Gap Between AI Planning and Reinforcement Learning" is happening on Aug 5 and 6. We have 5 invited talks, and 25 accepted papers that will be contributed talks or posters. prl-theworkshop.github.io

Lecturer Stefan Hugtenburg @tudelft came up with a 'gamified' knowledge platform to motivate the increasing number of IT students of different learning years to share more knowledge. #computerscience @EEMCS_TUD #TUDelftAI bit.ly/tud-ai-platform

Paper: ifaamas.org/Proceedings/aa… Code: github.com/AlgTUDelft/Alw… More: tdsimao.github.io/publications/S…

Here is a short intro (~90 sec) to our #AAMAS paper on how to develop reinforcement learning agents that respect the safety constraints during training. Joint work with Nils Jansen and Matthijs Spaan. @TUDelft_AI @aamas2021 @tudelft @Radboud_Uni @delftswtech @EEMCS_TUD #AI