postdoc in robot learning @ university of oxford previously @DFKI @ias_tudarmstadt @TUDarmstad @DeepMind intern, @CMRSurgical, @Cambridge_Eng
Oxford, United Kingdom
Joined January 2012
- Tweets107
- Following957
- Followers801
- Likes2K
Joe Watson retweeted
How well do world models reuse what they've already learnt, without forgetting it? 🤔
I am excited to introduce our new work that looks into this! 💡
Website: object814.github.io/Composit…
Paper: arxiv.org/abs/2609.22055
1 / N👇
#robotics #machinelearning #worldmodel
Check out Tianyou's first paper on cross-environment imitation learning using latent action spaces! 💪🦾
I am excited to introduce GLAM (Grounded Latent-Action World Models), our latest effort to enable robot learning on heterogeneous data using world models.
Website + videos: viccccciv.github.io/glam/
Paper: arxiv.org/abs/2606.21672 1 / N 🧵👇
#robotics #machinelearning #worldmodel
Stoked to be in Seoul 🇰🇷 for #ICML2026! Happy to chat all things Bayesian, reinforcement learning, robot learning and imitation learning this week 👇
Come chat at 10.30am tomorrow Korean time 🇰🇷!
Happy to share that @JoeMWatson and I are finally co-authors (with @IngmarPosner and @Jan_R_Peters). Our paper about posterior sampling for reinforcement learning with Gaussian processes (GP-PSRL) is appearing at #ICML2026 this week (some info below)
Feedback is most welcome
Joe Watson retweeted
We've lost an absolute giant today. RIP Dimitri Bertsekas. His probability and optimization books got me through my masters. Massive loss for the MIT community and the field.
Congrats to this year's pioneers, the selection was extremely challenging!
Please join us! Room 211 in the Convention Center at 12
🌍World models can play an important role towards building general agents,
but what should be their role in decision-making?🕹️
@JoeMWatson and I are organizing a Social @iclr_conf on this topic🎙️
🗓️Feel free to join the conversation on Friday 24th April, at noon!!
Come chat to us!
Headed to Rio for #ICLR 🇧🇷 come say hi at our poster!
XQC: a principled look at critic optimization. BN+WN+cross-entropy → condition numbers orders of magnitude smaller than baselines. SOTA on 70 continuous ctrl tasks w/ 4.5× less params.
📅 Thu, 1030–1300
📍 Pavilion 4, #4518
Joe Watson retweeted
🎉 Really excited, our paper "XQC: Well-conditioned Optimization Accelerates Deep Reinforcement Learning" has been accepted at #ICLR2026 .
If you are interested in reinforcement learning, sample-efficiency, compute-efficiency go check it out. See you in Rio!
🚀 New preprint! Introducing XQC— a simple, well-conditioned actor-critic that achieves SOTA sample efficiency in #RL
✅ ~4.5× fewer parameters than SimbaV2
✅ Scales to vision-based RL
👉 arxiv.org/pdf/2509.25174
Thanks to Florian Vogt @JoeMWatson @Jan_R_Peters
@Hessian_AI @DFKI
Joe Watson retweeted
🚀 New preprint! Introducing XQC— a simple, well-conditioned actor-critic that achieves SOTA sample efficiency in #RL
✅ ~4.5× fewer parameters than SimbaV2
✅ Scales to vision-based RL
👉 arxiv.org/pdf/2509.25174
Thanks to Florian Vogt @JoeMWatson @Jan_R_Peters
@Hessian_AI @DFKI
Joe Watson retweeted
Could geometric cues help improve goal inference in robotics?
We explore this question at #RLDM today | Spot 86.
Stop by if you're curious about bridging motion planning and intent prediction.
Scaling sample-efficient RL often relies on artisanal architectures (extra LayerNorm, etc)
Daniel and Florian found a major issue with vanilla MLPs: the larger network weights slow optimization, so simply adding weight norm unlocks sample efficiency for much harder tasks! 🚀
🚀 New preprint "Scaling Off-Policy Reinforcement Learning with Batch and Weight Normalization"🤖
We propose CrossQ+WN, a simple yet powerful off-policy RL for more sample-efficiency and scalability to higher update-to-data ratios. 🧵arxiv.org/abs/2502.07523v2
#RL @ias_tudarmstadt
We recently collaborated with @ABBgroupnews to survey the recent literature on physics-based machine learning ⚡️, with a focus on knowledge- and data-driven inductive biases 🔭
Check out the survey here 👇 and please reach out if we missed anything
arxiv.org/abs/2408.09840
This was a big collaboration, with the help of @theo_grune82772, @an_thai_le, @kay_hansel, @AHendawy19, @OlegArenz, @CarloDeramo , @MilesCranmer and @Jan_R_Peters, and Chen, Oliver, Will, Fabian, Tanmay and Martin who aren't on twitter
Active learning! 🔥Entropy-regularized control! ⚡️Sequential Monte Carlo! 🎲 Make sure to check it out
#ICML2024
@sahel_iqbal is presenting our paper "Nesting Particle Filters for Experimental Design in Dynamical Systems" next week. A novel way to amortize sequential Bayesian experimental design:
arxiv.org/pdf/2402.07868
Jointly with @AdrienCorenflos and @simosarkka
@FCAI_fi
1/*
I’m presenting CSIL at the poster session this morning, come find me at #1906! #NeurIPS2023
Sandy and I also open-sourced the implementation a few weeks back, you can find it at
github.com/google-deepmind/c…
Excited to share my internship project from my time at @DeepMind, looking at sample-efficient imitation learning using entropy-regularized reinforcement learning
TL;DR: do behavioral cloning (BC), get inverse reinforcement learning (IRL) for free! [1/6]
CSIL has been accepted at @NeurIPSConf as a spotlight! ✨
Big thanks to my internship hosts Sandy and Nicolas at @GoogleDeepMind Robotics
We hope to share the code in the near future 🤖
Excited to share my internship project from my time at @DeepMind, looking at sample-efficient imitation learning using entropy-regularized reinforcement learning
TL;DR: do behavioral cloning (BC), get inverse reinforcement learning (IRL) for free! [1/6]
Excited to share my internship project from my time at @DeepMind, looking at sample-efficient imitation learning using entropy-regularized reinforcement learning
TL;DR: do behavioral cloning (BC), get inverse reinforcement learning (IRL) for free! [1/6]
This project was a collaboration with my internship hosts, Sandy Huang and Nicolas Heess!
For more information and results, check out
the project page arxiv.org/abs/2305.16498
the paper arxiv.org/abs/2305.16498