📢 Save the date! Join us on 12th May 2021 to meet our future leaders in data science and hear about their cutting edge research. Full programme & registration info to follow. Find out more about our Centre for Doctoral Training 👉 bit.ly/2Oi15Xi

Don't you just hate it when people drive up your bum! Need to up that optimal distance but nice target tracking eg using soft actor critic, a variational autoencoder and the #donkeycar Unity sim. Tensorboard graphs not looking healthy though, more work needed! #RL #madewithunity

Join us on 12th May as we dive into the world of #conservation tech, an exciting area focussing on utilising #AI and #MachineLearning to help the world’s plants and animals. bit.ly/3o4umSH

Shout out to BOSH! powering our come dine with me efforts #vegan #Foodie #comedinewithme

Very pleased with our new survey boat that we picked up today. Will be used for our collaborative Engineering & Marine Science dolphin research off Northumberland. Many thanks to Nick Wright for making it happen. Matt Sharpe Cameron @KN_Richardson Georgia Atkinson Sciences at Newcastle

My new little man, Freddy. #whippets #DogsofTwittter #30thbirthday #frommetome

🇪🇸 -25C snowstorm 🇺🇸 Texas deep freeze 🇦🇺 Historic floods in New South Wales 🇨🇦 Hundreds of sudden deaths in 49C heatwave 🇺🇸 ‘Monster’ wildfires 🇩🇪 & 🇨🇳 Hundreds dead in flash floods 🇬🇷 & 🇹🇷 'Apocalyptic' wildfires A year of climate disasters: youtu.be/xeY9miTqzEA

We’ve acquired the MuJoCo physics simulator (mujoco.org) and are making it free for all, to support research everywhere. MuJoCo is a fast, powerful, easy-to-use, and soon to be open-source simulation tool, designed for robotics research: dpmd.ai/mujoco-blog

It is critical for Reinforcement Learning (RL) practitioners to properly evaluate and compare results. I wrote a visual explanation about the tools used by the rliable library (NeurIPS Oral): araffin.github.io/post/rliable/ 🧵⬇️

The 4th & final #Reith lecture on Living with #AI from Prof. Stuart Russell last night, in our gorgeous venue The Catalyst. Listen to BBC Radio 4 22nd December 🤙🏽

Fantastic episode of Stuart Russell’s Reith Lectures, recorded at National Innovation Centre for Data bbc.co.uk/sounds/play/m0… Particularly fun to hear friends and colleagues @rsfrankl Paul Watson Jen Wood Kirsten Crane in the Q&A afterwards

Recurrent PPO (aka PPO LSTM) implementation, one of our most requested feature, is now on SB3 Contrib master branch =)! It was benchmarked against PPO with frame-stack on a set of environments where velocity was masked (see below for more details). github.com/Stable-Baselin…

Fantastic talk at King's College London from Oussama Khatib of Stanford University #robotics

Poster-ing @ #SMM2022 🐋

There are places where times stops & you truly are "in the moment"... #Magical #mindful Earlier today in the #NorthSea off the #Northumberland coast Thanks to Billy Shiel Nick Wright Newcastle University Damian Beeley

Fantastic being out on the water yesterday, two separate encounters with white-beaked dolphin pods, and a minke whale to top it off - just the boost I needed to start writing thesis chapter 3 of 7 on #sim2real

A different kind of morning, filming at Blyth Marine Station for an upcoming series profiling some of the amazing work taking place across Faculty of Science, Agriculture & Engineering Also ft Freddy Whippet, who tried very hard to be 'helpful' all morning

Amazing to see our NDD20 dataset used by Meta in their ‘Segment Anything’ paper 🐬 Can’t wait to share our marine RL data generation environment very soon! Segment Anything: lnkd.in/erWSsDy7 NDD20: lnkd.in/ea6H6bM7

Hyperparameters can make or break your ML model. But who has time for endless trial and error or manual guesswork? 🤔 Discover my visual guide to automatic hyperparameter tuning so you can spend more time on important tasks, like napping 🦥. Blog: araffin.github.io/post/hyperpara… 🧵⬇️

Fantastic time at IEEE Conference on Games in Boston last week, presenting our DRL Gym wrapped Unity simulation SWiMM DEEPeR. IEEE Conference on Games thank you for the brilliant organisation, content and generous catering! CDT in Cloud Computing for Big Data thank you for sending us ! Look forward to sharing the paper