This account reflects my current interests: #AI #RL, #multiagent, #deeplearning, #ReinforcementLearning
France, the Alps
Joined February 2011
- Tweets1.4K
- Following393
- Followers143
- Likes1.4K
Marc-Philippe HUGET retweeted
We are pleased to announce that the 2020 @NeurIPSConf MineRL Competition has officially launched! This is an exciting collaboration between @mldcmu, @OpenAI, @PreferredNet, @AIcrowdHQ, @MSFTResearch, and @DeepMind. Check it out and sign up here: minerl.io/competition
Marc-Philippe HUGET retweeted
Slime Volleyball Gym Environment
A simple gym environment for testing single and multi-agent reinforcement learning algorithms. Includes a few self-play training examples in python.
github.com/hardmaru/slimevol…
Marc-Philippe HUGET retweeted
Our paper (arxiv.org/abs/1905.01072) on deep residual reinforcement learning has been nominated for a Best Paper award at AAMAS!
Well done to @ShangtongZhang, @WendelinBoehmer, and @shimon8282 as they have been nominated for the best paper award at @AAMAS2020c. Read more here:
cs.ox.ac.uk/news/1799-full.h…
Marc-Philippe HUGET retweeted
OpenAI's multi-agent team is currently hiring research scientists! Interested in multiple, interacting agents, AI-generating algorithms, open-ended algorithms, automatically generating training environments, deep RL, and more? Please apply! openai.com/jobs
Marc-Philippe HUGET retweeted
Want to learn about meta-learning? Lecture videos for CS330 are now online!
youtube.com/playlist?list=PL…
Topics incl. MTL, few-shot learning, Bayesian meta-learning, lifelong learning, meta-RL & more:
cs330.stanford.edu
+ 3 guest lectures from Kate Rakelly, @svlevine, @jeffclune
Marc-Philippe HUGET retweeted
We're standardizing OpenAI's deep learning framework on PyTorch to increase our research productivity at scale on GPUs (and have just released a PyTorch version of Spinning Up in Deep RL): openai.com/blog/openai-pytor…
Marc-Philippe HUGET retweeted
Over 300% funded for the Neural Network from Scratch book @daniel_kukiela and I are writing! Over a hundred people in the draft doc
Special shout out for all the people contributing to the doc too. Got some grammar experts shaping things up nicely :) <3
kickstarter.com/projects/sen…
Marc-Philippe HUGET retweeted
Deep RL agents are data hungry and often learn task-specific representations. Our model learns object-centric abstractions from raw videos. This enables highly data-efficient RL and structured exploration. arxiv.org/abs/1906.11883
Marc-Philippe HUGET retweeted
We’re releasing a Neural MMO — a massively multiagent game environment that supports numerous populations of agents: blog.openai.com/neural-mmo/
- Code: github.com/openai/neural-mmo
- 3D Client: github.com/jsuarez5341/neura…
Marc-Philippe HUGET retweeted
Model-Based Reinforcement Learning for Atari
They show that the simple, iterative method of learning a world model is enough to get ~SOTA data-efficient results on various Atari games. Only 100K interactions between agent and game (2hrs of realtime play)
arxiv.org/abs/1903.00374
Marc-Philippe HUGET retweeted
Evolving neural net controllers for a team of self-organizing robots. This 2010 work used self-play tournament selection to evolve a team of cooperative agents that plays a simulated soccer game. Interesting strategies emerged after only 500 generations. goo.gl/XvjJNN
Marc-Philippe HUGET retweeted
Multi-Agent Soccer Environment from DeepMind
Training agents to get the ball over to the other side is really my thing.
GitHub: goo.gl/aBFoqM
Paper: goo.gl/eHV4kz
Marc-Philippe HUGET retweeted
One more step in the journey of making cooperative multi-agent RL reproducible 🧐. Considering the number of papers that use one-off toy environments in this field, this benchmark environment was long overdue..
Excited about @StarCraft? Let's use it for cooperative multi-agent RL!
Introducing #SMAC: The StarCraft Multi-Agent Challenge, a diverse set of decentralised unit-level micro challenges.
Preprint: arxiv.org/abs/1902.04043
Blog: whirl.cs.ox.ac.uk/blog/smac
Code: github.com/oxwhirl/smac
Marc-Philippe HUGET retweeted
Creating a Zoo of Atari-Playing Agents to Catalyze the Understanding of Deep Reinforcement Learning. Great work led by Joel Lehman @joelbot3000, w/ many excellent collaborators from Uber AI Labs, OpenAI, & Google Brain. Blog: eng.uber.com/atari-zoo-deep-… Code: github.com/uber-research/ata…