We looked inside some of the tweets by @mir_k and here's what we found interesting.
Inside 100 Tweets
Check out a novel #ReinforcementLearning environment where agents aim to master the world’s most popular sport—football! The Google Research Football Environment includes benchmarks & progressive RL training scenarios, and is available in open source beta→http://goo.gle/2Mz8IqG
Ever wanted to train a T-Rex to play soccer? Check out our new paper on learning composable motor skills with multiplicative compositional policies (MCP): http://xbpeng.github.io/projects/MCP/ Big thanks to all my collaborators: @mmmbchang, Grace Zhang, @pabbeel, @svlevine
🐣 New Tutorial, open-source code & demo! Building a SOTA Conversational AI with transfer learning & OpenAI GPT models -Code/pretrained model from our NeurIPS 2018 ConvAI2 competition model, SOTA on automatic track -Detailed Tutorial w. code -Cool demo http://convai.huggingface.co 👇
Straight from Google I/O, I'm happy to announce that @Arm and Google @TensorFlow are joining forces to accelerate TinyML. We will be merging uTensor, the first open source embedded inference project, with TensorFlow Lite Micro. https://os.mbed.com/blog/entry/uTensor-and-Tensor-Flow-Announcement/ … #tensorflow #tinyml #mbed
Instead of computing the best racing line, I computed a huge grid of the "value" of being in any given position / angle, which lets a robot select the best action just by looking up all moves in a table. Bottom: table for the car's current heading. It recovers from any position.
This year, we have 3 papers at RSS, all on robotic RL: learning to walk, tool use w/ model-based RL, and learning from pixels w/o hand-designed rewards. All w/o using sim to train and on real robots https://arxiv.org/abs/1812.11103 https://arxiv.org/abs/1904.05538 https://arxiv.org/abs/1904.07854
Our new paper, “Reinforcement Learning, Fast and Slow”, reviews recent techniques in deep RL that narrow the gap in learning speed between humans and agents, & demonstrate an interplay between fast and slow learning w/ parallels in animal/human cognition: https://www.cell.com/trends/cognitive-sciences/fulltext/S1364-6613(19)30061-0 …