maggieliuzzi/reinforcement_learning

A mix of resources related to Reinforcement Learning.

Reinforcement Learning (RL)

Fundamentals

• David Silver's UCL Course   https://www.youtube.com/playlist?list=PLbWDNovNB5mqFBgq7i3MY6Ui4zudcvNFJ
• LazyProgrammer's Udemy Course   https://www.udemy.com/course/artificial-intelligence-reinforcement-learning-in-python

Deep Reinforcement Learning (Deep RL)

• Deepmind & UCL's Advanced Deep Learning and Reinforcement Learning course   https://www.youtube.com/playlist?list=PLqYmG7hTraZDNJre23vqCGIVpfZ_K2RZs
• LazyProgrammer's Udemy Courses   https://www.udemy.com/course/deep-reinforcement-learning-in-python   https://www.udemy.com/course/cutting-edge-artificial-intelligence

Latest Trends

• Meta Learning

  • Zero-Shot Learning
  • One-Shot Learning
  • Few-Shot Learning

• Imitation Learning

  • Inverse Reinforcement Learning (IRL)

• Self-Supervised Learning

• Multi-Objective Reinforcement Learning (MORL)

• Distributional Reinforcement Learning

• Network Pruning/ Policy Pruning for Deep RL

   Other Concepts

• Curriculum Learning

Courses

• Chelsea Finn's Deep Multi Task and Meta Learning course   https://www.youtube.com/playlist?list=PLoROMvodv4rMC6zfYmnD7UG3LVvwaITY5

Papers

• Curiosity-driven Exploration by Self-supervised Prediction   https://pathak22.github.io/noreward-rl/resources/icml17.pdf

• Contextual Imagined Goals for Self-Supervised Robotic Learning   https://arxiv.org/pdf/1910.11670.pdf

• Model Agnostic Meta Learning   https://arxiv.org/pdf/1703.03400.pdf

• Deep Visual Foresight for Planning Robots   http://phys.csail.mit.edu/papers/6.pdf

• More   https://twimlai.com/twiml-talk-217-trends-in-reinforcement-learning-with-simon-osindero/

   Summaries

• Multi-Agent Hide and Seek   https://www.youtube.com/watch?v=Lu56xVlZ40M

• Self-Supervised Deep RL   https://www.youtube.com/watch?v=txHQoYKaSUk

• Locomotion   https://www.youtube.com/watch?v=14zkfDTN_qo

• OpenAI's Safety Gym   https://www.youtube.com/watch?v=_s7Bg6yVOdo

Blogs & Presentations

• Meta-Learning: from Few-Shot Learning to Rapid Reinforcement Learning   https://icml.cc/media/Slides/icml/2019/halla(10-09-15)-10-13-00-4340-meta-learning_.pdf

• Unsupervised Meta Learning   https://bair.berkeley.edu/blog/2020/05/01/umrl/

• Exploring Evolutionary Meta-Learning in Robotics   https://ai.googleblog.com/2020/04/exploring-evolutionary-meta-learning-in.html

• Meta-Learning in Deep RL   http://rail.eecs.berkeley.edu/deeprlcourse-fa17/f17docs/lecture_16_meta_learning.pdf

Podcasts

• Sam Harrington's TWiML AI   https://twimlai.com/

• Women in AI   https://www.listennotes.com/podcasts/women-in-ai-rework-9Ae7qJTht0E/

• Lex Fridman's   https://lexfridman.com/ai/

• Practical AI   https://changelog.com/practicalai

   Episodes

• Chelsea Finn
  • https://twimlai.com/twiml-talk-335-trends-in-reinforcement-learning-with-chelsea-finn/
  • https://twimlai.com/twiml-talk-029-robotic-perception-control-chelsea-finn/

Implementation Tools

• OpenAI's Spinning Up in Deep RL   https://openai.com/blog/spinning-up-in-deep-rl/

• OpenAI's Gym   https://gym.openai.com/

• TensorFlow Agents (tf_agents)   https://www.tensorflow.org/agents

• Google Dopamine   https://github.com/google/dopamine   https://medium.com/the-21st-century/google-dopamine-new-rl-framework-f84a35b7fb3f

• Facebook's Higher   https://github.com/facebookresearch/higher

• Omniglot Dataset   https://www.tensorflow.org/datasets/catalog/omniglot   https://github.com/brendenlake/omniglot   https://www.groundai.com/project/the-omniglot-challenge-a-3-year-progress-report/1