Dopamine is a research framework for fast prototyping of reinforcement learning algorithms. - google/dopamine
Denny BritzDopamine - a research framework for fast prototyping of reinforcement learning algorithms from Google: github.com/google/dopamine
This looks great - I just really hope it won’t be abandoned like the many other RL and TF frameworks out there.
After years of detective work, it's still unclear why a Japanese doctor faked dozens of clinical trials
Denny BritzResearcher at the center of an epic fraud remains an enigma to those who exposed him sciencemag.org/news/2018/08/r…
This is a fascinating read on how difficult it is to correct academic fraud and how wrong results will continue to live on
Recent progress in artificial intelligence through reinforcement learning (RL)
has shown great success on increasingly complex single-agent environments and two-player turn-based games. However, the real world contains multiple agents, each...
Denny BritzJust had time to catch up on the DeepMind CTF (deepmind.com/blog/capture-t…) post/paper. I don’t know what to say. This is a whole new level of production quality with these visualizations and videos. When do we start getting full-length theater style movies with research papers?
We've trained an agent to achieve a high score of 74,500 on Montezuma's Revenge from a single human demonstration, better than any previously published result. Our algorithm is simple: the agent plays a sequence of games starting from carefully...
Denny BritzLearning Montezuma's Revenge from a Single Demonstration buff.ly/2IWUvwq
I like this approach. While not applicable to all problems, it makes very smart use of the available resources: One demonstration and full control over the environment (simulator). pic.twitter.com/evvBVREyh3