Scalable machine learning,
Scalable Python, for everyone
The increasing prevalence of high-fidelity industrial digital twins is providing a range of opportunities to apply Reinforcement Learning (RL) techniques outside of traditional academic examples and video games. We will present how we supported Emirates Team New Zealand in winning the 36th America’s Cup by leveraging some of the latest AI techniques and technology platforms. The America’s Cup is the oldest sports trophy in the world, but is first and foremost a technology competition. It’s a race against time to design, test, and deliver the fastest yacht on race day, with a team of Olympic-level athletes ready to sail it and make it fly thanks to cutting-edge hydrofoils. We will describe how we used deep Reinforcement Learning to speed up their design process. In particular, we will deep dive on the underlying technologies we used, from scaling RL training on a Ray cluster, to using Tune to manage results variability.
Chief Data Scientist, McKinsey/QuantumBlack Australia