We’ve partnered with the This Week in Machine Learning & AI podcast for a 7 part series on Industrial AI. Check out Episode 3 below and download our latest paper exploring the unique challenges and requirements of Industrial AI.
In Part 3 of TWIML’s Industrial AI series, Sam Charrington digs into robotics and reinforcement learning with Berkeley PhD student, Chelsea Finn. This talk gets into some of the technical weeds of cutting-edge robotics technologies, including inverse reinforcement learning, meta learning and the benefits and challenges of training robots in simulations. Chelsea also talks about what it’s like pursuing a PhD in machine learning and how to keep up with such a rapidly advancing field.
“Another approach is to train in simulation. It’s practical to acquire a lot of data and then try to use what you learned in simulation to be able to effectively act in the real world either with zero shot transfer, where you get zero data in the real world, or with few shot transfer or just fine tuning the real world, where you need less data from the real world than you would need if you didn’t have that simulated data.” - Chelsea Finn
Check out the full conversation with Chelsea below. If you want to explore using reinforcement learning in your own organization, learn more at bons.ai.