ep5 - Sean Meyn: Markov chains, networks, reinforcement learning, beekeeping and jazz
inControl
00:00
When Do You Need Stochastic Control?
When do you need to go stochastic? That would be a question that I would ask you as a giant in the field. You get a family of decision roles based on the simpler model, and then you fine-tune it based on a more realistic model. Maybe using some statistical techniques like stochastic approximation or some formulation of RL but that's a fine-tuning after you get a good policy. In finance, volatility is everything. But in so many cases, it's back to the fluid model idea.
Transcript
Play full episode
