AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
Markov Chain Monte Carlo - A Simple Method for Learning Unnormalized Distributions
Markov chan monte carlo is an increasingly popular sampling method for obtaining asemtotic information about unnormalized distributions or energy functions. It's used to estimate the post rio distribution in basian inference, which was where you'v probably heard of it before. The real world is expensive, so why not train an imagination in our mind until we're ready to produce good questions? So benjoe's big idea is that we could have an interacting loop between a generative model and the real world. For g flow nets to work, we need a reward function and a deterministic episodic environment. Does that sound familiar? Yes, just like reinforcement learning.