Opening Variational Inference, Part 4 (CS 285, Lecture 18)
Variational Inference, Part 4 (CS 285, Lecture 18)
- Length
- 25 min
- Level
- Graduate
- Recorded
- UC Berkeley2023
Chapter 1 of 4: Before the lecture
Variational autoencoders give deep reinforcement learning a practical form of amortized variational inference. In the fourth part of the CS 285 variational inference sequence, Levine trains an encoder and decoder jointly on the evidence lower bound using the reparameterization trick. Learned latent states then stand in for raw images, with Higgins et al. on disentangled representations and Embed to Control as examples. Conditional variational autoencoders follow, applied to multimodal action sequences as in Learning Latent Plans from Play. The final part builds sequence-level models over whole trajectories for partially observed systems, citing Stochastic Latent Actor-Critic, and assumes the evidence lower bound, Gaussian distributions and neural networks.
Your comment is kept while you sign up.Sign in
Levine trains variational autoencoders by reparameterization for latent states, plans and trajectories.
Variational autoencoders give deep reinforcement learning a practical form of amortized variational inference. In the fourth part of the CS 285 variational inference sequence, Levine trains an encoder and decoder jointly on the evidence lower bound using the reparameterization trick. Learned latent states then stand in for raw images, with Higgins et al. on disentangled representations and Embed to Control as examples. Conditional variational autoencoders follow, applied to multimodal action sequences as in Learning Latent Plans from Play. The final part builds sequence-level models over whole trajectories for partially observed systems, citing Stochastic Latent Actor-Critic, and assumes the evidence lower bound, Gaussian distributions and neural networks.
Higgins et al., 2017
Learning Latent Plans from Play
Embed to Control: A Locally Linear Latent Dynamics Model for Control from Raw Images
SOLAR: Deep Structured Latent Representations for Model-Based Reinforcement Learning
Stochastic Latent Actor-Critic: Deep Reinforcement Learning with a Latent Variable Model
This lecture
Lecture 55 of 73 of Deep Reinforcement Learning.
Choose a node to see how it connects to this lecture. What comes before is on the leftat the top. What comes after is on the rightat the bottom. The ring is what it teaches. People and places are above. Other lectures and books are below.
Press a chosen node again to open it.