Explore the concept of "Exploration via Randomized Value Functions" in this 50-minute lecture from the 2019 ADSI Summer Workshop on Algorithmic Foundations of Learning and Control. Delve into topics such as deep exploration, linear sanity checking, stochastic policies, and Thompson sampling. Examine algorithmic ideas, intuition, and theory behind shortest path problems, episodic reinforcement learning, and model-based tabular learning. Investigate scalable approximations, value function approximations, and the real challenges facing the field. Gain insights into deep questions surrounding exploration in learning and control algorithms.
2019 ADSI Summer Workshop- Algorithmic Foundations of Learning and Control
Paul G. Allen School via YouTube
Overview
Syllabus
Introduction
Balancing
Outline
Deep Exploration
Large Theory
Linear sanity checking
The problem
Stochastic policies
Algorithm template
greedy
Thompson sampling
Does it work
One algorithmic idea
Intuition
Theory
Shortest Path
episodic reinforcement learning
modelbased tabular learning
scalable approximations
value function approximations
the real challenge
deep questions
Taught by
Paul G. Allen School
Related Courses
-
ADSI Summer Workshop: Algorithmic Foundations of Learning and Control - Emma Brunskill
-
ADSI Summer Workshop- Algorithmic Foundations of Learning and Control
-
Reinforcement Learning From Small Data in Feature Space - 2019 ADSI Summer Workshop
-
Linear Quadratic Control and Online Learning - 2019 ADSI Summer Workshop
-
An Online Learning Approach to Model Predictive Control - 2019 ADSI Summer Workshop
-
2019 ADSI Summer Workshop- Algorithmic Foundations of Learning and Control, Yinyu Ye
