ConceptAI & ML
Reward function
A scalar-valued function of the state and action that indicates which states and actions are better.
Know first
Named as a prerequisite by the lectures that define this. The number is how many say so. A grey one has no page yet.
Opens up
Concepts that name this one as a prerequisite.
Inverse reinforcement learning3Markov decision process2Active learning for reward function learning1Explicit policy constraint methods1Goal-conditioned policies1Intrinsic motivation1Markov reward process1Model1Monte-Carlo evaluation in go1Multi-reward systems1Off-dynamics reinforcement learning1Optimal control1
Lecturers
2Each in their own terms, from the easiest level.
Pieter Abbeel
A function that returns a positive, negative, or zero reward for taking an action from a state and landing in a new state.
Lectures that assume it
17The first 12 are shown.
- Markov Processes, Reward Processes & Bellman Equations
- Offline Reinforcement Learning II
- Introduction to Reinforcement Learning (CS234, 2024, Lecture 1)
- Markov Decision Processes (MDPs) Part 2/2
- What Humans Want: Models of Preference and Reward in Minds and Machines
- Control as Inference, Part 1
- Inverse Reinforcement Learning, Part 1
- Digital vs. Biological Intelligence (Vienna Gödel Lecture)
- Sequential Decision-Making: Policy Gradients vs. Value-Based Architectures
- Beyond Brute Force: The Era of Adaptive Intelligence
- Integrating Learning and Planning
- Robotics, Language, and Vision
Connections
27The lectures that define it or assume it, and the concepts next to it.
Concepts8
- Action space
- Markov decision problem
- State space
- Inverse reinforcement learning
- Markov decision process
- Active learning for reward function learning
Show 2 moreShow fewer
Lectures19
- Markov Decision Processes (MDPs) Part 1/2
- Policy Gradient Foundations: Trajectory Distributions & High-Dimensional State Spaces
- Markov Processes, Reward Processes & Bellman Equations
- Offline Reinforcement Learning II
- Introduction to Reinforcement Learning (CS234, 2024, Lecture 1)
- Markov Decision Processes (MDPs) Part 2/2
Show 13 moreShow fewer
- What Humans Want: Models of Preference and Reward in Minds and Machines
- Control as Inference, Part 1
- Inverse Reinforcement Learning, Part 1
- Digital vs. Biological Intelligence (Vienna Gödel Lecture)
- Sequential Decision-Making: Policy Gradients vs. Value-Based Architectures
- Beyond Brute Force: The Era of Adaptive Intelligence
- Integrating Learning and Planning
- Robotics, Language, and Vision
- Offline Reinforcement Learning II
- Transfer Learning & Meta-Learning
- Language Models: Prompting, Chain-of-Thought, and Instruction-Tuning
- Exploration of advanced topics in deep reinforcement learning
- Guest Lecture: Dorsa Sadigh (UC Berkeley CS 285)