Value Function
A value function is an object in Reinforcement Learning and Control Theory. It corresponds to the sum of (potentially discounted) rewards of a control problem as defined by Richard Ernest Bellman.
Links to this note:
- Current Learning Objectives
- white2015developing: Developing a predictive approach to knowledge
- veeriah2019discovery: Discovery of Useful Questions as Auxiliary Tasks
- vanhasselt2015learning: Learning to Predict Independent of Span
- Temporal Difference Learning
- sutton2011horde: Horde: A Scalable Real-time Architecture for Learning Knowledge from Unsupervised Sensorimotor Interaction
- Predictive Processing
- Predictive Knowledge
- Off-policy Reinforcement Learning
- kearney2019making: Making Meaning: Semiotics Within Predictive Knowledge Architectures
- jaderberg2017reinforcement: Reinforcement Learning with Unsupervised Auxiliary Tasks
- Interview Review Material
- StudyPlan