01 Decision-Making Under Uncertainty
How should an intelligent system act when it cannot know the world completely?
Area 01: Decision-Making Under Uncertainty
Master Question
How should an intelligent system act when it cannot know the world completely?
What We Want to Discover
Real agents rarely observe the true state of the world directly. They observe noisy, partial, or delayed signals and must still choose an action. This area studies how an agent should represent what it does not know, how it should update that representation as evidence arrives, and how uncertainty should shape the action it finally takes.
Why It Matters
Almost every other area in this portfolio assumes some solution to this problem, whether the agent is a mobile robot with a noisy sensor, a satellite with a limited observation window, or a multi-agent system reasoning about what another agent believes. Getting this foundation wrong propagates errors into every system built on top of it.
Core Concepts
- Partially observable Markov decision processes (POMDPs)
- Belief states and belief updating
- Bayesian inference and probabilistic filtering
- Value of information and information gain
- Exploration versus exploitation
- Risk sensitive decision making
Relevant Disciplines
- Decision theory
- Bayesian statistics
- Control theory
- Cognitive science
- Epistemology
Potential Mimicry Sources
- Animal foraging under incomplete information
- Human active sensing and gaze control
- Statistical decision making in medicine
Projects
No projects yet.
Active Questions
No projects yet, so there are no active derived questions to report.
Key Findings Across Projects
Pending. No projects in this area have produced findings yet.
Unresolved Questions
Pending.
Connections to Other Areas
Pending. Connections will be identified as projects in this area develop.