CIS 5603. Artificial Intelligence
Uncertain Reasoning
In Artificial Intelligence, uncertainty has been an active field for decades, with many approaches explored.
1. Probabilistic reasoning
There have been various attempts to carry out reasoning according to probability theory.
Bayesian Networks, also known as Belief Networks, use assumptions on conditional independence to reduce the demand on joint probability and computational complexity.
Various types of inference (like deduction, induction, and abduction) can be uniformly treated as Bayesian conditioning, which also covers certain types of learning (example).
Other related approaches:
Issues:
2. Decision making
In practical reasoning, "to achieve a goal" can be generalized to "to achieve the highest reward/utility".
Decision theory: When each state has a utility associated, and each action has a probability distribution on the states it may lead to, the best action is the one that has the Maximum Expected Utility (MEU).
Markov Decision Process (MDP) is one formalization of multi-step decision making or probabilistic planning, where the optimal policy maximizes the expected (discounted) total rewards.
3. Alternative approaches
There are also approaches beyond probability theory:
Reading