COMP538ReasoningandDecisionunderUncertainty.pptVIP

* * 1. Java Basics for C++ Programmer COMP 538 Reasoning and Decision under Uncertainty Introduction Readings: Pearl (1998, Chapter 1 Shafer and Pearl, Chapter 1 Objectives Course objectives Course contents Uncertainty Uncertainty: the quality or state of being not clearly known. Uncertainty appears in many tasks Partial knowledge of the state of the world Noisy observations Phenomena that is not covered by our models Inherent randomness Probability and Decision Theory Well-known and well-understood framework for uncertainty Clear semantics Provides principled answers for: Combining evidence Predictive Diagnostic reasoning Incorporation of new evidence Intuitive (at some level) to human experts Can be learned Course Objectives When applied to real-world problems, probability theory and decision theory suffer from Complexity of model construction Complexity of problem solving This course covers methodologies developed recently in AI community for dealing with those complexity problems. The methodologies combine ideas from several disciplines Artificial Intelligence, Machine Learning Decision Theory, Theory of Computer Science Statistics, Information Theory, Operations Research Complexity Problem of Applying Probability Theory Example: Patients in hospital are described by several attributes: Background: age, gender, history of diseases, … Symptoms: fever, blood pressure, headache, … Diseases: pneumonia, heart attack, … A joint probability distribution needs to assign a number to each combination of values of these attributes, exponential model size. 20 attributes require 2020 ( roughly 106 ) numbers Real applications usually involve hundreds of attributes Because of the exponential model size problem, it was believed that probability theory is not practical for dealing with uncertainty in AI. Alternative uncertainty calculi were introduced: uncertainty factors, non-monotonic logic, fuzzy logic, etc. Complexity Problem of Applying Probability Theory