Outline
Logistics
Sources of Uncertainty
Agents With Uncertainty
Knowledge Representation
Ways to Represent Uncertainty
Numerical Repr of Uncertainty
Probability As “Softened Logic”
Probabilistic Knowledge Representation and Updating
Example: Is This Cow a Menace?
Basics
Conditional Independence
Summary so Far
Computational Models for Probabilistic Reasoning
Causality
A Different Network
Creating a Network
Example
Structural Models
Possible Bayes Network
Graphical Models and Problem Parameters
Complete Bayes Network
NOISY-OR: A Common Simple Model Form
More Complex Example
A First-Cut Graphical Model
Structural Relationships and Independence
Cond. Independence in Bayes Nets
More on D-Separation
Two Remaining Questions
Adding Evidence
Inference
Algorithm
Simplest Causal Case
Simplest Diagnostic Case
Normalization
General Case
Multiply connected nets
Review Question
Decision Networks (Influence Diagrams)
Evaluation
Course Topics by Week
Unifying View of AI
Specifying a search problem?
Example: AI Planning
How Represent Actions?
Planning as Search
Forward-Chaining World-Space Search
Planning as Search 2
Plan-Space Search
Planning as Search 3
Planning as Search 4
Search Summary
Binary Constraint Network
Constraint Satisfaction Summary
Backjumping (BJ)
Other Strategies
Nodes Explored
Knowledge Repr. Summary
Resolution
Davis Putnam (DPLL)
GSAT
Immobile Robots Cassini Saturn Mission
Solution: Part 1 Model-based Programming
Solution: Part 2 Model-based Deductive Executive
Solution: Part 3 Risc-like Best-first, Deductive Kernel
A family of increasingly powerful deductive model-based optimal controllers
Specifying a valve
Mode identification + reconfiguration
Example: Cassini propulsion system
MI/MR as combinatorial optimization
Propositional. Logic vs First Order
IIIIIS Representation III
Query Planning
Overview of Construction
Inverse Rules
Efficient & Robust Execution
Defining a Learning Problem
DT Learning as Search
Search thru space of Decision Trees
Resulting Tree ….
Information Gain
Overfitting…
Comparison
Ensembles of Classifiers
Constructing Ensembles
PAC model
Wrapper Induction
Wrapper induction algorithm
MDP Model of Agency
Trajectory
Properties of the Model
Value Iteration
Policy Iteration
Reinforcement Learning
PPT Slide
Email: weld@cs.washington.edu
Other information: CSE 592, Lecture 10
Download presentation source