Problemsolving_Course_Lecture_06 (1).Pdf
Bạn đang xem bản rút gọn của tài liệu. Xem và tải ngay bản đầy đủ của tài liệu tại đây (1.63 MB, 17 trang )
Problem-Solving
& Decision Making
Introduction to Decision-Making II
Problem-Solving & Decision Making
Introduction to Decision-Making II
Abdel-Rahman Hedar
Artificial Intelligence Professor
BUA
UNI1103
Agenda
Players in a decision
Representation of
Decision Problems
Expected Utility
Function
Section 1: Players in a Decision
Players in a decision
•
The decision makers (DMs) are responsible for making the
decision: they ‘own the problem’.
To be able to take and implement a decision, DMs need hold the
appropriate responsibility, authority and accountability.
•
•
•
•
Responsibility: An individual or group is responsible for a decision. Their
task to see that the choice is made and implemented.
Authority: An individual or group has the authority to take a decision.
They have power over the resources needed to analyze and implement the
choice.
Accountability: An individual or group is accountable for a decision. They
are the ones to take the credit or blame for the decision process and the
choice that is made and implemented.
5
Section 2: Representation of Decision Problems
Data, Information and Knowledge
8
Representation of Decision Problems
• The action space, A = {a1, a2,…,am}
o the set of options from which the DM may choose.
• The state space, Θ = {θ1,θ2,…,θn}.
• The DM will receive a consequence, cij, lying in some
consequence space, C, determined both by the chosen action
ai and the state θj that pertains:
9
Decision
Table
10
Decision Table Example
Section 3: Expected Utility Function
Decision Problems
Decisions under certainty
• DM either knows or learns the true state before making the
decision.
Decisions with risks
• DM does not know the true state but based on some knowledge
it is likely to make some possible state.
Decisions under strict certainty
• Can say nothing at all about the true state.
• Identify only what states may be possible.
13
Subjective Expected Utility Model
• Subjective Expected Utility (SEU) Model can be used to
represent the decision problem with risks.
• SEU Model has two elements:
o Subjective Probability Distribution, P(.).
- Give a probability for a state.
o Utility Function, u(.).
- Give preferences for consequences.
14
Subjective Expected Utility Model
• P(θ) > P(θ′): DM believes state θ to be more likely to
occur than θ′.
• u(c) > u(c′): DM strictly prefers consequence c to
consequence c′.
• The SEU model asserts that to combine her beliefs and
preferences coherently in order to rank the actions the DM
should form expected utilities:
15
SEU Example
Questions?
Comments!
