Normal and Extensive Form Games
Can we use normal and extensive form games for
infinite games?
Can normal and extensive form games be used in
games where the agent does not have perfect
knowledge of everyones payoffs?
Can we use the normal and extensiv
Deals with the design of a multiagent strategy or algorithm to optimize the air traffic
flow system
It works by assigning some intelligent agents as controllers of aircrafts in a limited 2D
space.
The objective is to minimize delays and congestion in the
Normal Form Games
Finite, n-person normal form game: (N, A, u):
N is a finite set of n players, indexed by i
A = A1, . . . , An is a set of actions for each
player i
a
A is an action profile
u = cfw_u1, . , un, a utility function for each
player, whe
Insincere Agents
The agents chose the best strategy for
them.
A protocol implements a specific social
choice function if the protocol has an
equilibrium that results in the same
outcome even when insincere agents do
not truthfully reveal their preferenc
Distributed Planning
A combination of traditional AI planning
and distributed problem solving.
Either plan creation is centralized, but the
plan execution is distributed
or the plan creation is distributed, but the
plan execution is centralized
or the
Learning Process Definition
Learning is defined as:
activities that are executed with the
intention to achieve a particular learning
goal.
Multiagent Learning Characterizations
Two primary learning techniques for
multiagents:
Centralized learning (iso
2/22/2011
OVERVIEW
Adversarial system definition
Formalization of adversarial environment
Connect four
Risk
Future work
BACKGROUND
Early work cooperation
Rise of BDI
ADVERSARIAL ACTIVITY
(AA)
Purpose:
Framework, not directly implementable
Overcoming
Why Multiagents?
Distribute the problem solving.
Sometimes no centralized solution will work.
Complex problems.
Geographically distributed
Numerous components
Large amount of content
Broad scope
Multiagent Environments
Characteristics of multiagent
en
2/2/2011
Authors: Adam Campbell and Annie S. Wu
Presenter: Caroline Harriott
Roles vs. Tasks
Roles in Multi-agent systems
o es
u t age t syste s
3 Models of Role Allocation Problem
OAP
E-GAP
RMTDP
Choosing a Role Allocation Procedure
Types of Role Allo
Distributed Optimization
Distributed methods of satisfying
global constraints.
Assumes cooperative agents
Path-Finding Problem
Path finding can be represented as a
graph with weighted directed links.
The link weight can represent the distance
between
Linear Programming
In mathematics, linear programming
(LP) is a technique for optimization of a
linear objective function, subject to linear
equality and linear inequality constraints.
Informally, linear programming determines
the way to achieve the best
Distributed Problem Solving
Assume agents share a common goal.
Types of problems
Path-finding problems: Find a path from the
start state to the goal state.
8 or 15 puzzle
Constraint satisfaction problems (CSP): Find a
configuration that satisfies the
2/4/2011
Game Theory
Non-cooperative game theory
Coalitional (or cooperative) game theory
Self-Interested Agents
Self-interested (SI) agents maintain a personal
description of the world states and choose
d
i i
f h
ld
d h
actions that attempt to create the
Extensive Form Games
Normal form game representation does
not incorporate any notion of sequence,
or time, of the players actions.
Extensive form games are an alternative
representation that explicitly specifies the
temporal structure.
Two variants:
Perfe
Nash Equilibrium
A strategy profile s = (s1, , sn) is a Nash
Equilibrium if for all agents i si is a best
if,
i,
response to s-i.
two strategies si and s-i are in Nash equilibrium
if:
1. under the assumption that agent i plays si,
agent j can do no better