Title: Hierarchical Learning in Complex System a Research outlook
1Hierarchical Learning in Complex System - a
Research outlook
- Jie BAO , PhD Student
- Artificial Intelligence Lab
- Department of Computer Science
- Iowa State University
- baojie_at_cs.iastate.edu
- http//www.cs.iastate.edu/baojie
-
2Hierarchy is common in Biological Intelligent
System
From Human Physiology Chapter 8 CNS, School of
Medicine and Health Sciences, University of North
Dakota,
Microstructure of brain - Hypercolumns in Visual
Cortex
Macrostructure of the brain
http//www.cs.iastate.edu/baojie/acad/current/cas
/cas.htm
3Outline for Hierarchical Learning
- General Hierarchy Formation
- Nature system are mostly hierarchical system
- Divide and conquer in Engineering
- The power and stableness from cooperation of
subsystems - Easier to design and implement
- Direction Order increasing
- Grow
- Multiplication
- Accelerate
- Interaction
- Higher level hierarchy appears!
- Differentiation
- Centralization
A general induction for Hierarchy in InfoSys _at_
http//www.public.iastate.edu/baojie/pub/2002-03-
06_hierarchical.ppt, See some philosophy
background _at_ http//www.cs.iastate.edu/baojie/ess
ay/ph/ph.htm
4Hierarchical Algorithms
- in Neural Network
- Hierarchy of neural system
- Multi-net System
- Modular Learning
- Ensemble Learning
- Hybrid learning
- in Multi Agent System
- Hierarchical Graphical Game
- Evolution of Cooperation
- Team Formation
- in Genetic Algorithm
- Hierarchy in Selection Social/Ecological
structure in population - Hierarchy in Coding Behavior Evolution in GA
(BEGA) Genetic Programming
http//www.cs.iastate.edu/baojie/acad/current/hnn
/hnn.htm
http//www.cs.iastate.edu/baojie/acad/current/mas
/mas.htm
http//www.cs.iastate.edu/baojie/acad/current/beg
a/bega/2001-11-22_bega.htm
5Current Focus- Hierarchical Graphical Game for
Multiagent
- Bayesian Network -gt Game Network Single Game
- Includes Decision, Chance and Utility in the
network - Find Strategy Independency in the network
- Find Nash Equilibria in multi-player game
- Game Network for repeated Game
- A graphical representation of Markov Game
- Use Graph to find Evolutionary Stable Strategy
(ESS) - Team (Hierarchy) formation in repeated game
Decision
Chance
Utility