Professional Documents
Culture Documents
Artificial Intelligence
Outline
Genetic Algorithms.
Particle Swarm Optimization.
Ant Colony Optimization.
Genetic Algorithms
History of GAs.
Applications of GAs.
Components of GAs.
General GA evolutionary loop.
Examples.
Genetic Algorithms
Invented by John Holland (1975).
Using population-based approach.
Based on Natural Selection ('survival of the fittest).
Use usually fixed length and linear representation
of chromosomes.
Genetic Operators for modification: Crossover and
Mutation.
Genetic Algorithms -
Applications
Domain Application Types
Control gas pipeline, pole balancing, missile evasion, pursuit
YES
Are stopping criteria satisfied ? Stop &Report
NO
Select individuals based on fitness
00010110100111010010
Permutation representation for problem 3.
(2 5 7 6 8 1 3 4)
(, ) selection.
(+) selection.
……
Genetic Algorithms -
Crossover and Mutation
Problem 1, 2:
Crossover
01011011110 01011011110
01011011110 01011011110
Mutation
01011011110
01011001110
Genetic Algorithms -
Crossover and Mutation
Problem 3:
Crossover
Mutation
43521 41523
Genetic Algorithms -
Crossover and Mutation
Problem 4:
Crossover
Mutation
Genetic Algorithms -
Parameters
History of PSO
Full PSO model
Aspects of the basic PSO
Performance Measures
INVENTORS
Proposed in 1995 by Eberhart and Kennedy
Used for solving optimization problems of
continuous nonlinear functions
sk 1 gbest k
gbest k
v k 1 d
k
v
k
pbest
pbest k
s k
d
Particle Flying Model
k 1 k k 1
s s v
w1 c1 rand ()
w2 c2 rand ()
i i i
sk 1 gbest k
gbest k
k 1 d
k 1 k k v
v i v v
i i
v k
pbest k
pbest k
s k
d
Do
For each particle
Calculate fitness value
If the fitness value is better than the best fitness value (pbest) in history
set current value as the new pbest
End
Choose the particle with the best fitness value of all the particles as the gbest
For each particle
Calculate particle velocity according equation (*)
Update particle position according equation (**)
End
While stopping condition is true
Gbest PSO:
Neighbor of particles is the
entire swarm
Lbest PSO:
Neighbor of a particle is some
particles that are near to the
particle.
INITIALISING PSO
The position:
Should be initialized to uniformly cover the search space
The velocity:
can be initialized to zero: vij (0) 0
STOPPING CONDITIONS
Swarm size:
from 10 to 30
Neighborhood size:
Smaller neighborhood sizes are
less susceptible to local optimal
Starting with small neighborhood
and increase the neighborhood
size.
Learning factors:
c =c =2.0
1 2
PERFORMANCE MEASURES
Accuracy : accuracy( S , t ) | f ( y ' (t ) f ( x*)) |
n
Reliability : reliabilit y ( S (t ), )
N
100
Robustness : robustness( S (t )) [ , ]
ACO history
ACO applications.
Implementation Steps.
ACO basic algorithms
ACO-History
Goss et al. 1989,
Deneuborg et al.
1990, experiments
with Argentine ants
Dorigo et al. 1991,
applications to
shortest path
problems
Now: established
method for various
optimization problems
ACO-Applications
Efficiently Solves NP hard Problems 1
Routing
5
TSP (Traveling Salesman Problem) 2
Vehicle Routing
Sequential Ordering
Assignment 3 4
QAP (Quadratic Assignment Problem)
Graph Coloring
Generalized Assignment
Frequency Assignment
University Course Time Scheduling
ACO-Applications
Scheduling
Job Shop
Open Shop
Flow Shop
Total tardiness (weighted/non-weighted)
Project Scheduling
Group Shop
Subset
Multi-Knapsack
Max Independent Set
Redundancy Allocation
Set Covering
Weight Constrained Graph Tree partition
Arc-weighted L cardinality tree
Maximum Clique
ACO-Applications
Other
Shortest Common Sequence
Constraint Satisfaction
2D-HP protein folding
Bin Packing
Machine Learning
Classification Rules
Bayesian networks
Fuzzy systems
Network Routing
Connection oriented network routing
Connection network routing
Optical network routing
Problem Solving by ACO
Daemon
activities
General Ant Colony Heuristic
Ants generation and
activity: compute
make decision
while resources transition
and move
probabilities
available: create ant
for each ant:
1. initialize
evaluate
update the
2. let ant run until a the current state
solution is found state
k
p (t )
ij
(t ) . ij
(t ) . il
ij
il
J ik
k
p (t )
ij (t ) .
ij
(t ) . il
ij
il
J ik
k k k
Q / L (t ) if (i, j ) T (t ) else 0.
ij
ij (t ) (1 ). ij (t ) ij (t )
Ant Colony System (ACS)
k
j arg max uJ i { ij (t ) .iu } if q qo j J
Modifications to AS.
New transition rule:
qo is a parameter that can be tweaked
It is similar to tuning temperature in SA.
J is a city randomly selected according to the
probability calculated previously.
This helps ACS to improvise on the best solutions.
Ant Colony System (ACS)
ij (t ) (1 ). ij (t ) . ij (t )
Pheromone update rule (new):
However, only applied to the best ant.
The change in the pheromone concentration =
1/L+.
Local updates done as follows:
ij (t ) (1 ). ij (t ) 0
ACO - Further Readings
A. P. Engelbrecht, Fundamentals of
Computational Swarm Intelligence, John Wiley &
Sons, 2006.
E. Bonabeau, M. Dorigo, G. Theraulaz, Swarm
Intelligence: From Natural to Artificial Systems,
Oxford University Press, 1999.
ACO tutorials by Marco Dorigo.