Swarm Intelligence

From Natural to Artificial Systems

Ukradnut kde sa dalo, a adaptovan
Swarming The Definition
aggregation of similar animals, generally
cruising in the same direction


Termites swarm to build colonies
Birds swarm to find food
Bees swarm to reproduce
Why do animals swarm?
To forage better
To migrate
As a defense against predators



Social Insects have survived for millions of
years.
Swarming is Powerful
Swarms can achieve things that an individual
cannot
Swarming Example
Bird Flocking

Boids model was proposed by Reynolds
Boids = Bird-oids (bird like)

Only three simple rules

Collision Avoidance
Rule 1: Avoid Collision with neighboring birds
Velocity Matching
Rule 2: Match the velocity of neighboring
birds
Flock Centering
Rule 3: Stay near neighboring birds
Swarming - Characteristics
Simple rules for each individual

No central control
Decentralized and hence robust

Emergent
Performs complex functions
Learn from insects
Computer Systems are getting complicated
Hard to have a master control


Swarm intelligence systems are:
Robust
Relatively simple
Swarm Intelligence - Definition
any attempt to design algorithms or
distributed problem-solving devices inspired
by the collective behavior of social insect
colonies and other animal societies
[Bonabeau, Dorigo, Theraulaz: Swarm
Intelligence]
Solves optimization problems

Applications
Movie effects
Lord of the Rings

Network Routing
ACO Routing

Swarm Robotics
Swarm bots

Roadmap
Particle Swarm Optimization
Applications
Algorithm
Ant Colony Optimization
Biological Inspiration
Generic ACO and variations
Application in Routing
Limitations of SI
Conclusion
Particle Swarm Optimization
Particle Swarm Optimization
Particle swarm optimization imitates human
or insects social behavior.
Individuals interact with one another while
learning from their own experience, and
gradually move towards the goal.
It is easily implemented and has proven both
very effective and quick when applied to a
diverse set of optimization problems.







Bird flocking is one of the best example of
PSO in nature.
One motive of the development of PSO was
to model human social behavior.
Applications of PSO
Neural networks like Human tumor analysis,
Computer numerically controlled milling
optimization;
Ingredient mix optimization;
Pressure vessel (design a container of
compressed air, with many constraints).
Basically all the above applications fall in a
category of finding the global maxima of a
continuous, discrete, or mixed search space,
with multiple local maxima.
Algorithm of PSO
Each particle (or agent) evaluates the
function to maximize at each point it visits in
spaces.
Each agent remembers the best value of the
function found so far by it (pbest) and its co-
ordinates.
Secondly, each agent know the globally best
position that one member of the flock had
found, and its value (gbest).
Algorithm Phase 1 (1D)
Using the co-ordinates of pbest and gbest,
each agent calculates its new velocity as:

v
i
= v
i
+ c
1
x rand() x (pbestx
i
presentx
i
)
+ c
2
x rand() x (gbestx presentx
i
)

where 0 < rand() <1

presentx
i
= presentx
i
+ (v
i
x t)
Algorithm Phase 2 (n-dimensions)
In n-dimensional space :

Ant Colony Optimization
Ant Colony Optimization - Biological
Inspiration
Inspired by foraging behavior of ants.
Ants find shortest path to food source from nest.
Ants deposit pheromone along traveled path which
is used by other ants to follow the trail.
This kind of indirect communication via the local
environment is called stigmergy.
Has adaptability, robustness and redundancy.


Foraging behavior of Ants
2 ants start with equal probability of going on
either path.
Foraging behavior of Ants
The ant on shorter path has a shorter to-and-
fro time from its nest to the food.
Foraging behavior of Ants
The density of pheromone on the shorter
path is higher because of 2 passes by the ant
(as compared to 1 by the other).
Foraging behavior of Ants
The next ant takes the shorter route.
Foraging behavior of Ants
Over many iterations, more ants begin using
the path with higher pheromone, thereby
further reinforcing it.
Foraging behavior of Ants
After some time, the shorter path is almost
exclusively used.
Generic ACO
Formalized into a metaheuristic.
Artificial ants build solutions to an
optimization problem and exchange info on
their quality vis--vis real ants.
A combinatorial optimization problem
reduced to a construction graph.
Ants build partial solutions in each iteration
and deposit pheromone on each vertex.

Ant Colony Metaheuristic
ConstructAntSolutions: Partial solution extended by adding
an edge based on stochastic and pheromone
considerations.
ApplyLocalSearch: problem-specific, used in state-of-art
ACO algorithms.
UpdatePheromones: increase pheromone of good
solutions, decrease that of bad solutions (pheromone
evaporation).
Various Algorithms
First in early 90s.
Ant System (AS):
First ACO algorithm.
Pheromone updated by all ants in the iteration.



Ants select next vertex by a stochastic function
which depends on both pheromone and problem-
specific heuristic n
ij
=

1
dij
Probability ofantkgoingfrom
cityitojatiterationt
( )
( )
( ) | |
not visited
, not visited
ij ij
k
ij
ik ik
k
t
p t j
t
o |
o
|
t q
t q
( (

=
(

o=1, |=5, poet mravcov m=poet miest, Q=100,

poiaton mnostvo feromnu t
0
=10
-6
Alpha = 0 : represents a greedy approach
Beta = 0 : represents rapid selection of tours
that may not be optimal.
Thus, a tradeoff is necessary.

Various Algorithms - 2
MAX-MIN Ant System (MMAS):
Improves over AS.
Only best ant updates pheromone.
Value of pheromone is bound.



L
best
is length of tour of best ant.
Bounds on pheromone are problem specific.
Theoretical Details
Convergence to optimal solutions has been
proved.
Cant predict how quickly optimal results will
be found.
Suffer from stagnation and selection bias.
Scope
List of applications using SI growing fast
Routing
Controlling unmanned vehicles.
Satellite Image Classification
Movie effects
Conclusion
Provide heuristic to solve difficult problems
Has been applied to wide variety of
applications
Can be used in dynamic applications
References
Reynolds, C. W. (1987) Flocks, Herds, and Schools: A Distributed Behavioral
Model, in Computer Graphics, 21(4) (SIGGRAPH '87 Conference
Proceedings) pages 25-34.

James Kennedy, Russell Eberhart. Particle Swarm Optimization, IEEE Conf.
on Neural networks 1995

www.adaptiveview.com/articles/ ipsop1

M.Dorigo, M.Birattari, T.Stutzle, Ant colony optimization Artificial Ants as a
computational intelligence technique, IEEE Computational Intelligence
Magazine 2006

Ruud Schoonderwoerd, Owen Holland, Janet Bruten - 1996. Ant like agents
for load balancing in telecommunication networks, Adaptive behavior, 5(2).