Natural Computing
Lecture 1
Michael Herrmann
mherrman@inf.ed.ac.uk
phone: 0131 6 517177
Informatics Forum 1.42
Problem Solving in Nature
Natural Computing 20/09/2011 J. M. Herrmann
Natural Computation
Physics
Chemistry
Biology
Geology
Astronomy
Natural Co

Prediction
Models of nancial markets
Doyne Farmer has formed Prediction Co. in Santa Fe, N.M., which
uses genetic algorithms and other methods to create models of
nancial markets that are drawn from statistical history and the
computer's randomly generate

Prediction
Models of nancial markets
Doyne Farmer has formed Prediction Co. in Santa Fe, N.M., which
uses genetic algorithms and other methods to create models of
nancial markets that are drawn from statistical history and the
computer's randomly generate

NAT 2011: Assignment 1
This is an exercise worth 10% of the course mark. It requires you to write a program and to use it to
carry out a number of investigations. You are to write up your investigations in the form of a con ference paper with a maximum le

Natural Computing 2011: Assignment 2
Particle Swarm Optimisation
This is an exercise worth 20% of the course mark. It requires you to write programs and to use them
to carry out a number of investigations. You choose any of the common programming language

Natural Computing: Tutorial 1
(week 3)
1. Go (step-by-step and then fast) through example at (cf. lecture 3)
http:/www.obitko.com/tutorials/genetic-algorithms/example-function-minimum.php
2. The knapsack problem is as follows: Given a set of weights W and

NAT Tutorial 1
2. The knapsack problem is as follows: given a set of weights W, and a target weight T, find a
subset of W whose sum is as close to T as possible.
Example:
W = cfw_5, 8, 10, 23, 27, 31, 37, 41,
T = 82
- Solve the instance of the knapsack pr

NAT Tutorial 2
(week 4)
1. (Goldberg) You are asked to minimize a function f(x,y,z) where
-20 < x < 125,
0 < y < 1200000,
-0.1 < z < 1.0
and the desired precisions for x, y and z are 0.5, 10000 and 0.001 respectively. Using
the customary grid-based binary

NAT Tutorial 2
1. (Goldberg) You are asked to minimize a function f(x,y,z) where
-20 < x < 125,
0 < y < 1200000,
-0.1 < z < 1.0
and the desired precisions for x, y and z are 0.5, 10000 and 0.001 respectively. Using
the customary grid-based binary encoding

NAT Tutorial 3
(week 5)
1. [Baldwin effect] It has been observed that some organisms seem to pass on behaviours learned
during their lifetime to their offspring. Lamarcks hypothesis was that traits acquired during the
lifetime of an individual could someh

NAT Tutorial 3
(week 5)
1. [Baldwin effect] It has been observed that some organisms seem to pass on behaviours learned
during their lifetime to their offspring. Lamarcks hypothesis was that traits acquired during the
lifetime of an individual could someh

NAT Tutorial 4: Genetic programming
1. A GP system is employed to evolve a controller for a mobile robot. The fitness
function evaluates the robot performance starting from 50 initial positions. In a
long series of tests the system is observed to produce

NAT Tutorial 4: Genetic programming
1. A GP system is employed to evolve a controller for a mobile robot. The fitness
function evaluates the robot performance starting from 50 initial positions. In a
long series of tests the system is observed to produce

NAT Tutorial 5: Particle Swarm Optimization
1. Consider one of the following problems (or any other one that seems to be
interesting) and explain how you would use ant colony optimization to find an
acceptable solution: Sequential ordering, classification

NAT Tutorial 5: Particle Swarm Optimization
1. Consider one of the following problems (or any other one that seems to be
interesting) and explain how you would use ant colony optimization to find an
acceptable solution: Sequential ordering, classification

NAT Tutorial 6: Metaheuristic Algorithms
1. Recall the main algorithms that we were dealing with (i.e. GA, ES, GP, ACO, PSO
and possibly variants of these, if this makes a difference) and classify them
according to Dorigos the criteria for the classificat

NAT Tutorial 6: Particle Swarm Optimization
1. Recall the main algorithms that we were dealing with (i.e. GA, ES, GP, ACO, PSO
and possibly variants of these, if this makes a difference) and classify them
according to Dorigos the criteria for the classifi

Computer Science Department
June 1990
GENETIC PROGRAMMING: A PARADIGM FOR GENETICALLY
BREEDING POPULATIONS OF COMPUTER PROGRAMS TO
SOLVE PROBLEMS
John R. Koza
(Koza@Sunburn.Stanford.Edu)
Computer Science Department
Stanford University
Margaret Jacks Hall

Natural Computing
Lecture 2: Genetic Algorithms
J. Michael Herrmann
michael.herrman@ed.ac.uk
phone: 0131 6 517177
Informatics Forum 1.42
INFR09038
23/9/2011
Meta-heuristic algorithms
Similar to stochastic optimization
Iteratively trying to improve a possi

Natural Computing
Lecture 3
Michael Herrmann
mherrman@inf.ed.ac.uk
phone: 0131 6 517177
Informatics Forum 1.42
27/09/2011
The Canonical Genetic Algorithm
The Canonical Genetic Algorithm: Conventions
1
Old population
2
Selection
3
Intermediate population
4

Natural Computing
Lecture 4
Michael Herrmann
mherrman@inf.ed.ac.uk
phone: 0131 6 517177
Informatics Forum 1.42
30/09/2011
The Schema Theorem
and the Building Block Hypothesis
Reminder: The Canonical Genetic Algorithm
1
Old population
2
Roulette-wheel sele

Natural Computing
Lecture 5
Michael Herrmann
mherrman@inf.ed.ac.uk
phone: 0131 6 517177
Informatics Forum 1.42
04/10/2011
The Building Block Hypothesis and
GA Variants
The Schema Theorem
From last lecture
E (m (H , t + 1) uf(Ht,)t ) m (H , t )
(
1
Pc d (

Natural Computing
Lecture 6
Michael Herrmann
mherrman@inf.ed.ac.uk
phone: 0131 6 517177
Informatics Forum 1.42
07/10/2011
Hybrid GA
and Practical Applications
Lamarckism
Characterised by
Inheritance of acquired traits
Use and disuse determine characterist

Computer Science Department
June 1990
GENETIC PROGRAMMING: A PARADIGM FOR GENETICALLY
BREEDING POPULATIONS OF COMPUTER PROGRAMS TO
SOLVE PROBLEMS
John R. Koza
(Koza@Sunburn.Stanford.Edu)
Computer Science Department
Stanford University
Margaret Jacks Hall