Introduction - Self-Adaptive Heuristics for Evolutionary Computation - page 3

Chapter 2
Evolutionary Algorithms

Computational Intelligence

Neural Networks

Evolutionary
Algorithms

classification
representation

Self Organizing Networks
Backpropagation
Spiking Networks

optimization
evolution

Evolution Strategies
Genetic Algorithms
Genetic programming

Reinforcement
Learning
Value Iteration
Q-Learning

Memetic Algorithms
Evolution and local search
Information sharing and
communication

learning behavior
optimization information

sharing

control
inference

optimization

emergent
behavior

Artificial Immune
Systems
Pattern recognition
Hypermutation

Fuzzy Logic
Fuzzy rule bases & inference
Fuzzy representation
and modelling

Chapter 3
Self-Adaptation

Swarm Intelligence

Particle Swarm Optimization
Ant Colony Optimization

Parameters
- exogenous
- endogenous
- population level
- individual level
- component level

Parameter setting

Parameter tuning Parameter control

by hand DOE metaevolution deterministic adaptive self-adaptive

Chapter 4
Biased Mutation

Human
experience

Sequential parameter
optimization
Response surface
modeling

Metaevolutionary
angle control ES
Coevolutionary
approaches

Dynamic penalty
function
Annealing penalty
function

Rechenberg's
1/5th rule
Covariance matrix
adaptation (CMS/CSA)

Step size control
of ES
SA-PMX

Methods

Hybridization

1

P

2

Chapter 5
Self-Adaptive Inversion
Mutation

b 1

2

A

A

B B

Chapter 6
Self-Adaptive Crossover

C C

D D

Chapter 7
Constraint Handling

infeasible search
space

feasible search
space

infeasible search
space

feasible search
space

P t-1 P t

optimum at
vertex of boundaries

optimum at
vertex of boundaries

optimum of
unconstrained
objective function

optimum of
unconstrained
objective function

Fig. 1.1. Survey of the chapters. After the introductive chapter 2 about EC, chapter 3
describes an extended taxonomy of parameter control and an estimation of distribution
view on self-adaptation. In chapter 4 the biased mutation is described and evaluated,
while chapter 5 equips inversion mutation with self-adaptation. Chapter 6 answers why
the self-adaptation of the crossover structure fails and chapter 7 evaluates methods for
optimization at the edge of feasibility.

is complemented by a history of adaptation techniques and an overview over
typical adapted parameters. We define self-adaptation in two kinds of ways. In
the standard definition self-adaptation is the evolutionary optimization of strat-
egy parameters, which are connected to the objective variables. From the point
of view of estimation of distribution algorithms (EDAs) self-adaptation is the
evolutionary control of a mixture distribution which estimates the location of
the optimum. But self-adaptation also has drawbacks, e.g. the premature step
size reduction of ES at the constraint boundary.

Chapter IV: Self-Adaptive Biased Mutation for Evolution Strategies

The Gaussian mutation operator of ES exhibits the feature of unbiasedness of its
mutations. But offering a bias to particular directions can be a useful degree of