Evolutionary Computation

1
Evolutionary Computation
Learning outcomes

• Overview,
• Simple evolutionary algorithm
• Differentiating the techniques
• Multi-modal-objective
• Mutation effects
• Building blocks
• Fitness function
• Representation
• Explore v exploit
• Hybridise (or die)
• Presentation competition

2
Evolutionary Computation
Learning instances

• Simple Evolutionary Algorithm
• Searching for techniques
• Advanced Genetic Algorithm
• Evolutionary Strategies
• Estimation of Distribution Algorithm
• Evolutionary Programming
• Genetic Programming
• Learning Classifier Systems
• Memetic algorithms
• Your choice!
• Many many more instances and are worth exploring

3
Representation

• Genetic information can be any symbol.
• Require dictionary and rules to manipulate the
  symbols.
• Some symbols make the search space easier to
  explore/exploit
• Some symbols are easier to store, manipulate and
  test
• Different problems may be suited to different
  symbol sets

4
Multiple Representations

• Binary, ternary, Grey or enumerated.
• Integer, real, floatingpoint or mantisssa
• Rank, order, series, histogram or array
• Bounded,
• Horn clauses and second-order logic
• S-type expressions
• Problem specific

5
Genetic Programming

• Genetic information can be any symbol.
• Require dictionary and rules to manipulate the
  symbols.
• - /
• AND, NOT, NOR
• Input, Output, Variable
• Derivative, Integrator, Lead, Lag
• Gain, Add-block, Junction

6
Genetic Programming

• Genetic information expressed in a symbolic
  decision tree

7
Genetic Programming

• Genetic information can be hard to interpret!

8
Genetic Programming

• Genetic operators adjusted to new symbols
• E.g. Lead mutates to Lag
• Blocks of symbols crossed over
• String size can increase and decrease.

9
Genetic Programming

• G(s) K (1 ? s) -3
• Task Design controller including
• Topology and parameter values.
• Minimise the integral of the
• time-weighted absolute error,
• Overshoot in response to step input is
• lt 2
• Controller functions over a range of
• K and ?.

10
Fitness Function

• 10 Elements
• Time to reach reference signal
• Avoidance of overshoot
• Settling time
• Variance in a reference signal
• Robustness to change in ?
• Robustness to change in K
• Response to step input
• Response to ramp input
• Stability with extreme spiked signal
• Frequency response
• Fitness is sum of elements.

11
Set-up

• Population size 66,000
• Maximum size of string 150
• Manually terminated
• Large Beowulf computational power
• Only 31 generations to result

12
Controller
13
Time-Domain Response
14
Comparison
15
Comparison
16
Websites of the week

• http//www.genetic-programming.org/
• (not http//www.geneticprogramming.com/)
• http//liinwww.ira.uka.de/bibliography/Ai/genetic.
  programming.html

17
Bloat!

• Very computationally expensive and
• The technique suffers from BLOAT
• The genome can expand without bound
• Redundant code can actually help an offspring
  survive
• Multiple techniques have been introduced to
  control bloat
• Still an ongoing research area