Genetic Algorithms (GA)

grandgoatΤεχνίτη Νοημοσύνη και Ρομποτική

23 Οκτ 2013 (πριν από 3 χρόνια και 11 μήνες)

117 εμφανίσεις

Genetic Algorithms (GA)

Vavilin Andrey {andy@ulsan.islab.ac.kr}

2

Intelligent Systems Lab.

What is GA?

GA is an heuristic search algorithm which generates solutions to optimization problems using
techniques inspired by natural evolution, such as inheritance, mutation, selection, and crossover.


Problem domain

-

graph
-
based problems (e.g. traveling salesman problem)

-

global optimization problems

-

scheduling and task planning problems

-

artificial intelligence tasks

-

computer vision

-

etc


3

Intelligent Systems Lab.

Function minimization example

-
2

-
2

2

2

4

Intelligent Systems Lab.

Function minimization

Gradient descent

Best point:
-
3.567

Coordinates: 1.823, 1.549

-
2

-
2

2

2

5

Intelligent Systems Lab.

Function minimization

Random search

Iterations: 5000

Best point:
-
3,560

Coordinates:
-
1.899,
-
1.639

-
2

-
2

2

2

6

Intelligent Systems Lab.

Function minimization

Genetic algorithm

Iterations: 200

Best point:
-
3,949

Coordinates:
-
2,
-
1.960

-
2

-
2

2

2

7

Intelligent Systems Lab.

Typical genetic algorithm

Population

Parents

Offspring

Recombination and
mutation

Parent selection

Survivor selection

initialization

termination

8

Intelligent Systems Lab.

Image processing examples

P.W.M. Tsang and Z. Yu,
“Genetic algorithm for model
-
based matching of projected images of three
-
dimensional objects”,

IEE Proceedings on Vision, Image and Signal Processing, vol.150, issue 6, pp.351
-
359, Dec. 2003

9

Intelligent Systems Lab.

Image processing examples

P.W.M. Tsang and Z. Yu,
“Genetic algorithm for model
-
based matching of projected images of three
-
dimensional objects”,

IEE Proceedings on Vision, Image and Signal Processing, vol.150, issue 6, pp.351
-
359, Dec. 2003

10

Intelligent Systems Lab.

Image processing examples

P.W.M. Tsang and Z. Yu,
“Genetic algorithm for model
-
based matching of projected images of three
-
dimensional objects”,

IEE Proceedings on Vision, Image and Signal Processing, vol.150, issue 6, pp.351
-
359, Dec. 2003

11

Intelligent Systems Lab.

Conclusions

Advantages:

-

Easy to implement

-

Better than random search and faster than brute force algorithm

-

Good for various classes of problems

-

Easy to use with GPU
-
based computation


Weak points

-

Specialized algorithms provide better solutions

-

GA do not scale well with increasing complexity

-

Bad implementation may cause algorithm converges to a local optima instead of a global one

12

Intelligent Systems Lab.

Image processing example

13

Intelligent Systems Lab.

Image processing example

Initial population

14

Intelligent Systems Lab.

Image processing example

Crossover

M1

M2

Initial individuals

Individuals produced by crossover

(changing position)

(changing position and angle)

(changing all)

15

Intelligent Systems Lab.

Image processing example

Mutations

Randomly change random number of parameters in randomly select individuals. Number of
individuals is 5% of population.

16

Intelligent Systems Lab.

Image processing example

Evaluating individuals using NN

Pixel values





Solid model

Edge model

Probability what the tested
individual is arrowhead

17

Intelligent Systems Lab.

Image processing example

NN training

Training set

Pixel values





Solid model

Edge model

18

Intelligent Systems Lab.

False detection example by reference method