Stochastic Optimization Algorithms

Written by
  1. Recent updates
  2. Course description
  3. Schedule
  4. Problem sets
  5. Exams
  6. Links

  

1. Recent updates

This course is held once a year in study period 1. The next course will start in September 2015.

 

2. Course description

Course specific prerequisites

Programming, basic engineering mathematics.

Aim

The aim of the course is for the students to attain basic knowledge of new methods in computer science inspired by evolutionary processes in nature, such as genetic algorithms, genetic programming, and artificial life. These are both relevant to technical applications, for example in optimization and design of autonomous systems, and for understanding biological systems, e.g., through simulation of evolutionary processes.

Learning outcome (after completion of this course, the student should be able to)

 

    • Implement and use several different classical optimization methods, e.g. gradient descent and penalty methods.

 

    • Describe and explain the basic properties of biological evolution, with emphasis on the parts that are relevant for evolutionary algorithms.

 

    • Define and implement (using Matlab) different versions of evolutionary algorithms, particle swarm optimization, and ant colony optimization, and apply the algorithms in the solution of optimization problems.

 

    • Compare different types of biologically inspired computation methods and identify suitable algorithms for a variety of applications.

 

Content

The course consists of the following topics:
- Classical optimization methods. Gradient descent. Convex functions. The lagrange multiplier method. Penalty methods.
- Evolutionary algorithms. Fundamentals of genetic algorithms, representations, genetic operators, selection mechanisms. Theory of genetic algorithms. Analytical properties of evolutionary algorithms. (Linear) genetic programming: representation and genetic operators.
- Particle swarm optimization. Fundamentals and applications.
- Ant colony optimization. Fundamentals and applications.
- Comparison of the different algorithms

Organisation

The course is organized as a series of lectures. Some lectures are devoted to problem-solving.

Literature

Wahde, M. Biologically inspired optimization methods: An introduction

Examination

The examination is based on a written exam and home problems.

3. Schedule

Please have a look on the external course webpage for more information.

4. Problem sets

Please have a look on the external course webpage for more information.

5. Exams

Please have a look on the external course webpage for more information.

6. Links

External course webpage: http://www.me.chalmers.se/~mwahde/courses/soa/2015/FFR105.html

 

Mattias Wahde

Professor Mattias Wahde is researcher and teacher within the research group Adaptive Systems.The ultimate aim of his research is to generate autonomous robots capable of carrying out a variety of relevant tasks, particularly dangerous or tedious tasks which are presently carried out by people. His research is focused on generating robotic brains (control systems) rather than hardware (robots). In particular, he is developing a method (the utility function method) for behavioral selection. This method, as well as his research in general, is based on biologically inspired computation methods, particularly evolutionary algorithms (EAs).

Position: Professor