Posted: November 2nd, 2015

Assignment Title

Student no.

1 INTRODUCTION

Concise description of task asked and your approach(es).

2 BACKGROUND RESEARCH

Evolutionary computing has been used in optimization in many ways. Give a brief summary of some examples you can find, connecting them to your assignment perhaps too. Use clear referencing and avoid including material directly unless carefully indicated as from another source. Half to a whole page expected.

3 EXPERIMENTATION

3.1 Function 1

Describe your basic algorithm so anyone who knows about basic GAs could hopefully repeat what you have done. Describe the representation, the parameters, the fitness calculation, etc.

Present your results as averaged behaviour over more than one run, eg (graph not of assignment):

Figure 1: Initial performance on task 1.

Show the effects of varying parameters and give potential explanations as to why the behaviour/performance changes. Eg vary the mutation rate or population size.

3.2 Function 2

Again, show results from varying parameters/aspects and explain them. The use of different selection and crossover mechanisms, and/or other modifications, would be good here.

3.3 Function 3

Clearly describe how you changed the representation to deal with the real-valued function. And then present results from its use, with graphs and explanations.

For the very keen, it would be great to see you using other well-known functions from the literature, comparing performance, and/or implementing other mechanisms or algorithms.

4 CONCLUSIONS

Concise summary of what you found and learned. Identification of ways you might do things differently next time, and why.

REFERENCES

Bernado Mansilla, E. & Garrell, J. (2003) Accuracy-Based Learning Classifier Systems: Models, Analysis and Applications to Classification Tasks. *Evolutionary Computation* 11(3): 209-238.

And please include either a link to your source code so that more than one person can access it over the next few months or as a full listing in an Appendix. The second marker and external examiner need to be able to see what you actually produced.

MODULAR PROGRAMME

COURSEWORK ASSESSMENT SPECIFICATION

Module Details

Module Code UFCFY3-15-3

Run 15SEP/1

Module Title ADVANCES IN ARTIFICIAL INTELLIGENCE

Module Leader Larry Bull

Module Coordinator

Module Tutors

Component and Element Number B: CW1

Weighting: (% of the Module’s assessment) 50

Element Description PRACTICAL ASSIGNMENT REQUIRING THE PRODUCTION OF PROGRAM CODE (Practical Assignment Requiring the Production of Program Code)

Total Assignment time

Dates

Date Issued to Students 19/10/15

Date to be Returned to Students 11/12/15

Submission Place

Blackboard

Submission Date 26/11/2015

Submission Time 2.00 pm

Deliverables

Report in specified format, source code.

Module Leader Signature

UFCFY3-15-3 ADVANCES IN AI: Assignment

[Hand-in Deadline: November 26th 2015]

Write a report on your attempts to solve a set of optimization problems as effectively as possible using any evolutionary computing mechanisms covered in the course, building upon your own genetic algorithm code developed in the first few lab sessions.

Function 1:

f(x) = x2 Where x is an integer in the range 0-255, i.e., 0≤ x ≤ 255

Function 2:

f(x,y) = 0.26.( x2 + y2 ) – 0.48.x.y Where -15 ≤ x,y ≤ 15

Function 3:

n

f(x) = 10n + xi2– 10.cos(2.xi) Where -5.12 ≤ xi ≤ 5.12, and use n=10, 20

i=1

The worked example of solving f(x) = x2 in the first lecture on evolutionary algorithms will provide you with a starting point for Function 1. Function 1 is a maximization problem, whereas Functions 2 and 3 are minimization problems.

Approaches to consider alongside parameter varying include different types of selection and crossover. Suggested further extensions are other standard benchmark functions from the literature, permutation problems, non-stationary problems, and/or implementing other metaheuristics, adding local search or parameter adaptation, etc. Generally, more marks will be given to the effective use of more sophisticated approaches.

Include a research section which briefly reviews optimization and evolutionary computing therein. In an experimentation section, describe the encoding and operators used, show and explain example runs with various parameter settings, and solutions found.

Outline Marking Scheme:

Assignments will be assessed according to the following criteria:

General approach, quality of writing and visual impression (10%)

Research (20%), Experimental Method (30%), Analysis and Discussion (30%)

Conclusions (5%)

Citation and Reference Scheme (5%)

Guideline:

Depending on font size, and line spacing, around 8 pages is a reasonable target length. The intention is your hand-in approximates to a research paper. Please use the report style file and include commented source code as an Appendix or via an active link.

Place an order in 3 easy steps. Takes less than 5 mins.