I=
Pakistan Research Repository Home
 

Title of Thesis

Combining Pso Algorithm and Honey Bee Food Foraging Behavior for Solving Multimodal and Dynamic Optimization Problems

Author(s)

Muhammad Rashid

Institute/University/Department Details
Department of Computer Science / National University of Computer & Emerging Sciences, Islamabad
Session
2010
Subject
Computer Science
Number of Pages
180
Keywords (Extracted from title, table of contents and abstract of thesis)
Combining, Pso, Algorithm, Honey Bee, Food, Swarm, Foraging, Behavior, Multimodal, Dynamic, Optimization, neural, networks

Abstract
Swarm intelligence algorithms are taking the spotlight in the field of function optimization. In this research our attention centers on combining the Particle Swarm Optimization (PSO) algorithm with food foraging behavior of honey bees. The resulting algorithm (called HBF-PSO) and its variants are suitable for solving multimodal and dynamic optimization problems. We focus on the niching and speciation capabilities of these algorithms which allow them to locate and track multiple peaks in environments which are multimodal and dynamic in nature. The HBF-PSO algorithm performs a collective foraging for fitness in promising neighborhoods in combination with individual scouting searches in other areas. The strength of the algorithm lies in its continuous monitoring of the whole scouting and foraging process with dynamic relocation of the bees (solution/particles) if more promising regions are found. We also propose variants of the algorithm in which each bee has a different position update equation and we utilize genetic programming (GP) for continuous evolution of these position update equations. This process ensures adaptability and diversity in the swarm which leads to faster convergence and helps to avoid premature convergence. We also explore the use of opposite numbers in our algorithm and incorporate opposition based initialization, opposition based generation jumping and opposition based velocity calculation. The proposed algorithm and its variants are tested on a suite of benchmark optimization problems. In the final portion of our work we report our experiments on the training of feedforward neural networks utilizing our proposed algorithms.

Download Full Thesis
4,416 KB
S. No. Chapter Title of the Chapters Page Size (KB)
1 0 CONTENTS

 

iv
62 KB
2

1

INTRODUCTION

1.1 Swarm Intelligence for Optimization problems
1.2 Particle Swarm Optimization
1.3 Types of Optimization Problems
1.4 Research Overview
1.5 Contribution
1.6 Thesis Organization

1
741 KB
3 2 HONEY BEE FORAGING BEHAVIOR INSPIRED PSO ALGORITHM

2.1 Introduction
2.2 Subswarms and Niching
2.3 Honey Bee Foraging Behavior Inspired Particle Swarm Optimization Algorithm (HBF-PSO)

9
168 KB
4 3 EXPERIMENTS ON HBF-PSO

3.1 Single Optimum in Multimodal Environments
3.2 Multiple Optima in Multimodal Environments
3.3 Dynamic Optimization Experiments

25
1,583 KB
5 4 SWARM AREA, SWARM SIZE AND NUMBER OF SWARMS

4.1 Self Adjusting Neighborhood (HBF-PSOn)
4.2 Species Identification Algorithm

39
194 KB
6

5

GP BASED ADAPTABLE VELOCITY UPDATE EQUATIONS

5.1 Introduction
5.2 Previous Work
5.3 Adaptable Velocity Update Equations
5.4 Adaptable and Non-Adaptable Velocity Update Equations
5.5 Adaptable and Non Adaptable Velocity Update Equations with Species Seeds Algorithm
5.6 SPSOGP embedded in HBF framework
5.7 Usage Frequency of Update Equations

53
389 KB
7

6

INCORPORATION OF OPPOSITION BASED LEARNING

6.1 Opposition-based Learning
6.2 Existing Approaches Utilizing Opposition-based Learning with PSO
6.3 Proposed Method
6.4 Experiments and Results

86
304 KB
8

7

FEEDFORWARD NEURAL NETWORK TRAINING

7.1 Feedforward Neural Networks
7.2 Training Feedforward Neural Networks with PSO
7.3 Experiments and Results

103
201 KB
9

8

CONCLUSION

8.1 Comparison and Analysis
8.2 Achievements
8.3
Future research directions

110
146 KB
10 9 REFERENCES

119


107 KB