You are on page 1of 9

Department of Information Systems and Computing BSc (Hons) Computer Science

Academic Year 2011-2012


Artificial Intelligence and Combinatorial OPtimisation Problems Akuha Aondoakaa (0926840) Shengxiang Yang

ABSTRACT:
The project based on the topic Ant Colony Optimisation (ACO) on the Travelling sales man problem (TSP). First of all, the Ant Colony Optimisation (ACO) is a populationbased approach for solving combinatory optimisation problems (example Travelling Sales man Problem). The ant colony optimisation came as a result of the studying of real ant behaviour in their natural habitat as they find the shortest part to their food source. Dr. Marco Dorigo originally researched on the computational intelligence approaches to combinatorial optimisation, in collaboration with Alberto Colorni and Vittorio Maniezzo with resulted to the Ant Colony Optimisation (ACO). The ACO is to be tested on a classic Travelling Sales man Problem that is all about the distance between the cities been independent of the direction of traversing arc. The expected out come of this project is to make or write a working ACO algorithm in java and also test it on the T.S.P to is it finds the best solution for the problem which is finding the shortest path.

BACK GROUND:
The problem this research will be addressing is how we can apply Ant Colony Optimisation (ACO) on the Travelling Salesman Problem (TSP). The Travelling Salesman Problem (TSP) is an NP-hard problem in combinatorial optimisation studied in operations research and theoretical computer science, Intuitively, the TSP is a classic problem for a salesman who is looking for the shortest path that will take him to all costumer cites starting from his home and back, visiting each city once, which is known by others as Hamilton circle or a Tour. The Travelling Salesman Problem (TSP) plays an important role in the research of Ant Colony Optimisation (ACO), which was first called Ant System according to Marco Dorigos book called Ant colony optimization. Ant Colony Optimisation can be applied directly to TSP. When using the ACO tours are constructed by applying the following three procedure to each ant according to Marco Dorigo: (1) choose, according to some criterion, a start city at which the ant is positioned; (2) use pheromone and heuristic values to probabilistically construct a tour by iteratively adding cities that the ant has not yet visited yet, until all cities have been visited; and (3) go back to the initial city.

AIMS AND OBJECTIVES:


In this section the aims and objectives on the topic Ant Colony Optimisation for the Travelling Salesman Problem are as follows, Design of an Ant Colony Optimisation algorithms and its metaheuristics: Frist of all a metaheuristics in computer science is the design of s computational method that optimizes a problem by repeatedly trying to improve a candidate solution solution with regard to a given degree of quality or worth. Combinatorial optimization problems are challenging in the sense that they are often easy to state but extremely hard to solve and therefore, most problems arising in the applications are NPhard, which implies with polynomially bounded computation time. Now a days, many researchers have focused their attention on new class of algorithms, called metahuristics according to Marco Dorigo, in which I agree with due to the extensive research I have seen and the mass demand of real life situations that needs the algorithm, examples are, vehicle routing and mapping. Ant system or later called Ant Colony Optimization is a successful metahruristic, which is inspired by the behaviour of real ants. The Ant Colony Optimization (ACO) metaheuristic has been suggested as a common framework or backbone for many existing applications and different algorithm versions of ant algorithms. In this chapter we are going to talk about the ACO and its

representation in particular. In Ant Colony Optimisation (ACO) an artificial ant is present, which is a stochastic constructive procedure that is increasing gradually by regular degrees or additions in building a solution by summing up opportunely defined solution under development, therefore, the ACO metaheuristic can be applied to all combinatorial optimisation problem for which a constructive heuristics can be defined. ACO algorithms can be assumed by the interplay or mixture of three procedures namely, constructions Ant Solution, Update Pheromones, and Daemon Actions in accordance to Marco Dorigo. Construct Ant Solutions manages a colony or in this case a group of artificial ants that simultaneously and asynchronously visit the nearest states of the considered problem by moving through neighbour nodes of the problems construction graph. These Ant movements are affected by the following decision policy of pheromone trails and heuristic information, which is applied stochastically. In this way, ants simultaneously construct solutions to the optimization problem. Immediately an ant has constructed a solution, or while in construction of the solution, the ant evaluates the solution that will be decide how much Pheromone to deposit. Update Pheromone is the process by which pheromone trails are modified or improved. The increase or decrease is either caused by the deposited pheromone by ants on the components or due to evaporation of pheromone. From a practical point of view, those elements or connections that were either used by many ants or that were used by at least one ant or which produce a very suitable solution that will be used again by future ants increases the probability of new deposited pheromone. While in terms of pheromone evaporation enforces a useful form of forgetting. Finally, the Daemon Actions procedure is used to implement centralised actions which cannot be performed by single ants according to Marco Dorigo. The local optimization procedure is as a result of the actions of the daemon or the accumulation of global information that can be used to decide whether it is useful or not to deposit additional pheromone to favour the search process from a nonlocal perspective. The daemon monitors the routs taken by the ant in the colony and selects some or a few ants (those that constructed the best solutions in the algorithms iteration), which are then allowed to deposit additional pheromone on the connection they used. Procedure ACO Metaheuristics Schedule Activities Construct Ant Solutions Update Pheromones Daeomon Actions % optional End- Schedule Activites End- procedure

The Ant Colony Optimization in pseudo code.

The N-P hard problem T.S.P:


This section is focused on the N-P hard problem, Travelling Sales Man (TSP) because its the man problem the Ant Colony Optimization (ACO) algorithm is going to be applied on, in this project. First of all, lets get a deep understanding of what the problem TSP is all about. Given a set of cities with cost between them, the problem here is to find the shortest path in which you can reach all the cities and return to the starting point, which is called a tour or circuit through the cities. This is achieved by the right order or arrangement of the cities been visited in the tour. The problem (TSP) is one of the most popular investigated problems in computational mathematics. It has inspired studies by mathematicians, computer scientists, chemists, physicists, psychologists and a host of non-professional researcher according to David L. Applegate and other wed searches I have done myself. Most educators in schools of all levels use the TSP to introduce discrete mathematics. The TSP problem has been applied in logistics, neuroscience and many other applications, which was acknowledged by David L. Applegate. The name Travelling Salesman Problem (TSP) is a very popular name but its origin is kind of a mystery because, after researching on it, still cant find an authoritative documentation that points the creator of the name and also dont have a clear guess when it first came to use. TSP researchers like Merrill Flood of Princeton University and the RAND Corporation was one of the most influential early researchers. According to most sources, in an interview, Flood covering the Princeton mathematics, made the following comment according to David L. Applegate Developments that started in the 1930s at Princeton have interesting later. For example, Koopmans first became interested in the 48 states Problem of Haster Whitney when he was with me in the Princeton surveys, as I tried to solve the problem in connection with the work of Bob Singleton and me on school bus routing for the state of West Virginia. I dont know who coined the peppeir name Travelling Salesman Problem for Whitneys problem, but that name certainly caught on, and the problem has turned out to be of way fundamental importance. This interview of Flood took place in 1984 with Albert Tucker, posing the questions. Although the origin of the name Travelling Salesman Problem cannot be identified, it is no argument that the name fits the problem of finding the shortest path or route through a number of cities in a region.

TSP and ACO:


ACO algorithm can be applied directly to the TSP problem, for instance if a graph is created, where the lengths and components of the graph are set, all possible tours corresponds to the states of the problem; and the ant construct feasible tours that correspond to permutations of the city indices as a result of the constraint enforced. All ACO algorithms for the TSP all have pheromone trails that are related to the arcs in the graph and the heuristic desirability distance between two cities in the graph or nodes are inversely proportional form the opposite cities as well, example; distance between cities A and B is the same as cities B and A. Tours in the ACO algorithm can be constructed by applying the following constructive procedure to each ant: (1) ants are positioned according to some criterion, which is used as the start city of the tour been made by each ant. (2) The tour been constructed by an ant is as a result of use of pheromone and heuristic values to help decide which city the ant has to choose next in construction of the tour. (3) Go back to the initial city. All ant that have completed their tour might deposit pheromone on the path they followed to construct their tour. In the end the effectiveness the ACO algorithm on the TSP will be compared to other algorithm that can the also used to solve the TSP, like, Simulating Annealing (SA), Simple Hill Climbing Method (SHCM) and others.

PROJECT APPRAOCH:
In this section there are a list of research methods and methodology used in order to achieve the desired objectives and goals of the project and also where these methods will be applied during the course of this project. Firstly, what is research all about? The answer to this is that The following below are the list of research methods and methodology were used and maybe used during the project. 1. Research Problem: in this project the research problem was presented in a list on our school student accounts called blackboard and were expected to pick a topic of our interest, get an insight of what its all about by surfing the web for literature and articles or reading books that relates with the topic. When these processes are completed, the next thing was to meet the supervisor that was in charge or linked with the research problem and have a discussion with him on the topic and if the research problem turns out to be what you had in mind earlier then a proposal can be requested which will be later on accepted by the researcher or student. 2. Research Design: Research design helps inlaying down

directions or paths in which to be followed by the researcher to help improve the quality of the project report, helping the researcher know exactly which path of the project is to be done at a certain stage. These are some of the questions asked to help keep the researcher on the right track of meeting the over all objectives of the projects; (i) what are the objectives of the project? (ii) What are the propositions to be tested? (iii) What methodology is going to be used in the project? These were some of the questions that were asked and by answering them it helping keep the researched on the right track thought the course of the project. 3. Collection of Data: After those steps were completed, data had to be collected to help solve the problem at hand, which is the effect of ACO algorithm of the TSP. The following below are the list of data collecting methods that were used and how, why they were used or going to be used during the course of the project:

Descriptive VS Analytical:
Descriptive research is all about fact findings, its major characteristics as its name implies is about describing the problem at hand, the researcher has no influence or control over variables been researched on and can only report what is going on or has already happened while the Analytical research method used data, facts or information that already exists and is in the researchers disposal to analyse the problem and make a critical evaluation. After clearly knowing where and how those methods above can be applied, for this project, the Analytical method was used and will be used because in the case of the topic ACO effects on the TSP, there is the need to analyse the variables at hand, there will also be the need of using data like distance between two cites and also data like the number of cities to be visited by an ant in the ACO on the TSP problem. There fore more critical evaluation and report will be presented when the ACO is fully implemented and functional.

Applied Research:
Applied research aims at getting results for real world problems faced by a society or an organisation This method of research was also chosen because the ACO algorithm is widely used in the society, not only for solving the TSP problem but also used involving other problems like, vehicle routing, scheduling problems, set problems, and many others. For this project, the ACO algorithm is strictly going to be applied on the TSP problem to help meet the objectives of the project, in this case the real life problem will be the TSP problem, so data will be collected to help achieve this aim.

Quantitative VS Qualitative:

Qualitative is more about judging with amount or quantity; it can be applied on phenomena that are expressed by quantity. On the hand qualitative research is the total opposite and therefore concerned with quality of a phenomenon. By their definitions the qualitative method was used or will be used because the ACO is an algorithm that can be analysed in quality rather than quantity in terms of comparison to other TSP solving algorithms like the Simulating Annealing (SA) and the Simple Hill Climbing Method (SHCM), which will help achieve one of the objective of this project which is to analyse the effects of ACO algorithm on the TSP problem and also compare with other algorithms.

Methodology:
The methodology been used during the course of this project is the Water Fall Methodology because in the process of competing the project in time, the project has been divided into steps according to the Water fall which are; (i) project analyses, which involves pointing out the requirements and objectives of the project, (ii) the design phase, which outlines the method going to be used; for example, java language is going to be used for implantation of code in this project, (iii) implementation phase, (iv) testing, this phase is all about testing the java code been implemented to see if it works correctly and can not be skipped on less all the implementations work , and finally, (v) evaluation phase, which ensures everything is running accordingly or as expected. This methodology is used because I wish to avoid confusion and also follow a very simple order.

PROGRAMME OF WORK:
In section is all about the project plan and objectives that are going to be completed with the help of my supervisor. Task one Literature acquiring Requirements Design framework Term 2. Task two- presentation Implementation & testing Milestone Evaluation. Task 3- Reporting & submission Software demonstration.

ETHICAL CONSIDERATIONS ON THE PROJECT:


The following below are the list of ethical considerations been used during the course of this project. 1.Planning Research: 2.Responsibity:

3.Reporting Results and Plagiarism: 4.Instutusional Approval:

BIBLOGRAPHY:

Kkk approch Supply a simple, concise description of the project and its expected outcomes. http://code.ulb.ac.be/dbfiles/Dar2003master.pdf backgrod Introduce the problem which your research will be addressing, explain its academic and/or industrial context, demonstrate some knowledge and understanding of past and current work in the subject area and provide a justification for undertaking the

research. Wiki. Aim & objectives Ref. http://en.wikipedia.org/wiki/Metaheuristic


Define the overall aims of the project and the specific measurable objectives which you plan to achieve. The aim should be general and broad. It can be phrased in terms of solving a practical problem; answering a question; or improving a user task or business in some way. The objectives should be specific and measurable; they should help structure your project and provide a means of evaluating whether it achieved its stated intentions.

Approach Program frame work.


describe the tasks you need to conduct in order to complete the project and identify the major milestones by which you and your supervisor will monitor the progress of your project (e.g. completion of literature survey, completion of software development). The programme of work should be illustrated by a simple diagrammatic workplan showing the major tasks and milestones against a timeline.

Ethical.
: highlight ethical considerations/principles that will be taken into account during the project. Ethical considerations in research.

(http://media.wiley.com/product_data/excerpt/28/04713691/047136 9128-3.pdf)

You might also like