|
NAD - A Discretization Approach for Improving Interdependency
|
|
E. Chandra Blessie, E. Karthikeyan, B. Selvaraj
|
|
Volume 2, Issue 1, 2010
pp.
9 - 17
|
|
Received
11 December 2009,
Accepted
09 March 2010
|
|
Abstract.
Many classification algorithms need continuous features to be discretized. We propose a new algorithm NAD (Normalized And Discretization method) for improving the interdependency between continuous and class features. Empirical comparison of five existing algorithms with the proposed algorithm showed that the proposed algorithm generate better result based on time complexity and storage space.
|
|
Keywords.
Classification algorithms; Discretization method; Continuous features.
|
|
Full Text:
PDF
|
|
|
A Crossbreed Quantum Evolutionary Algorithm
|
|
H.K. Khalafi
|
|
Volume 2, Issue 1, 2010
pp.
1 - 8
|
|
Received
25 July 2009,
Accepted
17 November 2009
|
|
Abstract.
Quantum Clonal Algorithms (QCAs) are based on the automatic self-protective behavior of the quantum evolutionary algorithms (QEAs). QCAs have high effective parallelism and replacing clone operator by mutation and selection of the traditional evolutionary algorithms, which can increase the variety and avoid hurriedness. However, the intricacy of the QCAs is growth at the same time, which will affect the efficiency of the QCAs. In this paper, we suggest a Crossbreed Quantum Evolutionary Algorithm (CQEA) which integrates Quantum Evolutionary Algorithms, with the Particle Swarm Optimization algorithm.
|
|
Keywords.
Quantum computing; Particle swarm optimization; Evolutionary algorithm; Clonal algorithms.
|
|
Full Text:
PDF
|
|
|
A Computational Approach of Lexical Refinement
|
|
Feng-Jen Yang
|
|
Volume 1, Issue 1, 2009
pp.
56 - 76
|
|
Received
05 August 2009,
Accepted
24 October 2009
|
|
Abstract.
Lexical refinement is a post process to refine machine planned discourses and make machine generated languages more coherent and more fluent. Overlooking this process, a system can still generate comprehensible languages but may sound unnatural and sometimes frustrate its users. To this end, generating coherent and natural sounding languages is a major concern in natural language systems. In this research, a lexical refinement approach is presented to improve the coherence and fluency of natural language generations.
|
|
Keywords.
Lexical refinement; Lexical selection; Lexicalization.
|
|
Full Text:
PDF
|
|
|
Visualization of Corpus Category Information using Gestalt Visual Perception Model
|
|
T. Mala, T.V. Geetha
|
|
Volume 1, Issue 1, 2009
pp.
44 - 55
|
|
Received
06 February 2009,
Accepted
22 September 2009
|
|
Abstract.
Visualization of a set of documents is a challenging task. There are various aproaches to visualize corpus of newspaper documents. In this paper we use a new method called the blobby modeling approach to visualize the category information of a set of documents. This blobby model is designed based on a novel Gestalt Visual Perception Model which uses Gestalt Perception Principles and a set of mathematical mapping functions to map the Category information to the blobby objects. First SRP algorithm is used to categorize Tamil newspaper articles. SRP algorithm is a Set, Rank and Priority based classifier which can classify multi category documents also. The categorized documents are visualized using blobby objects. Blobby objects or metaballs are implicitly modeled objects and they are dynamic in nature. Their dynamism makes them move randomly in different directions and they combine together to form blobbies. These blobbies are modeled using Gestalt perception model which transforms the category weight information obtained as output from SRP algorithm into visual forms. In this paper corpus of Tamil documents are categorized and the category weight information is used to create visual models. The visual generated follow Gestalt Perception laws. Gestalt Visual perception model helps us to view the category information in visual models by applying laws of Similarity and Continuity. These laws are also used to analyse and perceive the visual models. Multimedia elements are added to the visual model and also animated to make the visual more realistic.
|
|
Keywords.
Category Visualization; Gestalt Perception Model; Blobby object.
|
|
Full Text:
PDF
|
|
|
Software Agents as a Simulation Tool to Study Aggregate Consumers' Behaviors in Market Places
|
|
Filippo Neri
|
|
Volume 1, Issue 1, 2009
pp.
32 - 43
|
|
Received
11 April 2009,
Accepted
15 June 2009
|
|
Abstract.
An agent based tool for analyzing the aggregate behavior of fast moving consumer goods markets under several scenarios is described. This simulation methodology allows for the study of tradeoffs among several variables of information like product advertisement, consumers' memory capability, and word of mouth among friends in determining market shares. Insights gained by using this approach on an hypothetical marketplace are reported.
|
|
Keywords.
Software agents; market simulation; Organizational dynamics.
|
|
Full Text:
PDF
|
|
|
A New Optimization Model for Solving the Constraint Satisfaction Problem
|
|
Mohamed Ettaouil, Chakir Loqman
|
|
Volume 1, Issue 1, 2009
pp.
13 - 31
|
|
Received
15 March 2009,
Accepted
04 June 2009
|
|
Abstract.
A constraint satisfaction problem (CSP) involves the assignment of values to variables which are subject to a set of constraints. In this paper, we propose a new modelization of the binary CSP as 0-1 quadratic programming (QP)which consists in minimizing a quadratic function subject to linear constraints QP. Our model puts in interaction all constraints, and then the problem can be considered globally during the search. We apply a general method to solve QP based on semidefinite relaxations. We show that this method is able to determine whether a CSP has a solution or not. Then, we explain how we can apply this method to solve a non-binary CSP after converting the latter into an equivalent binary CSP. Furthermore, we give some hints on how to reduce as much as possible the initial size of the CSP. Some numerical examples and computational experiments assess the effectiveness of the theoretical results shown in this paper, and the advantages of the new modelization.
|
|
Keywords.
Constraint satisfaction problem; Quadratic 0-1 programming; Semidefinite relaxations; Filtering algorithms.
|
|
Full Text:
PDF
|
|
|
ClassifyChi2: A New Hybrid Discretization Algorithm
|
|
Saleh M. Abu-Soud, Ahmad M. Odat
|
|
Volume 1, Issue 1, 2009
pp.
1 - 12
|
|
Received
07 November 2008,
Accepted
24 February 2009
|
|
Abstract.
In this paper we consider new weak and stronger forms of contra $\delta$-precontinuous via the concept of g$\delta$p-closed sets which we call ap-$\delta$-precontinuous, and perfectly $\delta$-precontinuous maps. Moreover, we use ap-$\delta$-precontinuous to obtain a characterization of $\delta$-preT$_{1/2}$ spaces. Also we introduce the class of functions which is called completely $\delta$-precontinuous maps and we obtain some their basic properties. Discretization is a technique that converts continuous variables into discrete ones for inductive machine learning and data mining systems. This paper presents a new hybrid discretization algorithm called ClassifyChi2. ClassifyChi2 depends on chi-square statistical methods. The new algorithm is supervised, static, global, and belongs to the merging methods. It is compared with some well known discretization algorithms through applying the algorithms on some data sets of different sizes over three measures: starting number of intervals, number of cut points with high accuracy, and execution time.
|
|
Keywords.
Discretization; Learning by example; Inductive machine learning; Data mining; Machine learning.
|
|
Full Text:
PDF
|
|
|
A Novel Self-Adaptive Behavior QEA
|
|
H.K. Khalafi
|
|
pp.
|
|
Received
22 June 2009,
Accepted
|
|
Abstract.
In accordance with tradition quantum evolutionary algorithms can obtain excellent results in the optimization of multi-peak functions. In any case, they are easy to be trapped to hurriedness. In this article, A Novel Self-adaptive BQEA is recommending on the basis of the concepts and tenet of QEA, in order to enhance the efficiency. Firstly, Self-adaptive behavior triploid chromosome is constructed to keep the population variety; Secondly, double mutation is used to make sure the variety of the swarm, then individual chromosome cross will be imported into this new algorithm in order to achieve the information communication between the chromosomes and enlarge the search scope in the available space. Experiments on test functions of varied intricaciesare implemented and compared with other EAs. The result indicates that the newalgorithm in this article can search and get the global most efficient solution in a shorter time.
|
|
Keywords.
Self-adaptive behavior; Evolutionary algorithm; Quantum computing; Discrete cross.
|
|
Full Text:
PDF
|
|