Wong Man Leung

Followers

Following

Co-author

Public Views

Geng Cui

Lingnan University

Carlos Cotta

Universidad de Málaga

luis ortega

Thierry BROUARD

Université François-Rabelais, Tours

Hubert Cardot

Université François-Rabelais, Tours

Jiri Ocenasek

Geng Cui

Lingnan University

David Heckerman

Mohammad Homayounpour

AmirKabir University Of Technology

A. Khanteymoori

Interests

Uploads

Papers by Wong Man Leung

An Efficient Data Mining Method for Learning Bayesian Networks Using an Evolutionary Algorithm-Based Hybrid Approach

IEEE Transactions on Evolutionary Computation, 2004

Given the explosive growth of data collected from current business environment, data mining can p... more Given the explosive growth of data collected from current business environment, data mining can potentially discover new knowledge to improve managerial decision making. This study proposes a novel data mining approach that employs an evolutionary algorithm to discover knowledge represented in Bayesian networks. The approach is applied successfully to handle the business problem of finding response models from direct marketing data. Learning Bayesian networks from data is a difficult problem. There are two different approaches to the network learning problem. The first one uses dependency analysis, while the second one searches good network structures according to a metric. Unfortunately, both approaches have their own drawbacks. Thus, we propose a novel hybrid algorithm of the two approaches, which consists of two phases, namely, the Conditional Independence (CI) test and the search phases. In the CI test phase, dependency analysis is conducted to reduce the size of the search space. In the search phase, good Bayesian network models are generated by using an evolutionary algorithm. A new operator is introduced to further enhance the search effectiveness and efficiency. In a number of experiments and comparisons, the hybrid algorithm outperforms MDLEP, our previous algorithm which uses Evolutionary Programming (EP) for network learning, and other network learning algorithms. We then apply the approach to two data sets of direct marketing and compare the performance of the evolved Bayesian networks obtained by the new algorithm with those by MDLEP, the logistic regression models, the naïve Bayesian classifiers, and the tree-augmented naïve Bayesian network classifiers (TAN). In the comparison, the new algorithm outperforms the others.

Download

Discover dependency pattern among attributes by using a new type of nonlinear multiregression

International Journal of Intelligent Systems, 2001

Multiregression is one of the most common approaches used to discover dependency pattern among at... more Multiregression is one of the most common approaches used to discover dependency pattern among attributes in a database. Nonadditive set functions have been applied to deal with the interactive predictive attributes involved, and some nonlinear integrals with respect to nonadditive set functions are employed to establish a nonlinear multiregression model describing the relation between the objective attribute and predictive attributes. The values of the nonadditive set function play a role of unknown regression coefficients in the model and are determined by an adaptive genetic algorithm from the data of predictive and objective attributes. Furthermore, such a model is now improved by a new numericalization technique such that the model can accommodate both categorical and continuous numerical attributes. The traditional dummy binary method dealing with the mixed type data can be regarded as a very special case of our model when there is no interaction among the predictive attributes and the Choquet integral is used. When running the algorithm, to avoid a premature during the evolutionary procedure, a technique of maintaining diversity in the population is adopted. A test example shows that the algorithm and the relevant program have a good reversibility for the data.

Download

Grammar-Based Genetic Programming with Bayesian network

2014 IEEE Congress on Evolutionary Computation (CEC), 2014

Grammar-Based Genetic Programming (GBGP) improves the search performance of Genetic Programming (... more Grammar-Based Genetic Programming (GBGP) improves the search performance of Genetic Programming (GP) by formalizing constraints and domain specific knowledge in grammar. The building blocks (i.e. the functions and the terminals) in a program can be dependent. Random crossover and mutation destroy the dependence with a high probability, hence breeding a poor program from good programs. Understanding on the syntactic and semantic in the grammar plays an important role to boost the efficiency of GP by reducing the number of poor breeding. Therefore, approaches have been proposed by introducing context sensitive ingredients encoded in probabilistic models. In this paper, we propose Grammar-Based Genetic Programming with Bayesian Network (BGBGP) which learns the dependence by attaching a Bayesian network to each derivation rule and demonstrates its effectiveness in two benchmark problems.

Data mining in marketing using Bayesian networks and evolutionary programming

Give the explosive growth of customer data collected electronically from current electronic busin... more Give the explosive growth of customer data collected electronically from current electronic business environment, data mining can potentially discover new knowledge to improve managerial decision making in marketing. This study proposes an innovative approach to data mining using Bayesian Networks and evolutionary programming and applies the methods to marketing data. The results suggest that this approach to knowledge discovery can generate superior results than the conventional method of logistic regression. Future research in this area should devote more attention to applying this and other data mining methods to solving complex problems facing today\u27s electronic businesses

Parallel evolutionary algorithms on consumer-level graphics processing unit

In this research, we have implemented a parallel EP on consumer-level graphics processing units a... more In this research, we have implemented a parallel EP on consumer-level graphics processing units and proposed indirect indexing and many optimization skills to achieve maximal efficiency. The parallel EP is a hybrid of masterslave and fine-grained models. Competition and selection are performed by CPU (i.e. the master) while fitness evaluation, mutation, and reproduction are performed by GPU which is essentially a massively parallel machine with shared memory. Unlike other fine-grained parallel computers such as Maspar, GPU allows processors to communicate with, not only nearby processors, but also any other processors. Hence more flexible fine-grained EAs can be implemented on GPU. We have done experiments to compare our parallel EP on GPU and an ordinary EP on CPU. It is found that the speed-up factor of our parallel EP ranges from 1.25 to 5.02, when the population size is large enough. Moreover, there is a sub-linear relation between the population size and the execution time. Thu...

Evolutionary algorithms for data mining

Artificial evolution can be applied to machine learning thanks to genetic programming, for instan... more Artificial evolution can be applied to machine learning thanks to genetic programming, for instance, which is a very successful branch of Evolutionary Computing. One would therefore think that by transitivity, because data-mining is one of the main applications of machine learning, artificial evolution could be successful in solving data-mining problems. Whether this is the case or not, it seems that artificial evolution is not much used in data-mining, even though many papers show that EC can provide interesting alternative solutions to standard machine learning approaches

Download

Constrained optimization with genetic algorithm : improving profitability of targeted marketing

by Wong Man Leung and Geng Cui

Direct marketing forecasting models have focused on estimating the response probabilities of cons... more Direct marketing forecasting models have focused on estimating the response probabilities of consumer purchases and neglected the profitability of customers. This study proposes a method of constrained optimization using genetic algorithm to maximize the profitability at the top deciles of a customer list. We apply this method to a direct marketing dataset using tenfold cross-validation. The results from this method compare favorably with the unconstrained model and that of the DMAX model. The method of constrained optimization has distinctive advantages in augmenting the profitability of direct marketing campaigns. We explore the implications for targeted marketing problems and for assisting management decision-making and augmenting profitability of direct marketing

Discovering knowledge from noisy databases using genetic programming

Journal of the American Society for Information Science, 2000

In data mining, we emphasize the need for learning from huge, incomplete and imperfect data sets

Download

Expanding Self-Organizing Map for data visualization and cluster analysis

Information Sciences, 2004

The Self-Organizing Map (SOM) is a powerful tool in the exploratory phase of data mining. It is c... more The Self-Organizing Map (SOM) is a powerful tool in the exploratory phase of data mining. It is capable of projecting high-dimensional data onto a regular, usually 2-dimensional grid of neurons with good neighborhood preservation between two spaces. However, due to the dimensional conflict, the neighborhood preservation cannot always lead to perfect topology preservation. In this paper, we establish an Expanding SOM (ESOM) to preserve better topology between the two spaces. Besides the neighborhood relationship, our ESOM can detect and preserve an ordering relationship using an expanding mechanism. The computation complexity of the ESOM is comparable with that of the SOM. Our experiment results demonstrate that the ESOM constructs better mappings than the classic SOM, especially, in terms of the topological error. Furthermore, clustering results generated by the ESOM are more accurate than those obtained by the SOM.

Download

An Efficient Data Mining Method for Learning Bayesian Networks Using an Evolutionary Algorithm-Based Hybrid Approach

IEEE Transactions on Evolutionary Computation, 2004

Download

Discover dependency pattern among attributes by using a new type of nonlinear multiregression

International Journal of Intelligent Systems, 2001

Download

Grammar-Based Genetic Programming with Bayesian network

2014 IEEE Congress on Evolutionary Computation (CEC), 2014