Kevin S. Tickle

Association rule mining to detect factors which contribute to heart disease in males and females

This paper investigates the sick and healthy factors which contribute to heart disease for males and females. Association rule mining, a computational intelligence approach, is used to identify these factors and the UCI Cleveland dataset, a biological database, is considered along with the three rule generation algorithms - Apriori, Predictive Apriori and Tertius. Analyzing the information available on sick and healthy individuals and taking confidence as an indicator, females are seen to have less chance of coronary heart disease then males. Also, the attributes indicating healthy and sick conditions were identified. It is seen that factors such as chest pain being asymptomatic and the presence of exercise-induced angina indicate the likely existence of heart disease for both men and women. However, resting ECG being either normal or hyper and slope being flat are potential high risk factors for women only. For men, on the other hand, only a single rule expressing resting ECG being hyper was shown to be a significant factor. This means, for women, resting ECG status is a key distinct factor for heart disease prediction. Comparing the healthy status of men and women, slope being up, number of coloured vessels being zero, and oldpeak being less than or equal to 0.56 indicate a healthy status for both genders.

Computational intelligence for heart disease diagnosis: A medical knowledge driven approach

This paper investigates a number of computational intelligence techniques in the detection of heart disease. Particularly, comparison of six well known classifiers for the well used Cleveland data is performed. Further, this paper highlights the potential of an expert judgment based (i.e., medical knowledge driven) feature selection process (termed as MFS), and compare against the generally employed computational intelligence based feature selection mechanism. Also, this article recognizes that the publicly available Cleveland data becomes imbalanced when considering binary classification. Performance of classifiers, and also the potential of MFS are investigated considering this imbalanced data issue. The experimental results demonstrate that the use of MFS noticeably improved the performance, especially in terms of accuracy, for most of the classifiers considered and for majority of the datasets (generated by converting the Cleveland dataset for binary classification). MFS combined with the computerized feature selection process (CFS) has also been investigated and showed encouraging results particularly for NaiveBayes, IBK and SMO. In summary, the medical knowledge based feature selection method has shown promise for use in heart disease diagnostics.

Solving the traveling salesman problem using cooperative genetic ant systems

The travelling salesman problem (TSP) is a classic problem of combinatorial optimization and has applications in planning, scheduling, and searching in many scientific and engineering fields. Ant colony optimization (ACO) has been successfully used to solve TSPs and many associated applications in the last two decades. However, ACO has problem in regularly reaching the global optimal solutions for TSPs due to enormity of the search space and numerous local optima within the space. In this paper, we propose a new hybrid algorithm, cooperative genetic ant system (CGAS) to deal with this problem. Unlike other previous studies that regarded GA as a sequential part of the whole searching process and only used the result from GA as the input to subsequent ACO iterations, this new approach combines both GA and ACO together in a cooperative manner to improve the performance of ACO for solving TSPs. The mutual information exchange between ACO and GA in the end of the current iteration ensures the selection of the best solutions for next iteration. This cooperative approach creates a better chance in reaching the global optimal solution because independent running of GA maintains a high level of diversity in next generation of solutions. Compared with results from other GA/ACO algorithms, our simulation shows that CGAS has superior performance over other GA and ACO algorithms for solving TSPs in terms of capability and consistency of achieving the global optimal solution, and quality of average optimal solutions, particularly for small TSPs.

Computational intelligence for microarray data and biomedical image analysis for the early diagnosis of breast cancer

The objective of this paper was to perform a comparative analysis of the computational intelligence algorithms to identify breast cancer in its early stages. Two types of data representations were considered: microarray based and medical imaging based. In contrast to previous researches, this research also considered the imbalanced nature of these data. It was observed that the SMO algorithm performed better for the majority of the test data, especially for microarray based data when accuracy was used as performance measure. Considering the imbalanced characteristic of the data, the Naive Bayes algorithm was seen to perform highly in terms of true positive rate (TPR). Regarding the influence of SMOTE, a well-known imbalanced data classification technique, it was observed that there was a notable performance improvement for J48, while the performance of SMO remained comparable for the majority of the datasets. Overall, the results indicated SMO as the most potential candidate for the microarray and image dataset considered in this research.

Significant Cancer Prevention Factor Extraction: An Association Rule Discovery Approach

Cancer is increasing the total number of unexpected deaths around the world. Until now, cancer research could not significantly contribute to a proper solution for the cancer patient, and as a result, the high death rate is uncontrolled. The present research aim is to extract the significant prevention factors for particular types of cancer. To find out the prevention factors, we first constructed a prevention factor data set with an extensive literature review on bladder, breast, cervical, lung, prostate and skin cancer. We subsequently employed three association rule mining algorithms, Apriori, Predictive apriori and Tertius algorithms in order to discover most of the significant prevention factors against these specific types of cancer. Experimental results illustrate that Apriori is the most useful association rule-mining algorithm to be used in the discovery of prevention factors.

RBF neural networks for solving the inverse problem of backscattering spectra

This paper investigates a new method to solve the inverse problem of Rutherford backscattering (RBS) data. The inverse problem is to determine the sample structure information from measured spectra, which can be defined as a function approximation problem. We propose using radial basis function (RBF) neural networks to approximate an inverse function. Each RBS spectrum, which may contain up to 128 data points, is compressed by the principal component analysis, so that the dimensionality of input data and complexity of the network are reduced significantly. Our theoretical consideration is tested by numerical experiments with the example of the SiGe thin film sample and corresponding backscattering spectra. A comparison of the RBF method with multilayer perceptrons reveals that the former has better performance in extracting structural information from spectra. Furthermore, the proposed method can handle redundancies properly, which are caused by the constraint of output variables. This study is the first method based on RBF to deal with the inverse RBS data analysis problem.

Intelligent methods for solving inverse problems of backscattering spectra with noise: a comparison between neural networks and simulated annealing

This paper investigates two different intelligent techniques—the neural network (NN) method and the simulated annealing (SA) algorithm for solving the inverse problem of Rutherford backscattering (RBS) with noisy data. The RBS inverse problem is to determine the sample structure information from measured spectra, which can be defined as either a function approximation or a non-linear optimization problem. Early studies emphasized on numerical methods and empirical fitting. In this work, we have applied intelligent techniques and compared their performance and effectiveness for spectral data analysis by solving the inverse problem. Since each RBS spectrum may contain up to 512 data points, principal component analysis is used to make the feature extraction so as to ease the complexity of constructing the network. The innovative aspects of our work include introducing dimensionality reduction and noise modeling. Experiments on RBS spectra from SiGe thin films on a silicon substrate show that the SA is more accurate but the NN is faster, though both methods produce satisfactory results. Both methods are resilient to 10% Poisson noise in the input. These new findings indicate that in RBS data analysis the NN approach should be preferred when fast processing is required; whereas the SA method becomes the first choice should the analysis accuracy be targeted.

Assessing the capacity reliability of ageing water distribution systems

This paper presents two new efficient algorithms for estimating the capacity reliability of ageing water distribution systems recognising the uncertainties in nodal demands and the pipe capacity. Capacity reliability is defined as the probability that the nodal demand is met at or over the prescribed minimum pressure for a fixed network configuration. Uncertainties in the nodal demands and values of pipe roughness are modelled by a probabilistic approach. The impacts of these uncertainties on the hydraulic performance of water distribution systems are then assessed by probabilistic hydraulic models based on the mean value first order second moment (MVFOSM) method and the first order reliability method (FORM) respectively. The performances of the models are evaluated and compared by application to an example network. Results from this application indicate that both models provide reasonably accurate estimates of capacity reliability of a deteriorated distribution network in the cases that the uncertainty in the random variables is small. However, FORM performs much better in cases involving large variability in the nodal demands and pipe roughnesses.

A Comparison Between Rule Based and Association Rule Mining Algorithms

Recently association rule mining algorithms are using to solve data mining problem in a popular manner. Rule based mining can be performed through either supervised learning or unsupervised learning techniques. Among the wide range of available approaches, it is always challenging to select the optimum algorithm for rule based mining task. The aim of this research is to compare the performance between the rule based classification and association rule mining algorithm based on their rule based classification performance and computational complexity. We consider PART (Partial Decision Tree) of classification algorithm and Apriori of association rule mining to compare their performance. DARPA (Defense Advanced Research Projects Agency) data is a well-known intrusion detection problem is also used to measure the performance of these two algorithms. In this comparison the training rules are compared with the predefined test sets. In terms of accuracy and computational complexity we observe Apriori is a better choice for rule based mining task.

Finding a unique Association Rule Mining algorithm based on data characteristics

This research compares the performance of three popular association rule mining algorithms, namely apriori, predictive apriori and tertius based on data characteristics. The accuracy measure is used as the performance measure for ranking the algorithms. A wide variety of association rule mining algorithms can create a time consuming problem for choosing the most suitable one for performing the rule mining task. A meta-learning technique is implemented for a unique selection from a set of association rule mining algorithms. On the basis of experimental results of 15 UCI data sets, this research discovers statistical information based rules to choose a more effective algorithm.