Search

The diagnosis of cancer is presently undergoing a change of paradigm for the diagnostic panel using molecular biomarkers. MicroRNA (miRNA) is one of the most important genomic datasets presenting the genome sequences. Since several studies have shown the relationship between miRNAs and cancers, data mining and machine learning methods can be incorporated to extract a large amount of knowledge from cancer genomic datasets. However, previous research works on the identification of cancers from miRNAs have made it possible to diagnose cancer, and the accuracy of some classes is not quite satisfactory. Therefore, this research is aimed at promoting a super-class (meta-label) approach and deep learning in a three-phase method to diagnose cancers from miRNAs. The steps in the first phase of the proposed method, named Representation learning, are partitioning data into super-classes, meta-data creation and super-classes classification. This phase helps data to be split into some subsets to improve classification accuracy. In other words, the first phase groups labels based on the separability of classes into a meta-label, and then a multi-label learner is built to predict these meta-labels. In the second phase, a feature selection to reduce the dimensions of the problem is applied to each super-class to help to focus the attention of an induction algorithm in those features that are more important to predict the target concept. In the third phase of the proposed method, an evolutionary deep neural network for the classification of labels in each super-class is performed. The last two phases are done separately for each subset in which five super-classes and subsequently five deep neural networks are trained. The experimental results reveal that the proposed method achieved more efficient results than 19 recent machine learning methods. Despite the fact that evaluating the dataset which consists of 29 types of cancers provides a more complicated situation for the convolutional neural network to be learned, the performance of the method is noticeably better than other existing methods. The other success which can be considered here is a significant reduction in running time comparing to other methods.

Since recommended treatment for Non-small cell lung cancer (NSCLC) after surgery is chemotherapy, the prediction of effectiveness or futileness of adjuvant chemotherapy (ACT) in early stage is important for future decision. Classification of NSCLC in gene expression data is performed to predict effectiveness or futileness of ACT. Selection of genes highly correlated with the class attribute, affects the classification accuracy. In this paper, a new cell separation algorithm is proposed which it imitates the action of cell separation using differential centrifugation process involving multiple centrifugation steps and increasing the rotor speed in each step. The CSA uses the application of centrifugal force to separate the solutions based on their objective function in different steps while the velocity is increased in each step. The CSA contributes to automatic trade-off between exploration and exploitation by control of selection rate during the search process. To examine the CSA, 25 test functions were used first and then the CSA was applied to predict effectiveness or futileness of ACT. The number of genes in candidate subsets is handled by increasing the subset size if after a certain number of iterations there is no improvement in fitness of the subset. This contributes to less time consideration and memory usage. In this experiment, the NSCLC data contain 280 samples collected from four institutes are used. As results, the minimum number of five genes with dependency degree equal to one and classification accuracy of higher than 94% for SVM, KNN and MLP classifiers is obtained.

PurposeMetaheuristic algorithms are classified into two categories namely: single-solution and population-based algorithms. Single-solution algorithms perform local search process by employing a single candidate solution trying to improve this solution in its neighborhood. In contrast, population-based algorithms guide the search process by maintaining multiple solutions located in different points of search space. However, the main drawback of single-solution algorithms is that the global optimum may not reach and it may get stuck in local optimum. On the other hand, population-based algorithms with several starting points that maintain the diversity of the solutions globally in the search space and results are of better exploration during the search process. In this paper more chance of finding global optimum is provided for single-solution-based algorithms by searching different regions of the search space.Design/methodology/approachIn this method, different starting points in initial step, searching locally in neighborhood of each solution, construct a global search in search space for the single-solution algorithm.FindingsThe proposed method was tested based on three single-solution algorithms involving hill-climbing (HC), simulated annealing (SA) and tabu search (TS) algorithms when they were applied on 25 benchmark test functions. The results of the basic version of these algorithms were then compared with the same algorithms integrated with the global search proposed in this paper. The statistical analysis of the results proves outperforming of the proposed method. Finally, 18 benchmark feature selection problems were used to test the algorithms and were compared with recent methods proposed in the literature.Originality/valueIn this paper more chance of finding global optimum is provided for single-solution-based algorithms by searching different regions of the search space.

Feature (biomarker) selection of microRNA data is an important step for the cancer classification task. In this paper, the cell separation meta-heuristic algorithm (CSA) which imitates the cell separation action using differential centrifugation process is improved. Enhanced search behaviour embedded in movement of virtual cells in this algorithm is provided. This enhancement contributes to an effective trade-off between global exploration and local exploitation of the search space which is the key factor to improve the performance of any meta-heuristic. In order to perform feature selection, different levels and periods to handle the number of features in the selected subset is automatically used during the search process. This process is set to avoid the effect of the dimension of data on computational time and effort. To examine the improved CSA (I-CSA), 25 test functions were initially used. Next, six benchmark classification problems, including four biological and two other datasets, were employed to test the I-CSA. Furthermore, an experiment for feature selection from microRNA data is conducted. The classification accuracy of the selected features obtained by applying I-CSA on microRNA data is compared with the accuracy of 22 well-known classifiers on a real-world dataset. This dataset contains genomic information of 8,129 patients for 29 different types of cancer with 1,046 gene expressions. The classification accuracy of each cancer type is also ranked with the results of 77 classifiers reported in previous works. As a result, the proposed approach achieved 100% accuracy in 25 out of 29 classes. In seven cases out of 29, the method achieved 100% accuracy, which no classifier in other studies has reached.

A broad variety of real-world problems have been solved using multilayer perceptron (MLP) artificial neural networks (ANNs). Optimization techniques aid ANNs to select suitable weights and achieve correct results. Recently, the kidney-inspired algorithm (KA) has been proposed for optimization problems. This algorithm is based on the filtration, reabsorption, secretion, and excretion processes that take place in the kidneys of the human body. In the KA, the value of in the filtration rate formula is a constant value in the range of 0, 1 that is set in the initialization stage of the algorithm. In this paper, an improved KA for optimization of the ANN model is presented in which the filtration rate is controlled by changing the value of from minimum to maximum during the search process, which helps in achieving a better balance between exploration and exploitation in the algorithm. In this algorithm if more solutes are filtered and moved to filtered blood it means that the algorithm has more exploration. In contrast, if more solutes move to waste it means that more exploitation is performed by the algorithm. In addition, the separate use of three chaotic maps instead of a random number in the movement formula of the modified KA is investigated in order to assess the ability of each map to help to achieve superior results. The proposed method is tested on benchmark classification and time series prediction problems. The method is also applied to a real-world rainfall forecasting problem. The results of a statistical analysis prove the ability of the method.

Display OmittedCooperative-competitive master-slave global-best harmony search algorithm. A multi-population is switched between cooperative master-slave and competitive master-slave strategies.The proposed method is compared against single population, cooperative master-slave and competitive master-slave.Finally, proposed method is applied on real-world water quality prediction problem. The artificial neural network (ANN) is one of the most accurate and commonly used machine-learning techniques and can learn even complex data by employing metaheuristic algorithms. Harmony search (HS) is a metaheuristic algorithm that imitates the process by which musicians tune their instruments to achieve perfect harmony. Global-best harmony search(GHS) is an effective variant of the HS algorithm that borrows the concept of gbest (globalbest) from particle-swarm optimization (PSO) to improve the performance of HS. Employing a multi-population technique improves the convergence of the algorithm. The master-slave technique is one of the most powerful multi-population techniques. This paper proposes a cooperative-competitive master-slave multi-population GHS (CC-GHS) to train the ANN. To provide the proposed CC-GHS algorithm with strong abilities in both exploration and exploitation, a competitive master-slave strategy (Com-GHS)is interacted with a cooperative master-slave strategy(Coo-GHS). A probabilistic variable is employed to achieve a good balance between cooperativeness and competitiveness. The method is tested on benchmark classification and time-series prediction problems, and statistical analyses demonstrate the ability of the proposed method. The CC-GHS is also applied to a real-world water-quality prediction problem with promising results.

In this paper, a population-based algorithm inspired by the kidney process in the human body is proposed. In this algorithm the solutions are filtered in a rate that is calculated based on the mean of objective functions of all solutions in the current population of each iteration. The filtered solutions as the better solutions are moved to filtered blood and the rest are transferred to waste representing the worse solutions. This is a simulation of the glomerular filtration process in the kidney. The waste solutions are reconsidered in the iterations if after applying a defined movement operator they satisfy the filtration rate, otherwise it is expelled from the waste solutions, simulating the reabsorption and excretion functions of the kidney. In addition, a solution assigned as better solution is secreted if it is not better than the worst solutions simulating the secreting process of blood in the kidney. After placement of all the solutions in the population, the best of them is ranked, the waste and filtered blood are merged to become a new population and the filtration rate is updated. Filtration provides the required exploitation while generating a new solution and reabsorption gives the necessary exploration for the algorithm. The algorithm is assessed by applying it on eight well-known benchmark test functions and compares the results with other algorithms in the literature. The performance of the proposed algorithm is better on seven out of eight test functions when it is compared with the most recent researches in literature. The proposed kidney-inspired algorithm is able to find the global optimum with less function evaluations on six out of eight test functions. A statistical analysis further confirms the ability of this algorithm to produce good-quality results.

This paper presents a new solution representation for genetic algorithm to optimize the neural network model. During the optimization process, the weights, biases and structure of the neural network are considered for altering. The quality of the model is examined by a cost function that deliberates over both minimization of error and complexity of the neural network model. The performance of the proposed method is investigated by applying it on two time series prediction problems. The results show promising results when we compare it with other methods in the literature. New solution representation for ANN optimization.Better performance for the time series prediction problem.Less complex neural network structure.

A modified bat algorithm with a new solution representation for both optimizing the weights and structure of ANNs is proposed.To improve the exploration and exploitation capability of bat algorithm some modifications based on chaotic map on bat algorithm are studied.The Taguchi method is used to tune the parameters of the algorithm.Six classifications and two time series benchmark datasets are used to test the performance of the proposed approach in terms of classification and prediction accuracy.Finally, our best method is applied to a real-world problem, namely to predict the future values of rainfall data in Selangor at Malaysia. The success of an artificial neural network (ANN) strongly depends on the variety of the connection weights and the network structure. Among many methods used in the literature to accurately select the network weights or structure in isolate; a few researchers have attempted to select both the weights and structure of ANN automatically by using metaheuristic algorithms. This paper proposes modified bat algorithm with a new solution representation for both optimizing the weights and structure of ANNs. The algorithm, which is based on the echolocation behaviour of bats, combines the advantages of population-based and local search algorithms. In this work, ability of the basic bat algorithm and some modified versions which are based on the consideration of the personal best solution in the velocity adjustment, the mean of personal best and global best solutions through velocity adjustment and the employment of three chaotic maps are investigated. These modifications are aimed to improve the exploration and exploitation capability of bat algorithm. Different versions of the proposed bat algorithm are incorporated to handle the selection of the structure as well as weights and biases of the ANN during the training process. We then use the Taguchi method to tune the parameters of the algorithm that demonstrates the best ability compared to the other versions. Six classifications and two time series benchmark datasets are used to test the performance of the proposed approach in terms of classification and prediction accuracy. Statistical tests demonstrate that the proposed method generates some of the best results in comparison with the latest methods in the literature. Finally, our best method is applied to a real-world problem, namely to predict the future values of rainfall data and the results show satisfactory of the method.

The age of each individual can be predicted based on the alteration rule of DNA methylation with age. In this paper, an age prediction method is developed in order to solve multivariate regression problems from DNA methylation data, by optimizing the artificial neural network (ANN) model using a new proposed algorithm named the Cell Separation Algorithm (CSA). The CSA imitates cell separation action by using a differential centrifugation process involving multiple centrifugation steps and increasing the rotor speed in each step. The CSA performs similar to the centrifugal force in separating the solutions based on their objective function in different steps, with velocity increasing in each step. Firstly, 25 test functions are used to test the CSA. Secondly, the CSA is examined on three forms of age prediction problems from two body fluids (blood and saliva). The healthy blood samples, diseased blood samples and saliva samples are used to test the method's capability. The results of the CSA are compared not only with other methods proposed in previous studies, but also with the results from stochastic gradient descent (SGD), ADAM, and genetic algorithm (GA). The model results of CSA are extremely better than the four methods proposed in previous works that have not used ANN training process. The CSA also outperformed SGD, ADAM that employ the ANN model without ANN optimization by meta-heuristics. The CSA results are comparable (even superior) to the GA model which takes the advantages of both ANN and meta-heuristics.

This paper presents an investigation of the nonlinearity behavior in deformation that appears during the linear stretching of the elastomeric microstructures in a pillar-based microfilter. Determining the impact of the nonuniformity in strain on the geometry and performance of such a planar device under in-plane stretch is the motivation for this study. A semiempirical model is used to explain the physical strain–stress behavior from the root to the tip of the micropillars in the linear arrays in the device. For microfabrication of the device, the main substrate is elastomeric polyurethane methacrylate, which is utilized in an ultraviolet-molding method. Optical imaging and scanning electron microscopy were used to evaluate the deformation of the microstructures under different loading conditions. It was demonstrated that by applying mechanical strains of $\Delta L/L_{o})$ on the elastomeric device using a modified syringe pump, the spacing of the pillars is increased effectively to about three times the size of the initial setting of 5.5 $\mu $ m, which corresponds to a strain of above 180% in the absence of nonuniformity effects. This simple yet interesting behavior can be exploited to rapidly adjust a microfluidic device for application to the separation of microbeads or blood cells, which would normally require the geometrical redesign and fabrication of a new device. [2014-0213]

The performance of an artificial neural network (ANN) depends on the connection weights and network structure. Many optimization algorithms have been applied for ANN model selection. This paper presents an optimization algorithm based on the cooperative bat-inspired algorithm. The advantage of the bat algorithm lies in the combination of a population-based algorithm and local search; however, it is more powerful in local search. Therefore to better balance exploration and exploitation in the population some modifications to the velocity equation of the standard bat algorithm are applied. In addition, we propose two new topologies for cooperation between subpopulations to further improve the algorithm's capability to maintain the diversity of bats in the population. The first is a combination of two known mechanisms (Ring and Master-Slave), and the second inserts a Coevolving strategy of slave subpopulations in the Master-Slave strategy. The proposed methods are applied for the selection of an ANN model, where both the structure of the ANN and its weights are optimized by the method. Six classification and two time series prediction benchmark datasets are tested and the performance of the proposed algorithms is evaluated and compared with other methods in the literature. Statistical analysis shows that for the classification problem there is a significant improvement in the bat algorithm with Ring and Master-Slave strategies cooperation compared to the other methods in the literature in terms of classification error for three cases out of five and a significant enhancement in the number of connection weights in the network. The analysis also shows that for time series prediction there is a significant improvement in the prediction error for all the cases.

The attribute reduction problem is the process of reducing unimportant attributes from a decision system to decrease the difficulty of data mining or knowledge discovery tasks. Many algorithms have been used to optimize this problem in rough set theory. The genetic algorithm (GA) is one of the algorithms that has already been applied to optimize this problem. This paper proposes 2 kinds of memetic algorithms, which are a hybridization of the GA, with 2 versions (linear and nonlinear) of the great deluge (GD) algorithm. The purpose of this hybridization is to investigate the ability of this local search algorithm to improve the performance of the GA. In both of the methods, the local search (the GD algorithm) is employed to each generation of the GA. The only difference of these methods is the rate of increase in the `level' in the GD algorithm. The level is increased by a fixed value in the linear GD algorithm, while the nonlinear GD algorithm uses the quality of the current solution to calculate the increase rate of the level in each iteration. The 13 datasets taken from the University of California - Irvine machine learning repository are used to test the methods and compare the results with the on-hand results in the literature, especially with the original GA. The classification accuracies of each dataset using the obtained reducts are examined and compared with other approaches using ROSETTA software. The promising results show the potential of the algorithm to solve the attribute reduction problem.

Hydropower is one of the technologies in renewable energy that is commercially viable on a large scale. A hybrid of metaheuristic Artificial Neural Network (ANN) technique with Bat Algorithm (BA), a bio-inspired algorithm is proposed to forecast future electricity production and water consumption at Sultan Azlan Shah Hydropower Dam located upstream of Perak river. In this study, both the ANN and Hybrid ANN-Bat Algorithm coding was designed and written explicitly to tailor the time series input data and assumptions used in this study. Comparison on results obtained from ANN and the proposed hybrid ANN — BA was conducted. Simulations conducted in this study exhibited that the proposed hybrid algorithm is much superior then the conventional ANN.

A number of properties of Artificial Neural Networks (ANNs) make them suitable for many applications such as time series prediction problem. However, lack of training model which finds a global optimal set of weights has been disadvantaged in some real-world problems. Genetic algorithm is an optimization procedure which is superior at exploring a search space in an intelligent method. In this paper we present a genetic-based algorithm to optimize the weights and biases of the ANN. In this work we tune the parameters of the genetic algorithm using Taguchi method. To test the method two standard time series prediction problems are employed. The results are compared to the methods in the literature. The comparison showed the superiority of the proposed method.

Attribute reduction is the process of selecting a subset of features from the original set of features that forms patterns in a given dataset. It can be defined as a process to eliminate redundant attributes and at the same time is able to avoid any information loss, so that the selected subset is sufficient to describe the original features. In this paper, we present a great deluge algorithm for attribute reduction in rough set theory (GD-RSAR). Great deluge is a meta-heuristic approach that is less parameter dependent. There are only two parameters needed; the time to “spend” and the expected final solution. The algorithm always accepts improved solutions. The worse solution will be accepted if it is better than the upper boundary value or “level”. GD-RSAR has been tested on the public domain datasets available in UCI. Experimental results on benchmark datasets demonstrate that this approach is effective and able to obtain competitive results compared to previous available methods. Possible extensions upon this simple approach are also discussed.