Moth flame optimization (MFO) is a swarm-based algorithm with mediocre performance and marginal originality proposed in recent years. It tried to simulate the fantasy navigation mode of moth lateral positioning. The basic MFO has no specific, deep strategies in different periods of the algorithm and a fragile evolutionary basis, which may lead to the problem of falling into local optimum and slow convergence trend. Therefore, this paper introduces a double adaptive weight mechanism into the MFO algorithm, termed as WEMFO, to boost the search capability of the basic MFO and provide a more efficient tool for optimization purposes. The proposed WEMFO adjusts the search strategy adaptively in different periods of the algorithm, making it more flexible between global search (diversification) and local search (intensification). The WEMFO algorithm is compared with some illustrious metaheuristic solvers and advanced metaheuristic methods developed in recent years on thirty benchmark functions. The experimental results expose that the developed WEMFO has apparent compensations in terms of convergence speed and solution accuracy. Moreover, this paper analyzes the diversity and balance of WEMFO and applies the algorithm to several engineering problems. The experimental results show that the WEMFO algorithm has good performance in engineering problems. Additionally, the proposed WEMFO was also applied to train Kernel Extreme Learning Machine (KELM), the resultant optimized WEMFO-KELM model was applied to six clinical disease classification problems. By comparing with MFO-KELM and other five classification models, the experimental results showed that the proposed algorithm had shown better performance in practical problems. An online guide for the algorithm in this research WEMFO and proposed classifier WEMFO-KELM will be publicly available at https://aliasgharheidari.com .

Double adaptive weights for stabilization of moth flame optimizer: Balance analysis, engineering cases, and medical diagnosis

Breast cancer (BC) is one of the primary causes of cancer death among women. Early detection of BC allows patients to receive appropriate treatment, thus increasing the possibility of survival. In this work, a new deep-learning (DL) model based on the transfer-learning (TL) technique is developed to efficiently assist in the automatic detection and diagnosis of the BC suspected area based on two techniques namely 80–20 and cross-validation. DL architectures are modeled to be problem-specific. TL uses the knowledge gained during solving one problem in another relevant problem. In the proposed model, the features are extracted from the mammographic image analysis- society (MIAS) dataset using a pre-trained convolutional neural network (CNN) architecture such as Inception V3, ResNet50, Visual Geometry Group networks (VGG)-19, VGG-16, and Inception-V2 ResNet. Six evaluation metrics for evaluating the performance of the proposed model in terms of accuracy, sensitivity, specificity, precision, F-score, and area under the ROC curve (AUC) has been chosen. Experimental results show that the TL of the VGG16 model is powerful for BC diagnosis by classifying the mammogram breast images with overall accuracy, sensitivity, specificity, precision, F-score, and AUC of 98.96%, 97.83%, 99.13%, 97.35%, 97.66%, and 0.995, respectively for 80–20 method and 98.87%, 97.27%, 98.2%, 98.84%, 98.04%, and 0.993 for 10-fold cross-validation method.

/pdf/a-novel-deep-learning-model-for-automatic-detection-and-6978h97i3w.pdf

A Novel Deep-Learning Model for Automatic Detection and Classification of Breast Cancer Using the Transfer-Learning Technique

Harris hawks optimization (HHO) is a newly developed swarm-based algorithm and the most popular optimizer in the recent year, which mimics the cooperation behavior of Harris hawks. Although the original HHO has a specific weight compared to other state-of-the-art methods when working on the local exploitation of feasible solutions, it still may fail to achieve an excellent set of scales between locally accurate exploitation and globally exploratory search. This imbalanced behavior forms an overall perspective, which can be experienced by slow convergence, inaccuracy or inadequate search coverage, and quickly dropping into local solutions. In order to strengthen the performance of the HHO, two strategies of Gaussian mutation and a dimension decision strategy observed in the cuckoo search method are introduced into this optimizer. The mechanism of cuckoo search is very useful in improving the convergence speed of the search agents as well as sufficient excavation of the solutions in the search area, while the Gaussian mutation strategy performs well in increasing the accuracy and jumping out of the local optimum. In order to verify the remarkable performance of the proposed method, the enhanced paradigm is compared with other mate-heuristic algorithms on 30 IEEE CEC2017 benchmark functions and three typical engineering problems. The experimental results illustrate that the novel developed GCHHO has an excellent ability to achieve superior performance in competition with the original HHO as well as other well-established optimizers. Open source codes and information on original HHO algorithm is available at: https://aliasgharheidari.com/HHO.html . Supportive data and info for this research will be publicly provided in http://aliasgharheidari.com .

Dimension decided Harris hawks optimization with Gaussian mutation: Balance analysis and diversity patterns

One of the essential tasks for the planning and development of talents training programs in different colleges of universities is to find how we can reasonably guide students to pursue a master's degree concerning their comprehensive situations. The purpose of this study is to develop a modified fuzzy k-Nearest Neighbor (FKNN) framework to predict the college students' intentions for master programs in advance, that is, students choose to attend the postgraduate exam or find a job after graduation. The proposed integrated framework combines the random forest (RF), FKNN, and a new chaos-enhanced sine cosine-inspired algorithm (CESCA). In this model, RF is employed to evaluate the importance of features in the dataset, while the FKNN is utilized to establish the relationship framework between the features and the college students' decisions to earn a master's degree or not. The proposed CESCA is utilized to tune the key parameters of the FKNN automatically. All eight variants of SCA have been rigorously compared based on 13 benchmark problems to validate the effectiveness of the proposed CESCA. Then, the CESCA-based FKNN (CESCA-FKNN) has been further compared against the other three classical classifiers in terms of four common performance metrics. The experimental results indicate that the proposed CESCA-FKNN can obtain the best classification accuracy. The results indicate that the established adaptive FKNN framework can be served as a powerful tool for college students' intention before pursuing a master's degree.

/pdf/predicting-intentions-of-students-for-master-programs-using-12im4o4e1m.pdf

Predicting Intentions of Students for Master Programs Using a Chaos-Induced Sine Cosine-Based Fuzzy K-Nearest Neighbor Classifier

It is of great clinical significance to develop an accurate computer aided system to accurately diagnose the breast cancer. In this study, an enhanced machine learning framework is established to diagnose the breast cancer. The core of this framework is to adopt fruit fly optimization algorithm (FOA) enhanced by Levy flight (LF) strategy (LFOA) to optimize two key parameters of support vector machine (SVM) and build LFOA-based SVM (LFOA-SVM) for diagnosing the breast cancer. The high-level features abstracted from the volunteers are utilized to diagnose the breast cancer for the first time. In order to verify the effectiveness of the proposed method, 10-fold cross-validation method is used to make comparison among the proposed method, FOA-SVM (model based on original FOA), PSO-SVM (model based on original particle swarm optimization), GA-SVM (model based on genetic algorithm), random forest, back propagation neural network and SVM. The main novelty of LFOA-SVM lies in the combination of FOA with LF strategy that enhances the quality for FOA, thus improving the convergence rate of the FOA optimization process as well as the probability of escaping from local optimal solution. The experimental results demonstrate that the proposed LFOA-SVM method can beat other counterparts in terms of various performance metrics. It can very well distinguish malignant breast cancer from benign ones and assist the doctor with clinical diagnosis.

/pdf/a-new-fruit-fly-optimization-algorithm-enhanced-support-492trfoab0.pdf

A new fruit fly optimization algorithm enhanced support vector machine for diagnosis of breast cancer based on high-level features

Glyphosate, one of the most popular herbicides world-wide, is used as an active ingredient in many commercial formulations. So far, many studies have focused on reproductive toxicity on glyphosate which may trigger epigenetic changes through endocrine-mediated mechanisms. Generally, it is critical to identify the glyphosate poisoning status in order to minimize health risks. This paper proposed a machine learning approach using 110 rats’ samples to identify the glyphosate poisoning status. All rats were randomly divided into 2 groups ( $n=55$ each) in the study. The rats in the blank control group were given 0.9% sodium chloride solution orally for 15 days, while the other groups were administered orally at the dosage of 0.5g/kg of glyphosate once per day for 15 days consecutively. Consequently, the potential of an enhanced fuzzy k nearest neighbor approach was explored through blood routine test to recognize the glyphosate poisoning condition based on the blood indices. A real-life data set was used to evaluate the effectiveness of the proposed method in terms of classification accuracy (ACC), sensitivity, specificity, and Matthews Correlation Coefficients (MCC). The results demonstrated that there were significant differences in blood routine test indices between the control and the glyphosate groups (p-value < 0.01). It was observed in the feature selection that the most important correlated indices were white blood cell, lymphocyte, intermediate cell, and neutrophil granulocyte. A comparative study with support vector machine and artificial neural networks had shown that the proposed approach could achieve much more promising results with 95.45% ACC, 0.90 MCC, 98.89% sensitivity, and 88.89% specificity. The empirical analysis verifies that the proposed method is promising to act as a new, accurate method for identification of the glyphosate poisoning status in subjects.

An Effective Machine Learning Approach for Identifying the Glyphosate Poisoning Status in Rats Using Blood Routine Test

Phenanthrene, a PAH with three fused benzene rings, is usually used as a model for the study on PAHs. During 4 days, 166 male mice were equally and randomly divided into two groups. One group was given vehicle-corn oil by oral gavage, the other was given phenanthrene at a dose of 450 milligrams per kilogram per day. In this study, in order to predict mice’s phenanthrene poisoning by virtue of blood analysis indices, a new machine learning approach was put forward, which was based on an improved binary moth-flame optimizer combined with extreme learning machine. The results of the experiment have manifested that the blood analysis indices of the control and phenanthrene groups were significantly different ( $p ). The most important correlated indices including serum alanine aminotransferase (ALT), gamma-glutamyl transferase (GGT), plateletcrit (PCT) and red blood cell distribution width-standard deviation (RDW-SD) were screened through feature selection. The classification results demonstrated that the proposed method can achieve 93.38% accuracy and 98.33% specificity. Promisingly, there is a new and accurate way to detect the status of phenanthrene poisoning expectably.

/pdf/a-new-hybrid-machine-learning-approach-for-prediction-of-3ezxq4ejnl.pdf

A New Hybrid Machine Learning Approach for Prediction of Phenanthrene Toxicity on Mice

Phthalic acid esters (PAEs) are organic pollutants and synthetic compounds and have adverse effects on human health. In this study, we investigated whether Di-2-Ethylhexyl phthalate (DEHP), one of many PAEs, has adverse effects on rats. Adult male Sprague-Dawley rats were treated daily by oral gavage with vehicle (corn oil) or DEHP at a dose of 3000 mg/kg/day for 15 days. The results showed that DEHP caused hepatotoxicity in rats. When compared with the control group, relative liver weights, and serum alanine, aminotransferase levels significantly increased after DEHP exposure. Hepatocyte swelling and degeneration were also found in DEHP-exposed rats. This study proposes an effective intelligence framework for the prediction of DEHP poisoning. The framework is designed by integrating an enhanced Harris hawks optimization (HHO) with a support vector machine (SVM), which is called SGLHHO-SVM. The core characteristic of the developed methodology is the SGLHHO algorithm that integrates the levy mechanism and two core operators abstracted from the salp swarm algorithm and grey wolf optimizer to enhance and restore the search capabilities of the HHO. The presented SGLHHO approach is used to tackle the key parameter pair optimization of the SVM, and it is also utilized to grab the optimal feature subset. Regarding the optimal feature subset and the pair parameter simultaneously, SGLHHO-SVM can autonomously predict the DEHP poisoning. The developed SGLHHO was conducted on 23 benchmark problems and compared with other state-of-the-art and competitive methods. The results demonstrate that the designed SGLHHO performs superior to other competitors on most benchmark problems. Furthermore, the proposed SGLHHO-SVM is also compared with other machine learning algorithms on a real-life DEHP sampled data. Statistical results verify the proposal can show better predictive property and higher stability on all for metrics. Therefore, the SGLHHO-SVM may be served as a potential computer-aided tool for the prediction of DEHP poisoning.

/pdf/predicting-di-2-ethylhexyl-phthalate-toxicity-hybrid-3ws91hdlvc.pdf

Predicting Di-2-Ethylhexyl Phthalate Toxicity: Hybrid Integrated Harris Hawks Optimization With Support Vector Machines

Pyrene, composed of four fused benzene rings, is a polycyclic aromatic hydrocarbon (PAH) that has served as a model compound for evaluating the toxic effects of PAHs. In this paper, 114 male rats were dosed daily by oral gavage with either vehicle (corn oil) or pyrene (1500 mg/kg/day) for four days. A method based on the gray wolf optimization-enhanced machine learning approach was then developed to identify pyrene poisoning in rats using the indices from blood analysis. The results showed that there were significant differences in blood analysis indices between the control and the pyrene groups ( $p ). In terms of feature selection, the most important correlated indices were liver to body weight ratio, the absolute value of leukomonocyte, the percentage of monocyte, serum albumin, direct bilirubin, urea nitrogen, and uric acid. This method was shown to have an accuracy rate of 94.62% (ACC), 0.8988 Matthews correlation coefficients (MCCs), 91.71% sensitivity, and 98.33% specificity. Empirical analysis indicates that this method represents a new and accurate approach for detecting pyrene poisoning.

/pdf/a-new-evolutionary-machine-learning-approach-for-identifying-3u1ixvutnc.pdf

Jiayin Zhu

Papers

An Effective Machine Learning Approach for Identifying the Glyphosate Poisoning Status in Rats Using Blood Routine Test

A New Hybrid Machine Learning Approach for Prediction of Phenanthrene Toxicity on Mice

Predicting Di-2-Ethylhexyl Phthalate Toxicity: Hybrid Integrated Harris Hawks Optimization With Support Vector Machines

A New Evolutionary Machine Learning Approach for Identifying Pyrene Induced Hepatotoxicity and Renal Dysfunction in Rats