Sequential Feature Selection Algorithm Research Articles

A hybrid machine learning-based model for predicting flight delay through aviation big data

The prediction of flight delays is one of the important and challenging issues in the field of scheduling and planning flights by airports and airlines. Therefore, in recent years, we have witnessed various methods to solve this problem using machine learning techniques. In this article, a new method is proposed to address these issues. In the proposed method, a group of potential indicators related to flight delay is introduced, and a combination of ANOVA and the Forward Sequential Feature Selection (FSFS) algorithm is used to determine the most influential indicators on flight delays. To overcome the challenges related to large flight data volumes, a clustering strategy based on the DBSCAN algorithm is employed. In this approach, samples are clustered into similar groups, and a separate learning model is used to predict flight delays for each group. This strategy allows the problem to be decomposed into smaller sub-problems, leading to improved prediction system performance in terms of accuracy (by 2.49%) and processing speed (by 39.17%). The learning model used in each cluster is a novel structure based on a random forest, where each tree component is optimized and weighted using the Coyote Optimization Algorithm (COA). Optimizing the structure of each tree component and assigning weighted values to them results in a minimum 5.3% increase in accuracy compared to the conventional random forest model. The performance of the proposed method in predicting flight delays is tested and compared with previous research. The findings demonstrate that the proposed approach achieves an average accuracy of 97.2% which indicates a 4.7% improvement compared to previous efforts.

Nonlocal Controls on Tropical Cyclogenesis: A Trajectory-Based Genesis Potential Index

Abstract Tropical cyclone (TC) genesis is initiated by convective precursors or “seeds” and influenced by environmental conditions along the seed-to-TC trajectories. Genesis potential indices (GPIs) provide a simple way to evaluate TC genesis likelihood from environmental conditions but have two limitations that may introduce bias. First, the globally fixed GPIs fail to represent interbasin differences in the relationship between environments and genesis. Second, existing GPIs are only functions of local environmental conditions, whereas nonlocal factors may have a significant impact. We address the first limitation by constructing basin- and time-scale-specific GPIs (local-GPIs) over the eastern North Pacific (ENP) and North Atlantic (NA) using Poisson regression. A sequential feature selection (SFS) algorithm identifies vertical wind shear and a heating condition as leading factors controlling TC genesis in the ENP and the NA, respectively. However, only a slight improvement in performance is achieved, motivating us to tackle the second limitation with a novel trajectory-based GPI (traj-GPI). We merge adjacent nonlocal environments into each grid point based on observed seed trajectory densities. The seed activity, driven mainly by upward motion, and the transition to TCs, controlled primarily by vertical wind shear or heating conditions, are captured simultaneously in the traj-GPI, yielding a better performance than the original GPIs. This study illustrates the importance of seed activity in modeling TC genesis and identifies key environmental factors that influence the process of TC genesis at different stages. Significance Statement The genesis potential index (GPI) is an effective tool for modeling the likelihood of tropical cyclone (TC) genesis for a given time and location. This study reveals that existing GPIs are primarily biased by a lack of information about nonlocal TC seed activity, since they are based only on local large-scale environmental variables. According to our study, upward motion and vertical wind shear are the most influential environmental factors in seed genesis and the transition from seed to TC, respectively. Based on the observed seed trajectories, we build trajectory-based GPIs that include the information from seed activity. Spatiotemporal performances of TC genesis are significantly improved over the original GPIs.

Machine Learning Models in Prediction of Treatment Response After Chemoembolization with MRI Clinicoradiomics Features.

To evaluate machine learning models, created with radiomics and clinicoradiomics features, ability to predict local response after TACE. 188 treatment-naïve patients (150 responders, 38 non-responders) with HCC who underwent TACE were included in this retrospective study. Laboratory, clinical and procedural information were recorded. Local response was evaluated by European Association for the Study of the Liver criteria at 3-months. Radiomics features were extracted from pretreatment pre-contrast enhanced T1 (T1WI) and late arterial-phase contrast-enhanced T1 (CE-T1) MRI images. After data augmentation, data were split into training and test sets (70/30). Intra-class correlations, Pearson's correlation coefficients were analyzed and followed by a sequential-feature-selection (SFS) algorithm for feature selection. Support-vector-machine (SVM) models were trained with radiomics and clinicoradiomics features of T1WI, CE-T1 and the combination of both datasets, respectively. Performance metrics were calculated with the test sets. Models' performances were compared with Delong's test. 1128 features were extracted. In feature selection, SFS algorithm selected 18, 12, 24 and 8 features in T1WI, CE-T1, combined datasets and clinical features, respectively. The SVM models area-under-curve was 0.86 and 0.88 in T1WI; 0.76, 0.71 in CE-T1 and 0.82, 0.91 in the combined dataset, with and without clinical features, respectively. The only significant change was observed after inclusion of clinical features in the combined dataset (p = 0.001). Higher WBC and neutrophil levels were significantly associated with lower treatment response in univariant analysis (p = 0.02, for both). Machine learning models created with clinical and MRI radiomics features, may have promise in predicting local response after TACE. Level 4, Case-control study.

Impulsivity Classification Using EEG Power and Explainable Machine Learning.

Impulsivity is a multidimensional construct often associated with unfavorable outcomes. Previous studies have implicated several electroencephalography (EEG) indices to impulsiveness, but results are heterogeneous and inconsistent. Using a data-driven approach, we identified EEG power features for the prediction of self-reported impulsiveness. To this end, EEG signals of 56 individuals (18 low impulsive, 20 intermediate impulsive, 18 high impulsive) were recorded during a risk-taking task. Extracted EEG power features from 62 electrodes were fed into various machine learning classifiers to identify the most relevant band. Robustness of the classifier was varied by stratified [Formula: see text]-fold cross validation. Alpha and beta band power showed best performance in the classification of impulsiveness (accuracy = 95.18% and 95.11%, respectively) using a random forest classifier. Subsequently, a sequential bidirectional feature selection algorithm was used to estimate the most relevant electrode sites. Results show that as little as 10 electrodes are sufficient to reliably classify impulsiveness using alpha band power ([Formula: see text]-measure = 94.50%). Finally, the Shapley Additive exPlanations (SHAP) analysis approach was employed to reveal the individual EEG features that contributed most to the model's output. Results indicate that frontal as well as posterior midline alpha power seems to be of most importance for the classification of impulsiveness.

Multivariate Data-Driven Models for Wind Turbine Power Curves including Sub-Component Temperatures

The most commonly employed tool for wind turbine performance analysis is the power curve, which is the relation between wind intensity and power. The diffusion of SCADA systems has boosted the adoption of data-driven approaches to power curves. In particular, a recent research line involves multivariate methods, employing further input variables in addition to the wind speed. In this work, an innovative contribution is investigated, which is the inclusion of thirteen sub-component temperatures as possible covariates. This is discussed through a real-world test case, based on data provided by ENGIE Italia. Two models are analyzed: support vector regression with Gaussian kernel and Gaussian process regression. The input variables are individuated through a sequential feature selection algorithm. The sub-component temperatures are abundantly selected as input variables, proving the validity of the idea proposed in this work. The obtained error metrics are lower with respect to benchmark models employing more typical input variables: the resulting mean absolute error is 1.35% of the rated power. The results of the two types of selected regressions are not remarkably different. This supports that the qualifying points are, rather than the model type, the use and the selection of a potentially vast number of input variables.

Prediction of Response of Hepatocellular Carcinoma to Radioembolization: Machine Learning Using Preprocedural Clinical Factors and MR Imaging Radiomics

PurposeTo create and evaluate the ability of machine learning–based models with clinicoradiomic features to predict radiologic response after transarterial radioembolization (TARE). Materials and Methods82 treatment-naïve patients (65 responders and 17 nonresponders; median age: 65 years; interquartile range: 11) who underwent selective TARE were included. Treatment responses were evaluated using the European Association for the Study of the Liver criteria at 3-month follow-up. Laboratory, clinical, and procedural information were collected. Radiomic features were extracted from pretreatment contrast-enhanced T1-weighted magnetic resonance images obtained within 3 months before TARE. Feature selection consisted of intraclass correlation, followed by Pearson correlation analysis and finally, sequential feature selection algorithm. Support vector machine, logistic regression, random forest, and LightGBM models were created with both clinicoradiomic features and clinical features alone. Performance metrics were calculated with a nested 5-fold cross-validation technique. The performances of the models were compared by Wilcoxon signed-rank and Friedman tests. ResultsIn total, 1,128 features were extracted. The feature selection process resulted in 12 features (8 radiomic and 4 clinical features) being included in the final analysis. The area under the receiver operating characteristic curve values from the support vector machine, logistic regression, random forest, and LightGBM models were 0.94, 0.94, 0.88, and 0.92 with clinicoradiomic features and 0.82, 0.83, 0.82, and 0.83 with clinical features alone, respectively. All models exhibited significantly higher performances when radiomic features were included (P = .028, .028, .043, and .028, respectively). ConclusionsBased on clinical and imaging-based information before treatment, machine learning–based clinicoradiomic models demonstrated potential to predict response to TARE.

Channel identification with Improved Variational Mode Decomposition

The identification of the type of wireless propagation channel (e.g., Line of Sight (LOS) or Non Line of Sight (NLOS)) is an important function in the wireless communication design and deployment especially in rich propagation environments. The wireless channel characteristics can be quite specific not only between Line of Sight (LOS) and Non Line of Sight (NLOS) wireless propagation conditions but also in different NLOS environments.In recent times, machine learning approaches have been increasingly used to differentiate and classify channel characteristics and this paper is part of this trend. In particular, this paper proposes the combination of machine learning with a recently proposed signal processing tool called Variational Mode Decomposition (VMD), which is a decomposition algorithm that decomposes a time series into several modes which have specific sparsity properties. VMD itself is a refinement of the Empirical Mode Decomposition (EMD) and demonstrated a superior performance to EMD for classification problems. One issue for the practical deployment of VMD in channel identification problems is the presence of hyper-parameters, which must be tuned for the applied context. The main contribution of this paper is to propose a novel approach for channel identification based on an improvement of VMD called Improved Variational Mode Decomposition (IVMD), where the optimal values of the hyper-parameters of VMD are automatically identified on the basis of the Shannon entropy of the signal output from the channel. Then, various features are extracted from the modes generated by IVMD and a sequential feature selection algorithm is applied to select the optimal features. This paper applies the proposed approach with IVMD to a data set generated by the authors with a wireless channel emulator, where 6 different propagation scenarios (including no fading conditions) are created for WiFi 802.11g signals, where only the preamble is used for channel identification. Even if channel identification based on the normalized preamble is a challenging classification problem, the proposed IVMD is able to outperform significantly the application of basic VMD, EMD and the time and frequency domain representations (as commonly done in literature) of the WiFi signals.

The utility of texture analysis of kidney MRI for evaluating renal dysfunction with multiclass classification model

We evaluated a multiclass classification model to predict estimated glomerular filtration rate (eGFR) groups in chronic kidney disease (CKD) patients using magnetic resonance imaging (MRI) texture analysis (TA). We identified 166 CKD patients who underwent MRI comprising Dixon-based T1-weighted in-phase (IP)/opposed-phase (OP)/water-only (WO) images, apparent diffusion coefficient (ADC) maps, and T2* maps. The patients were divided into severe, moderate, and control groups based on eGFR borderlines of 30 and 60 mL/min/1.73 m2. After extracting 93 texture features (TFs), dimension reduction was performed using inter-observer reproducibility analysis and sequential feature selection (SFS) algorithm. Models were created using linear discriminant analysis (LDA); support vector machine (SVM) with linear, rbf, and sigmoid kernels; decision tree (DT); and random forest (RF) classifiers, with synthetic minority oversampling technique (SMOTE). Models underwent 100-time repeat nested cross-validation. Overall performances of our classification models were modest, and TA based on T1-weighted IP/OP/WO images provided better performance than those based on ADC and T2* maps. The most favorable result was observed in the T1-weighted WO image using RF classifier and the combination model was derived from all T1-weighted images using SVM classifier with rbf kernel. Among the selected TFs, total energy and energy had weak correlations with eGFR.

Automatic Diagnosis of Mild Cognitive Impairment Based on Spectral, Functional Connectivity, and Nonlinear EEG-Based Features.

Accurate and early diagnosis of mild cognitive impairment (MCI) is necessary to prevent the progress of Alzheimer's and other kinds of dementia. Unfortunately, the symptoms of MCI are complicated and may often be misinterpreted as those associated with the normal ageing process. To address this issue, many studies have proposed application of machine learning techniques for early MCI diagnosis based on electroencephalography (EEG). In this study, a machine learning framework for MCI diagnosis is proposed in this study, which extracts spectral, functional connectivity, and nonlinear features from EEG signals. The sequential backward feature selection (SBFS) algorithm is used to select the best subset of features. Several classification models and different combinations of feature sets are measured to identify the best ones for the proposed framework. A dataset of 16 and 18 EEG data of normal and MCI subjects is used to validate the proposed system. Metrics including accuracy (AC), sensitivity (SE), specificity (SP), F1-score (F1), and false discovery rate (FDR) are evaluated using 10-fold crossvalidation. An average AC of 99.4%, SE of 98.8%, SP of 100%, F1 of 99.4%, and FDR of 0% have been provided by the best performance of the proposed framework using the linear support vector machine (LSVM) classifier and the combination of all feature sets. The acquired results confirm that the proposed framework provides an accurate and robust performance for recognizing MCI cases and outperforms previous approaches. Based on the obtained results, it is possible to be developed in order to use as a computer-aided diagnosis (CAD) tool for clinical purposes.

Classification of retinoblastoma-1 gene mutation with machine learning-based models in bladder cancer

PurposeThis study aims to evaluate the potential of machine learning algorithms built with radiomics features from computed tomography urography (CTU) images that classify RB1 gene mutation status in bladder cancer. MethodThe study enrolled CTU images of 18 patients with and 54 without RB1 mutation from a public database. Image and data preprocessing were performed after data augmentation. Feature selection steps were consisted of filter and wrapper methods. Pearson’s correlation analysis was the filter, and a wrapper-based sequential feature selection algorithm was the wrapper. Models with XGBoost, Random Forest (RF), and k-Nearest Neighbors (kNN) algorithms were developed. Performance metrics of the models were calculated. Models’ performances were compared by using Friedman’s test. Results8 features were selected from 851 total extracted features. Accuracy, sensitivity, specificity, precision, recall, F1 measure and AUC were 84%, 80%, 88%, 86%, 80%, 0.83 and 0.84, for XGBoost; 72%, 80%, 65%, 67%, 80%, 0.73 and 0.72 for RF; 66%, 53%, 76%, 67%, 53%, 0.60 and 0.65 for kNN, respectively. XGBoost model had outperformed kNN model in Friedman’s test (p = 0.006). ConclusionsMachine learning algorithms with radiomics features from CTU images show promising results in classifying bladder cancer by RB1 mutation status non-invasively.

Alterations in Patients With First-Episode Depression in the Eyes-Open and Eyes-Closed Conditions: A Resting-State EEG Study.

Altered resting-state EEG activity has been repeatedly reported in major depressive disorder (MDD), but no robust biomarkers have been identified until now. The poor consistency of EEG alterations may be due to inconsistent resting conditions; that is, the eyes-open (EO) and eyes-closed (EC) conditions. Here, we explored the effect of the EO and EC conditions on EEG biomarkers for discriminating MDD subjects and healthy control (HC) subjects. EEG data were recorded from 30 first-episode MDD and 26 HC subjects during an 8-min resting-state session. The features were extracted using spectral power, Lempel-Ziv complexity, and detrended fluctuation analysis. Significant features were further selected via the sequential backward feature selection algorithm. Support vector machine (SVM), logistic regression, and linear discriminate analysis were used to determine a better resting condition to provide more reliable estimates for identifying MDD. Compared with the HC group, we found that the MDD group exhibited widespread increased β and γ powers ( ) in both conditions. In the EO condition, the MDD group showed increased complexity and scaling exponents in the α band relative to HC subjects ( ). The best classification performance of the combined feature sets was found in the EO condition, with the leave-one-out classification accuracy of 89.29%, sensitivity of 90.00%, and specificity of 88.46% using SVM with the linear kernel classifier when the threshold was set to 0.7, followed by the β and γ spectral features with an average accuracy of 83.93%. Overall, EO and EC conditions indeed affected the between-group variance, and the EO condition is suggested as the more separable resting condition to identify depression. Specially, the β and γ powers are suggested as potential biomarkers for first-episode MDD.

Evaluating the event-related potentials relevant to two groups of the Quran memorizers and non-memorizers during the retrieval phase of the visual memory

Quran memorizing leads to locating humans in a relaxation state, resulting in a slower reaction time (RT). This changes the event time and amplitude of the waves of event-related potentials (ERP). Nevertheless, studies on Quran memorizing have only employed the EEG signals to investigate the effects of Quran memorizing on the life quality of autistic patients and people with stroke. In this study, we accordingly analyzed ERP waves, especially P200, N200, and P300 waves, in two groups of Quran memorizers and non-memorizers recorded during the retrieval phases of pattern recognition memory (PRM) test. In fact, we first selected optimal channels according to the energy topography of ERP signals in the retrieval phase. Then, we extracted a set of features from ERP signals using discrete wavelet transform. Finally, we also evaluated the separability of these ERP signals using multilayer perceptron neural network (MLPNN) and support vector machine developed by the optimal combinations of features selected by T-test analysis, the sequential feature selection (SFS), and the genetic algorithm. The outcome of this analysis generally indicated that MLPNN developed by the optimal features obtained from SFS algorithm had higher accuracy to separate two groups of Quran memorizers and non-memorizers (99.3% and 93.2% for training and testing sets, respectively). Therefore, this finding generally represented that the Quran memorizing as a factor involved in the visual memory can be used as a protocol to change the attention and concentration in humans.

An improved feature selection approach for chronic heart disease detection

Irrelevant feature in heart disease dataset affects the performance of binary classification model. Consequently, eliminating irrelevant and redundant feature (s) from training set with feature selection algorithm significantly improves the performance of classification model on heart disease detection. Sequential feature selection (SFS) is successful algorithm to improve the performance of classification model on heart disease detection and reduces the computational time complexity. In this study, sequential feature selection (SFS) algorithm is implemented for improving the classifier performance on heart disease detection by removing irrelevant features and training a model on optimal features. Furthermore, exhaustive and permutation based feature selection algorithm are implemented and compared with SFS algorithm. The implemented and existing feature selection algorithms are evaluated using real world Pima Indian heart disease dataset and result appears to prove that the SFS algorithm outperforms as compared to exhaustive and permutation based feature selection algorithm. Overall, the result looks promising and more effective heart disease detection model is developed with accuracy of 99.3%.

Texture analysis of muscle MRI: machine learning-based classifications in idiopathic inflammatory myopathies

To develop a machine learning (ML) model that predicts disease groups or autoantibodies in patients with idiopathic inflammatory myopathies (IIMs) using muscle MRI radiomics features. Twenty-two patients with dermatomyositis (DM), 14 with amyopathic dermatomyositis (ADM), 19 with polymyositis (PM) and 19 with non-IIM were enrolled. Using 2D manual segmentation, 93 original features as well as 93 local binary pattern (LBP) features were extracted from MRI (short-tau inversion recovery [STIR] imaging) of proximal limb muscles. To construct and compare ML models that predict disease groups using each set of features, dimensional reductions were performed using a reproducibility analysis by inter-reader and intra-reader correlation coefficients, collinearity analysis, and the sequential feature selection (SFS) algorithm. Models were created using the linear discriminant analysis (LDA), quadratic discriminant analysis (QDA), support vector machine (SVM), k-nearest neighbors (k-NN), random forest (RF) and multi-layer perceptron (MLP) classifiers, and validated using tenfold cross-validation repeated 100 times. We also investigated whether it was possible to construct models predicting autoantibody status. Our ML-based MRI radiomics models showed the potential to distinguish between PM, DM, and ADM. Models using LBP features provided better results, with macro-average AUC values of 0.767 and 0.714, accuracy of 61.2 and 61.4%, and macro-average recall of 61.9 and 59.8%, in the LDA and k-NN classifiers, respectively. In contrast, the accuracies of radiomics models distinguishing between non-IIM and IIM disease groups were low. A subgroup analysis showed that classification models for anti-Jo-1 and anti-ARS antibodies provided AUC values of 0.646–0.853 and 0.692–0.792, with accuracy of 71.5–81.0 and 65.8–78.3%, respectively. ML-based TA of muscle MRI may be used to predict disease groups or the autoantibody status in patients with IIM and is useful in non-invasive assessments of disease mechanisms.

A major depressive disorder classification framework based on EEG signals using statistical, spectral, wavelet, functional connectivity, and nonlinear analysis

BackgroundMajor depressive disorder (MDD) is a prevalent mental illness that is diagnosed through questionnaire-based approaches; however, these methods may not lead to an accurate diagnosis. In this regard, many studies have focused on using electroencephalogram (EEG) signals and machine learning techniques to diagnose MDD. New methodThis paper proposes a machine learning framework for MDD diagnosis, which uses different types of EEG-derived features. The features are extracted using statistical, spectral, wavelet, functional connectivity, and nonlinear analysis methods. The sequential backward feature selection (SBFS) algorithm is also employed to perform feature selection. Various classifier models are utilized to select the best one for the proposed framework. ResultsThe proposed method is validated with a public EEG dataset, including the EEG data of 34 MDD patients and 30 healthy subjects. The evaluation of the proposed framework is conducted using 10-fold cross-validation, providing the metrics such as accuracy (AC), sensitivity (SE), specificity (SP), F1-score (F1), and false discovery rate (FDR). The best performance of the proposed method has provided an average AC of 99%, SE of 98.4%, SP of 99.6%, F1 of 98.9%, and FDR of 0.4% using the support vector machine with RBF kernel (RBFSVM) classifier. Comparison with existing methodsThe obtained results demonstrate that the proposed method outperforms other approaches for MDD classification based on EEG signals. ConclusionsAccording to the obtained results, a highly accurate MDD diagnosis would be provided using the proposed method, while it can be utilized to develop a computer-aided diagnosis (CAD) tool for clinical purposes.

A Comparative Analysis of Machine Learning Algorithms to Build a Predictive Model for Detecting Diabetes Complications

Diabetes complications have a significant impact on patients’ quality of life. The objective of this study was to predict which patients were more likely to be in a complicated health condition at the time of admission to allow for the early introduction of medical interventions. The data were 644 electronic health records from Alsukari Hospital collected from January 2018 to April 2019. We used the following machine learning methods: logistic regression, random forest, and k-nearest neighbor (KNN). The logistic regression algorithm performed better than the other algorithms achieving an accuracy of 81%, recall of 81%, and F1 score of 75%. Also, attributes such as infection years, swelling, diabetic ketoacidosis, and diabetic septic foot were significant in predicting diabetes complications. This model can be useful for the identification of patients requiring additional care to limit the complications and help practitioners in making decisions on whether the patient should be hospitalized or sent home. Furthermore, we used the sequential feature selection(SFS) algorithm which reduced the features to six, which is fewer than any model built before to predict diabetes complications. The primary goal of this study was achieved. The model had fewer attributes which means we have a simple and understandable model in addition to, it has a better performance.

Sequential Feature Selection and Machine Learning Algorithm-Based Patient’s Death Events Prediction and Diagnosis in Heart Disease

Due to the accessibility of data with multiple features, many feature determination techniques available in written form. These features promote data with extremely high measurement values. The feature determination strategy provides us with a way to reduce calculation time, improve prediction execution, and have a better understanding of data in machine learning, as well as a way to recognize applications. As pointed out by related works that have been reviewed, in general, existing works only focus on amplifying classification accuracy. For real-world applications, the selected subset of features must be continuous instead. In this research, proposes a sequential feature selection algorithm for detecting death events in heart disease patients during treatment to select the most important features. Several machine learning algorithms (LDA, RF, GBC, DT, SVM, and KNN) are used. In addition, the accuracy obtained by this method (SFS) is compared with the accuracy of the classifier. The confusion matrix, ROC curve, precision, recall rate, and f1-score are also calculated to verify the results obtained by the SFS algorithm. The experimental results show that for Random Forest Classifier_FS, the SFS method reaches 86.67% accuracy.

Computationally identifying hot spots in protein-DNA binding interfaces using an ensemble approach

BackgroundProtein-DNA interaction governs a large number of cellular processes, and it can be altered by a small fraction of interface residues, i.e., the so-called hot spots, which account for most of the interface binding free energy. Accurate prediction of hot spots is critical to understand the principle of protein-DNA interactions. There are already some computational methods that can accurately and efficiently predict a large number of hot residues. However, the insufficiency of experimentally validated hot-spot residues in protein-DNA complexes and the low diversity of the employed features limit the performance of existing methods.ResultsHere, we report a new computational method for effectively predicting hot spots in protein-DNA binding interfaces. This method, called PreHots (the abbreviation of Predicting Hotspots), adopts an ensemble stacking classifier that integrates different machine learning classifiers to generate a robust model with 19 features selected by a sequential backward feature selection algorithm. To this end, we constructed two new and reliable datasets (one benchmark for model training and one independent dataset for validation), which totally consist of 123 hot spots and 137 non-hot spots from 89 protein-DNA complexes. The data were manually collected from the literature and existing databases with a strict process of redundancy removal. Our method achieves a sensitivity of 0.813 and an AUC score of 0.868 in 10-fold cross-validation on the benchmark dataset, and a sensitivity of 0.818 and an AUC score of 0.820 on the independent test dataset. The results show that our approach outperforms the existing ones.ConclusionsPreHots, which is based on stack ensemble of boosting algorithms, can reliably predict hot spots at the protein-DNA binding interface on a large scale. Compared with the existing methods, PreHots can achieve better prediction performance. Both the webserver of PreHots and the datasets are freely available at: http://dmb.tongji.edu.cn/tools/PreHots/.

Controlled differential evolution based detection of neovascularization on optic disc using support vector machine.

The severe stage of Diabetic Retinopathy (DR) is characterized by the growth of new blood vessels which is called Neovascularization (NV). The abnormally grown blood vessels on the disc are breakable in nature thus the patient is at high risk of sudden blindness. Therefore, the significance of early and accurate detection of Neovascularization on Disc (NVD) should not be neglected. This paper presents an automatic detection of the optic disc using a Controlled Differential Evolution (CDE) algorithm. Further, the Region of Interest (ROI) is created automatically by extending the extreme boundaries of the optic disc by 100 pixels to ensure the presence of NV around the optic disc also. From the ROI so created, blood vessels are segmented using multi-scale Gabor filtering and subsequently, both the morphological and textural features are extracted. Simultaneously, statistical features are directly extracted from the earlier created ROI. Finally, the fundus image is classified by a Support Vector Machine (SVM) using the extracted features from all three feature sets. From each individual image, 16 features are extracted and the feature dimension is reduced to 13 using a sequential backward feature (SBF) selection algorithm. The optimal features are obtained from a total of 205 fundus images, which consists of 99 NVD positive and 106 NVD negative images. This paper attains an average accuracy of 98.75%, the specificity of 100%, the sensitivity of 97.8%, and area under the curve (AUC) as 100% when tested over image selected randomly.

Machine Learning for Pavement Performance Modelling in Warm Climate Regions

Accurate pavement performance modelling is an essential requirement for cost-effective pavement design and enhances pavement management decision making. Due to the complexity of the pavement structure and pavement response, dominant failure mechanisms may vary depending on the climate region: cold or warm. This study investigated the significance of pavement design factors on pavement performance in warm regions and compared them to set of factors previously identified for cold regions. An artificial neural network (ANN) supported by a forward sequential feature selection algorithm was employed to identify the most significant design factors prevailing in warm climate regions using data extracted from the Long-Term Pavement Performance database. In addition, five machine learning techniques were utilized to model the pavement performance in warm regions, namely: regression tree, support vector machine, ensembles, Gaussian process regression, and ANN. Moreover, conventional regression modelling was used for comparison assessment. The analysis revealed seven design factors that are significantly impacting asphalt pavement performance in warm regions: initial roughness, relative humidity, average wind velocity, average albedo, average emissivity, traffic volume, and pavement structural capacity. The results indicate that pavement performance in warm climate regions is dominated by different environmental factors than those found for cold climate regions. The ANN modelling technique produced the most accurate asphalt pavement performance models.