Polynomial Support Vector Machine Research Articles

Arabic Handwritten Word Recognition Based on Stationary Wavelet Transform Technique using Machine Learning

This paper is aimed at improving the performance of the word recognition system (WRS) of handwritten Arabic text by extracting features in the frequency domain using the Stationary Wavelet Transform (SWT) method using machine learning, which is a wavelet transform approach created to compensate for the absence of translation invariance in the Discrete Wavelets Transform (DWT) method. The proposed SWT-WRS of Arabic handwritten text consists of three main processes: word normalization, feature extraction based on SWT, and recognition. The proposed SWT-WRS based on the SWT method is evaluated on the IFN/ENIT database applying the Gaussian, linear, and polynomial support vector machine, the k-nearest neighbors, and ANN classifiers. ANN performance was assessed by applying the Bayesian Regularization (BR) and Levenberg-Marquardt (LM) training methods. Numerous wavelet transform (WT) families are applied, and the results prove that level 19 of the Daubechies family is the best WT family for the proposed SWT-WRS. The results also confirm the effectiveness of the proposed SWT-WRS in improving the performance of handwritten Arabic word recognition using machine learning. Therefore, the suggested SWT-WRS overcomes the lack of translation invariance in the DWT method by eliminating the up-and-down samplers from the proposed machine learning method.

Image enhancement methods on extracted texture features to detect prostate cancer by employing machine learning techniques

Prostate cancer (PCa) is the second most diagnosed cancer of men all over the world. The aim of this research was to improve PCa detection based on image enhancement methods including image adjustment and morphological erosion operations and then compute the texture features. We then employed robust Machine Learning (ML) techniques such as the Naïve Bayes, Support Vector Machine (SVM) kernels: Polynomial, Radial Base Function (RBF), Gaussian and Decision Tree (DT) based on extracted texture features. The Cross validation (Jack-Knife k-Fold) was performed, and performance was evaluated in term of Receiver Operating Curve (ROC), Specificity, Sensitivity, Positive Predictive Value (PPV), Negative Predictive Value (NPV), False Positive Rate (FPR). The highest detection performance based on morphological erosion operation on texture features was obtained using SVM polynomial with sensitivity (99.82%), specificity (96.63%), accuracy (98.59%) and AUC (0.9994). The image adjustment methods yielded the highest detection performance with sensitivity, specificity, and accuracy of 100% and AUC of 1.00 using ML SVM selected kernels. The results reveal that proposed image enhancement methods have the potential to accurately predict PCa. Thus, this approach can be better utilized by clinicians for early prediction of PCa for further diagnostic and treatment of the patients.

Detection of Tonic-Clonic Seizures using Wavelet Entropy of Scalp EEG.

Epilepsy is the most common chronic neurologic disorder characterized by the recurrence of unprovoked seizures. These seizures are paroxysmal events that result from abnormal neuronal discharges and are categorized into various types based on the clinical manifestations and localization. Tonic-Clonic seizures (TCSZ) may lead to injuries, and constitute the major risk factor for sudden unexpected death in epilepsy (SUDEP), especially in unattended patients. Therapeutic decisions and clinical trials rely on Video EEG which is not practical outside of clinical setting. In this study, wavelet entropy of scalp EEG signals are utilized to discriminate the seizures with and without clinical manifestations. The scalp EEG records from the publically available Temple University Hospital (TUH) dataset are considered for this work. A sevenlevel, fourth order Daubechies (db4) wavelet is utilized for the decomposition of first four seconds of scalp EEG during seizures. The entropy is extracted from the resultant coefficients and are used to develop SVM based models. Most of the extracted features found to have significant differences (p<0.05). The results show that polynomial SVM model achieves an accuracy of 95.5%, positive predictive value (PPV) of 99.4%, negative predictive value (NPV) of 91.57% and F-Score of 95.9%. Therefore, the proposed approach could be a support in detecting life-threatening seizures.

Algorithms for Classification of Signals Derived From Human Brain

Comparison of the Accuracy of different off-line methods for classification Electroencephalograph (EEG) signals, obtained from Brain-Computer Interface (BCI) devices are investigated in this paper. BCI is a technology that allows people to interact directly or indirectly with their environment only by using brain activity. But, the method of signal acquisition is non-invasive, resulting in significant data loss. In addition, the received signals do not contain only useful information. All this requires careful selection of the method for the classification of the received signals. The main purpose of this paper is to provide a fair and extensive comparison of some commonly employed classification methods under the same conditions so that the assessment of different classifiers will be more convictive. In this study, we investigated the accuracy of the classification of the received signals with classifiers based on AdaBoost (AB), Decision Tree (DT), k-Nearest Neighbor (kNN), Gaussian SVM, Linear SVM, Polynomial SVM, Random Forest (RF), Random Forest Regression ( RFR ). We used only basic parameters in the classification, and we did not apply fine optimization of the classification results. The obtained results show suitable algorithms for the classification of EEG signals. This would help young researchers to achieve interesting results in this field faster.

Sentiment Analysis on Twitter Data: Comparative Study on Different Approaches

Social media has become incredibly popular these days for communicating with friends and for sharing opinions. According to current statistics, almost 2.22 billion people use social media in 2016, hich is roughly one third of the world population and three times of the entire population in Europe. In social media people share their likes, dislikes, opinions, interests, etc. so it is possible to know about a person’s thoughts about a specific topic from the shared data in social media. Since, twitter is one of the most popular social media in the world; it is a very good source for opinion mining and sentiment analysis about different topics. In this research, SVM with different kernel functions and Adaboost are experimented using CPD and Chi-square feature extraction techniques to explore the best sentiment classification model. The reported average accuracy of Adaboost for Chi-square and CPD are 70.2% and 66.9%. The SVM radial basis kernel and polynomial kernel with Chi-square n-grams reported average accuracy of 73.73% and 68.67% respectively. Among the performed experimentation, SVM sigmoid kernel with Chi-square n-grams provided the maximum accuracy that is 74.4%.

Identification of vacancy defects in carbon nanotubes using vibration analysis and machine learning

Point vacancies are the most common defects in a carbon nanotube (CNT) lattice. However, there is no standard technique to identify these defects. Moreover, the effect of such vacancies on the vibrational properties of CNTs is unknown. Therefore, this paper presents the first-ever technique for identification of vacancy defects in single-walled carbon nanotubes (SWCNT) using their vibrational analyses and machine learning. 240 SWCNT samples were modelled using molecular-structural-mechanics-approach and their modes obtained using finite element analyses. Then, an analytical expression for fundamental vibration frequency of a vacancy defective SWCNT was derived using log-linear multiple regression. This frequency is found to decrease with: (i) increase in the magnitude of point vacancies; (ii) decrease in diameter; and (iii) increase in length. A polynomial support vector machine (SVM) was developed that successfully classifies pristine and vacancy defective SWCNT samples at test accuracy greater than 90%. The proposed technique for identification of point vacancies combines the developed SVM classification with inverse estimation using the derived expression. This technique will be useful for the most crucial characterization of SWCNTs, i.e., characterizing pristine and vacancy defective samples. The derived expression will be useful in predicting vibration response of vacancy defective SWCNTs, and deciding their applications.

Predicting time-varying, speed-varying dilemma zones using machine learning and continuous vehicle tracking

This paper proposes an innovative framework of predicting driver behavior under varying dilemma zone conditions using artificial intelligence-based machine learning. The framework utilizes multiple machine learning techniques to process vehicle attribute data (e.g., speed, location, and time-of-arrival) collected at the onset of the yellow indication, and eventually predicts drivers’ stop-or-go decision based on the data. A linear SVM was used to extract through vehicles from all approaching vehicles detected from radar sensors. A hierarchical clustering method was utilized to classify different traffic patterns by time-of-day. Finally, driver behavior prediction models were developed using three machine learning techniques (i.e., linear SVM, polynomial SVM, and ANN) widely adopted for binary classification problems. Model validation results showed that all the prediction models perform well with high prediction accuracies. The ANN model, which showed the best performance among the three, was selected to represent dilemma zone boundaries. Results show that the dilemma zone start- and end-points would both locate further from the stop bar with higher approaching speeds. Furthermore, the dilemma zone end-point would be more sensitive to the approaching speed than the start-point is. As a result, the dilemma zone length would become longer with higher approaching speeds. Results also showed that the dilemma zone length and location would vary by time-of-day regardless of the speed of approaching vehicles. The analysis showed that the dilemma zone length would be longer and its location would be much further from the stop bar for vehicles arriving during rush hours, as compared to those arriving during non-rush or nighttime hours. This indicates that drivers’ decision location to stop or go (when they are faced with a dilemma zone situation) is distributed farther from the intersection stop bar during rush hours. The proposed method shows an effective way of predicting driver behavior on signalized intersections. It is expected for the transportation agencies to use the method to improve intersection signal operations more effectively and safely.

Facial Emotion Recognition with Sparse Coding Descriptor

With the Corona Virus Disease 2019 (COVID-19) global pandemic ravaging the world, all sectors of life were affected including education. This led to many schools taking distance learning through the use of computer as a safer option. Facial emotion means a lot to teacher’s assessment of his performance and relation to his students. Researchers has been working on improving the face monitoring and human machine interface. In this paper we presented different types of face recognition methods which include: Principal component analysis (PCA); Speeded Up Robust Features (SURF); Local binary pattern (LBP); Gray-Level Co-occurrence Matrix (GLCM) and also the group sparse coding (GSC) and come up with the fusion of LBP, PCA, SURF GLCM with GSC. Linear Kernel Support Vector Machine (LSVM) Classifier out-performed Polynomial, RBF and Sigmoid kernels SVM in the emotion classification. Results obtained from experiments indicated that, the new fusion method is capable of differentiating different types of face emotions with higher accuracy compare with the state-of-the-art methods currently available.Keywords— Face Recognition, Principal Component Analysis, Speeded Up Robust Features, Grey Level Co-Occurrence Matrix

EVALUATION OF MACHINE LEARNING CLASSIFIERS FOR MAPPING FALCATA PLANTATIONS IN SENTINEL-2 IMAGE

Abstract. Efficient and accurate mapping of forest and other industrial tree plantations (ITPs) is essential to ensure better monitoring and sustainable management of these plantations. In Caraga Region, Mindanao, Philippines, ITPs planted with Falcata (Paraserianthes falcataria (L.) Nielsen) are widespread and has contributed to more than 50% of the nationwide log production. At present, there is limited information on the location and extent of existing plantations. This provides an opportunity to evaluate satellite remote sensing approaches for mapping these plantations from images, particularly those provided by the Sentinel-2 mission. The objective of this study is to evaluate machine learning classifiers for mapping Falcata plantations in Sentinel-2 image using a 9 × 9 km2 study area in Caraga Region. It also aims to find the best classifier that can provide acceptable levels of accuracy by utilizing only the four bands of the 10-m spatial resolution and the 9 bands of the 20-m spatial resolution Level 2A Sentinel-2 image, respectively. The following classifiers and their variants were evaluated: Linear Support Vector Machine (SVM), Polynomial SVM, Radial Basis Function (RBF) SVM, Artificial Neural Network (Neural Net), Random Forest (RF), and Maximum Likelihood (ML). One or more of these classifiers have been successfully used in in natural and plantation forest mapping, including tree species classification from remotely sensed images. However, their performance and accuracy in detecting and discriminating Falcata plantations is yet to be evaluated. Results of the evaluation showed that the ML classifier has the highest overall accuracy (OA) of 90.90% and has more consistent values for Producer’s Accuracy (PA), and User’s Accuracy (UA) for Falcata and Non-Falcata classes, and hence, provides better Falcata classification results than the other classifiers when the 10-m spatial resolution Sentinel-2 image was used. The accuracy assessment of the 20-m subset classification provides relative different results from that of the 10-m subset, perhaps due to the inclusion of more bands. The highest OA was obtained by the Linear and RBF SVM classifiers at 92.05% each. The SVM classifiers have consistent performance and produce more accurate classification results than the other classifiers (i.e., more than 90% OA, PA, and UA). From these results, it can be concluded that Maximum Likelihood classifier is best to use for Falcata mapping using the 10-m spatial resolution Sentinel-2 image. For the 20-m resolution image, any of the two SVMs (linear or RBF) is more appropriate to use. However, it should be noted that these results are based on classifications where default parameters were used. Improvement in the classification accuracy may be achieved if these parameters were optimized.

A novel automated autism spectrum disorder detection system

Autism spectrum disorder (ASD) is a neurological and developmental disorder that begins early in childhood and lasts throughout a person’s life. Autism is influenced by both genetic and environmental factors. Lack of social interaction, communication problems, and a limited range of behaviors and interests are possible characteristics of autism in children, alongside other symptoms. Electroencephalograms provide useful information about changes in brain activity and hence are efficaciously used for diagnosis of neurological disease. Eighteen nonlinear features were extracted from EEG signals of 40 children with a diagnosis of autism spectrum disorder and 37 children with no diagnosis of neuro developmental disorder children. Feature selection was performed using Student’s t test, and Marginal Fisher Analysis was employed for data reduction. The features were ranked according to Student’s t test. The three most significant features were used to develop the autism index, while the ranked feature set was input to SVM polynomials 1, 2, and 3 for classification. The SVM polynomial 2 yielded the highest classification accuracy of 98.70% with 20 features. The developed classification system is likely to aid healthcare professionals as a diagnostic tool to detect autism. With more data, in our future work, we intend to employ deep learning models and to explore a cloud-based detection system for the detection of autism. Our study is novel, as we have analyzed all nonlinear features, and we are one of the first groups to have uniquely developed an autism (ASD) index using the extracted features.

Analisis Sentimen Mahasiswa untuk Perbaikan Desain Afektif Ruang Kelas Jurusan Teknik Industri, Universitas Trisakti

Improving the affective classroom design is essential to maximize student performance and learning achievement. The comfort and performance achievement of Trisakti University Industrial Engineering students are influenced by the affective design of classrooms. This study aimed to use sentiment analysis in the classification of students’ perceptions of the affective classroom design. Student sentiment classification is done using a Support Vector Machine (SVM). The questionnaire analysis results also showed perceptions about the subjects that were considered the most difficult (statistics). The classrooms had a positive sentiment: FGTSC, a sample for the next stage of classroom design formulation. The results show what impressions and things the students consider in choosing the FGTSC class. Some examples of the dominant kansei word are “comfortable,” this shows that students really care about the comfort of a classroom in the learning process. The word kansei for the design concept was collected from students' perceptions of the “positive” label. Design elements that need to be improved include equipment that is less comfortable to use, less lighting, walls with graffiti and uncomfortable seating. The classification results using three SVM types Kernel linear, radial and polynomial obtained linear have the best accuracy value (76%). These results indicate that the classification of student sentiment has the maximum results with SVM linear kernel (dot) type. This method will be used in classifying student sentiment on the results of improving classroom design.

Identifying the Risk Factors Associated with Nursing Home Residents' Pressure Ulcers Using Machine Learning Methods.

Background: Machine learning (ML) can keep improving predictions and generating automated knowledge via data-driven predictors or decisions. Objective: The purpose of this study was to compare different ML methods including random forest, logistics regression, linear support vector machine (SVM), polynomial SVM, radial SVM, and sigmoid SVM in terms of their accuracy, sensitivity, specificity, negative predictor values, and positive predictive values by validating real datasets to predict factors for pressure ulcers (PUs). Methods: We applied representative ML algorithms (random forest, logistic regression, linear SVM, polynomial SVM, radial SVM, and sigmoid SVM) to develop a prediction model (N = 60). Results: The random forest model showed the greatest accuracy (0.814), followed by logistic regression (0.782), polynomial SVM (0.779), radial SVM (0.770), linear SVM (0.767), and sigmoid SVM (0.674). Conclusions: The random forest model showed the greatest accuracy for predicting PUs in nursing homes (NHs). Diverse factors that predict PUs in NHs including NH characteristics and residents’ characteristics were identified according to diverse ML methods. These factors should be considered to decrease PUs in NH residents.

Prediction of Worse Functional Status After Surgery for Degenerative Cervical Myelopathy: A Machine Learning Approach.

Surgical decompression for degenerative cervical myelopathy (DCM) is one of the mainstays of treatment, with generally positive outcomes. However, some patients who undergo surgery for DCM continue to show functional decline. To use machine learning (ML) algorithms to determine predictors of worsening functional status after surgical intervention for DCM. This is a retrospective analysis of prospectively collected data. A total of 757 patients enrolled in 2 prospective AO Spine clinical studies, who underwent surgical decompression for DCM, were analyzed. The modified Japanese Orthopedic Association (mJOA) score, a marker of functional status, was obtained before and 1 yr postsurgery. The primary outcome measure was the dichotomized change in mJOA at 1 yr according to whether it was negative (worse functional status) or non-negative. After applying an 80:20 training-testing split of the dataset, we trained, optimized, and tested multiple ML algorithms to evaluate algorithm performance and determine predictors of worse mJOA at 1 yr. The highest-performing ML algorithm was a polynomial support vector machine. This model showed good calibration and discrimination on the testing data, with an area under the receiver operating characteristic curve of 0.834 (accuracy: 74.3%, sensitivity: 88.2%, specificity: 72.4%). Important predictors of functional decline at 1 yr included initial mJOA, male gender, duration of myelopathy, and the presence of comorbidities. The reasons for worse mJOA are frequently multifactorial (eg, adjacent segment degeneration, tandem lumbar stenosis, ongoing neuroinflammatory processes in the cord). This study successfully used ML to predict worse functional status after surgery for DCM and to determine associated predictors.

National-scale spatial variations of soil magnetic susceptibility in Brazil

Brazil presents high soil variability, however, the soil magnetic susceptibility (χ) distribution on national scale is yet to be investigated. Thus, soil data (218 samples from the literature and 36 original samples) were utilized to estimate soil low frequency χ (χlf) across Brazil through machine learning algorithms (Generalized Additive Model Using Splines, Generalized Linear Model, Random Forest and Support Vector Machine) in order to assess the applicability of such methodology on a nationwide scale. Two variable groups were tested: texture and Fe2O3 contents – Group 1; and texture, Fe2O3, Al2O3, SiO2 contents – Group 2. The best performing model was applied to a database containing 3817 and 3973 georeferenced soil samples from superficial and subsuperficial horizons, respectively, obtained from soil surveys conducted by the Brazilian Agricultural Research Corporation. Polynomial Support Vector Machine using Group 2 variables presented the best performance (R2 = 0.91). Variable importance measurements showed that χlf is chiefly related to Fe2O3 and clay content. The spatial distribution of χlf on national scale varied from very high (>150 × 10−6 m3 kg−1) to very low (≤5 × 10−6 m3 kg−1), considering the interpretation bands that we propose for Brazilian soils. Extensive areas of magnetic soils were associated with their formation upon mafic rocks or iron-rich formations. Despite few exceptions, the observed tendency shows that χlf increases parallel to soil red color, drainage and weathering degree, and hence Rhodic Nitisol, Umbric Nitisol, Rhodic Ferralsol and Umbric Ferralsol presented the highest χlf values.

Intramuscular Fat Prediction Using Color and Image Analysis of Bísaro Pork Breed.

This work presents an analytical methodology to predict meat juiciness (discriminant semi-quantitative analysis using groups of intervals of intramuscular fat) and intramuscular fat (regression analysis) in Longissimus thoracis et lumborum (LTL) muscle of Bísaro pigs using as independent variables the animal carcass weight and parameters from color and image analysis. These are non-invasive and non-destructive techniques which allow development of rapid, easy and inexpensive methodologies to evaluate pork meat quality in a slaughterhouse. The proposed predictive supervised multivariate models were non-linear. Discriminant mixture analysis to evaluate meat juiciness by classified samples into three groups—0.6 to 1.1%; 1.25 to 1.5%; and, greater than 1.5%. The obtained model allowed 100% of correct classifications (92% in cross-validation with seven-folds with five repetitions). Polynomial support vector machine regression to determine the intramuscular fat presented R2 and RMSE values of 0.88 and 0.12, respectively in cross-validation with seven-folds with five repetitions. This quantitative model (model’s polynomial kernel optimized to degree of three with a scale factor of 0.1 and a cost value of one) presented R2 and RSE values of 0.999 and 0.04, respectively. The overall predictive results demonstrated the relevance of photographic image and color measurements of the muscle to evaluate the intramuscular fat, rarther than the usual time-consuming and expensive chemical analysis.

Classification and comparison of malignancy detection of cervical cells based on nucleus and textural features in microscopic images of uterine cervix

Cervical cancer is one of the major causes of death among women worldwide. Pap smear is the cytology based screening test which is used to detect abnormal cervical cells including pre-cancerous lesions. Accurate classification of Pap smear images is one of the challenging tasks in medical image processing and its performance can be enhanced by extracting and selecting the well-defined features and classifiers. The irregular chromatin structure is one of the prominent diagnostic features of abnormal cells. Classification is performed based on the extracted textural features and the benchmark Herlev dataset features. Radial basis function (RBF), polynomial and sigmoid SVM kernels are used for classification and comparison is performed with the features in benchmark database. Precision, recall and accuracy were calculated for all the combinations of the features. The classifier gives promising results when benchmark features are combined with textural features and also benchmark nucleus features with textural features. An effective integration of features for cervical cell classification had given good results for fast abnormal cell detection and primary Pap smear cell image classification.

Classification of Indian Classical Music With Time-Series Matching Deep Learning Approach

Music is a heavenly way of expressing feelings about the world. The language of music has vast diversity. For centuries, people have indulged in debates to stratisfy between Western and Indian Classical Music. But through this paper, an understanding can be fabricated while differentiating the types of Indian Classical Music. Classical music is one of the essential characteristics of Indian Cultural Heritage. Indian Classical Music is divided into two major parts, i.e. Hindustani and Carnatic. Models have been sculptured and trained to classify between Hindustani and Carnatic Music. In this paper, two approaches are used to implement classification models. MFCCs are used as features and implemented models like DNN (1 Layer, 2 Layers, 3 Layers), CNN (1 Layer, 2 Layers, 3 Layers), RNN-LSTM, SVM (Sigmoid, Polynomial & Gaussian Kernel) as one approach. A 3 channels input is created by merging features like MFCC, Spectrogram and Scalogram and implemented models like VGG-16, CNN (1 Layer, 2 Layers, 3 Layers), ResNet-50 as another approach. 3 Layered CNN and RNN-LSTM model performed best among all the approaches.

Automated metabolic assignment: Semi-supervised learning in metabolic analysis employing two dimensional Nuclear Magnetic Resonance (NMR)

Metabolomics is an expanding field of medical diagnostics since many diseases cause metabolic reprogramming alteration. Additionally, the metabolic point of view offers an insight into the molecular mechanisms of diseases. Due to the complexity of metabolic assignment dependent on the 1D NMR spectral analysis, 2D NMR techniques are preferred because of spectral resolution issues. Thus, in this work, we introduce an automated metabolite identification and assignment from 1H-1H TOCSY (total correlation spectroscopy) using real breast cancer tissue. The new approach is based on customized and extended semi-supervised classifiers: KNFST, SVM, third (PC3) and fourth (PC4) degree polynomial. In our approach, metabolic assignment is based only on the vertical and horizontal frequencies of the metabolites in the 1H–1H TOCSY. KNFST and SVM show high performance (high accuracy and low mislabeling rate) in relatively low size of initially labeled training data. PC3 and PC4 classifiers showed lower accuracy and high mislabeling rates, and both classifiers fail to provide an acceptable accuracy at extremely low size (≤9% of the entire dataset) of initial training data. Additionally, semi-supervised classifiers were implemented to obtain a fully automatic procedure for signal assignment and deconvolution of TOCSY, which is a big step forward in NMR metabolic profiling. A set of 27 metabolites were deduced from the TOCSY, and their assignments agreed with the metabolites deduced from a 1D NMR spectrum of the same sample analyzed by conventional human-based methodology.

A Cascaded Classification-Segmentation Reversible System for Computer-Aided Detection and Cells Counting in Microscopic Peripheral Blood Smear Basophils and Eosinophils Images

Computer-aided image analysis has a pivotal role in automated counting and classification of white blood cells (WBCs) in peripheral blood images. Due to their different characteristics, our proposed approach is based on investigating the variations between the basophils and eosinophils in terms of their color histogram, size, and shape before performing the segmentation process. Accordingly, we proposed a cascaded system using a classification-based segmentation process, called classification-segmentation reversible system (CSRS). Prior to applying the CSRS system, a Histogram-based Object to Background Disparity (HOBD) metric was deduced to determine the most appropriate color plane for performing the initial WBC detection (first segmentation). Investigating the local histogram features of both classes resulted in a 92.4% initial classification accuracy using the third-degree polynomial support vector machine (SVM) method. Subsequently, in the proposed CSRS approach, transformation-based segmentation algorithms were developed to fit the specific requirements of each of the two predicted classes. The proposed CSRS system is used, where the images from an initial classification process are fed into a second segmentation process for each class separately. The segmentation results demonstrated a similarity index of 94.9% for basophils, and 94.1% for eosinophils. Moreover, an average counting accuracy of 97.4% for both classes was achieved. In addition, a second classification was carried out after applying the CSRS, achieving a 5.2% increase in accuracy compared to the initial classification process.

Machine learning based congestive heart failure detection using feature importance ranking of multimodal features.

In this study, we ranked the Multimodal Features extracted from Congestive Heart Failure (CHF) and Normal Sinus Rhythm (NSR) subjects. We categorized the ranked features into 1 to 5 categories based on Empirical Receiver Operating Characteristics (EROC) values. Instead of using all multimodal features, we use high ranking features for detection of CHF and normal subjects. We employed powerful machine learning techniques such as Decision Tree (DT), Naïve Bayes (NB), SVM Gaussian, SVM RBF and SVM Polynomial. The performance was measured in terms of Sensitivity, Specificity, Positive Predictive Value (PPV), Negative Predictive Value (NPV), Accuracy, False Positive Rate (FPR), and area under the Receiver Operating characteristic Curve (AUC). The highest detection performance in terms of accuracy and AUC was obtained with all multimodal features using SVM Gaussian with Sensitivity (93.06%), Specificity (81.82%), Accuracy (88.79%) and AUC (0.95). Using the top five ranked features, the highest performance was obtained with SVM Gaussian yields accuracy (84.48%), AUC (0.86); top nine ranked features using Decision Tree and Naïve Bayes got accuracy (84.48%), AUC (0.88); last thirteen ranked features using SVM polynomial obtained accuracy (80.17%), AUC (0.84). The findings indicate that proposed approach with feature ranking can be very useful for automatic detection of congestive heart failure patients and can be very helpful for further decision making by the clinicians and physicians in order to decrease the mortality rate.