Search

Supervised learning has tremendous applications in cancer prediction, patient treatment, business, and engineering. Datasets for supervised learning are often corrupted by noise and non-biological effects, leading to model-overfitting and performance degradation in current methods of binary classification. In this research, we develop a new classification method that fully exploits unique characteristics, such as persistent outliers, anomalies from a batch effect, and hidden relationships between features and their classes in the datasets, hence improving classification performance of current methods. The proposed method, called dynamic multi-hyperplane partitioning (DMP), learns the model by using subclassifiers, which are random in number and each of which uses multiple hyperplanes for decision boundaries. We also develop a method to transform samples to improve classification performance of DMP. We prove that, under a mild condition, accuracy of DMP is as good as or supersedes that of support vector machine (SVM). We test DMP on comprehensive datasets, which span diverse fields of applications, and compare accuracy, sensitivity, specificity, F-measure, and the receiver operating characteristic of DMP to those of competitive baselines, including SVM, random forest, Bayes classifier, gradient boosting tree, and deep-belief nets and neural nets. From the comparison, the proposed method is most accurate in nine out of eleven datasets, when using the mean values alone for comparison. DMP achieves 100% accuracy, 100% sensitivity, and 100% specificity in three datasets. As a generalization, we perform statistical test of difference, at significance levels of 0.05, 0.01 and 0.001. From statistical tests, DMP is the most accurate or one of the most accurate classifiers in nine out of eleven benchmark datasets., and is not the most accurate classifier in the remaining two datasets. The DMP learning method is accurate, simple to implement, and does not require fine-tuning of parameters, making it attractive for binary classification. This research has practical applications and leads to a timely and accurate approach to binary classification in diverse fields.

In recent years, tissue oxygen saturation (StO2) based on hyperspectral imaging (HSI) technology has been deeply studied in the field of affective computing. HSI StO2 is a kind of non-contact physiological signal, which can reveal people's potential emotions. However, at this stage, the number of bands to generate StO2 is too many to achieve real-time emotion detection. Therefore, based on the publicly available HSI stress database, we used a Linear Prediction (LP) algorithm to select only 8 characteristic bands from the original 106 bands to generate StO2 and performed the task of identifying psychological stress and physical stress. The experimental results showed that the recognition rate of StO2 generated based on the selected 8 bands in stress detection is very close to (even higher) that of the original 106 bands, and reaches 84.44% when using the Bayes classifier.

The world is filled with information and getting the right information is a challenging task for internet users and online buyers. Recommender system helps internet users to get their information in a short span of time. It acts as an information extraction system that works behind users to perform their search easier. The recommender system comes under user’s content or item based search, similar users browsing behavior called collaborative and combination of both known as a hybrid. Here collaborative-based approach is adopted which recommends items to their users based on their past browsing behavior. In this article, the User-Item-Rating matrix is formulated concerning user personal profile, rating of the product, and reviews given by the users during their previous browsing history. In this research, user location is considered as an important attribute to group similar users. It also attempts to suppress the scalability and sparsity problems of the traditional collaborative filtering approach. So, the User-Item-Rating (UIR) matrix has considered the location, ratings and reviews for future recommendation. The Navie Bayes classifier algorithm is used to provide accurate topmost recommendations to internet users. The data set is taken from the MovieLens and IMDb database. The accuracy of the recommender system is measured based on the main metric f-measure. The experimental result has proven the improvement of the recommender system with the mentioned added attributes.

In the DC power system, arc fault caused by insulation defects and loose joints is easy to cause fire, explosion, and other safety accidents, but it is not easy to extinguish and detect. Therefore, it is of great significance to study the electrical characteristics of the DC arc and propose effective detection methods. Detection of DC arc faults in photovoltaic systems has attracted widespread attention from scholars at home and abroad. In this paper, several experimental conditions that may affect arc characteristics are analyzed. By using the photovoltaic strings and the arc generator, a lot of field experiments are carried out to obtain the current data during arc fault in an actual photovoltaic system under different conditions of current, cable lengths, and arc gaps. According to the experimental data under different conditions, several characteristic variables from both the time domain and the frequency domain are extracted to identify the arc fault. In this paper, three machine learning methods in the field of artificial intelligence: Random Forest, Support Vector Machine, and Bayes Classifier, are used to learn the experimental samples. Through training, a model is obtained in each method to distinguish between normal current and arc fault current. Then, the three classifier models are validated by hundreds of sets of experimental data under normal and arc conditions, and the classifier with the highest accuracy is found. According to the results, the proposed method can accurately identify arc fault conditions and normal working conditions. Among them, the random forest algorithm has the highest recognition accuracy.

Data Mining is the key to get the best conclusive intuition behind a large chunk of data. WEKA has been the center of the talk because of its wide functionalities which include Data Mining Algorithms like Naϊve Bayes Classifier, Support Vector Machine, Decision Tree, KStar etc. which is helping various industries today to make their work quicker and superior. Even though it provides functionalities there are some disadvantages to the same. This paper aims at overcoming the major disadvantages/drawbacks of the Tool by proposing different approaches. There have been experiments carried out for a few approaches to validate the approach as well.

Federated collaborative filtering (Fed-CF) is a variant of federated learning (FL) models, which can protect user privacy in recommender systems. In Fed-CF, the recommendation model is collectively trained across multiple decentralized clients by exchanging gradients only. However, the decentralized nature of Fed-CF makes it vulnerable to shilling attacks, which can be realized by inserting fake ratings of target items to distort recommendation results. Unfortunately, previous detection algorithms cannot work well in the FL framework, as all original data samples are not disclosed at all. In this paper, we are the first to systematically study the problem of shilling attacks in the context of federated learning, and propose an effective detection method called Federated Shilling Attack Detector (FSAD) to detect shilling attackers in Fed-CF. We first show the feasibility of shilling attacks in Fed-CF. Next, we dedicatedly design four novel features based on exchanged gradients among clients. By incorporating these gradient-based features, we train a semi-supervised Bayes classifier to identify shilling attackers effectively. Finally, we conduct extensive experiments based on real-world datasets to evaluate the performance of our proposed FSAD method. The experimental results show that FSAD can detect shilling attackers in Fed-CF with high accuracy, with the F1 value as high as 0.90 on the Netflix dataset, which approaches the performance of the optimal detector that utilizes complete private user information for detection.

Overall academic performance and social compulsion have created a pressurized psychological state for students. Earlier, the competition level used to be less. As the science evolves, the knowledge and workload of the faculty and the school students have conjointly been evolving to view with the society. Students find it tough at times after they had completed their standards in SSLC (Secondary School Leaving Certificate) and HSC (Higher Secondary School Certificate). Associating with the NEET exam, the result influences a crucial-end on countless students' life. Naive Bayes technique is an emerging discipline, used to analyze the accurate facts and helps in organizing the most stress-causing factors based on the probabilistic parameters. Abstract Reference Object characteristics and correlation are the two main attributes based on which the stress accuracy level is predicted. Information gathered from the survey is analyzed using Bayes classifier and the outcomes are inferred using visualization tools. Based on the output, the socioeconomic and physical stress causing factors have been encoded and prioritized. Frequently stated stress factors should be reduced to help the students to excel in their academic performance and activate the social activities, thereby decreasing individual health issues like migraine headaches, spectacle- wearing and so on.

Machine learning algorithms are nothing but programs in computes that try to forecast the decisions based on data driven assignment. In diagnosis of cancer the goal is to trained algorithm of machine learning that awards the appearance levels from cancer patient, can precisely predict what type and harshness of cancer they have. Breast cancer is the most delicate and deadly among all of the diseases in medicine. In this paper breast cancer classification is implemented using Find-S and Candidate elimination algorithm. These algorithms are used for the breast cancer detection. Navie Bayes Classifier is used for classification of breast cancer.

This paper proposes a novel approach for condition monitoring of power electronic systems. When monitoring the state of a power system, reliability is crucial, as this type of system is usually operated continuously for long periods of time, and as both missed faults as well as false detections can easily become prohibitively expensive. Recently, machine-learning-based methods for fault detection of power systems have gained popularity, since they can overcome many of the constrains of model-based techniques. Most of these methods train classifiers for different states of the system under test, and thus, the problem of fault detection becomes a problem of classification. In this paper we compare two of such recent techniques. We show that despite good results, it cannot reasonably be expected that the state classification is solved perfectly for every instant of time, which makes the application of such classifiers infeasible in practical systems. In order to overcome these issues, we propose to re-formulate the task into one of hybrid—neural and statistical—cross-temporal hypothesis testing. This novel hybrid framework allows us to build upon the previous machine-learning-based classification approaches, and to achieve full reliability on a challenging dataset of fault monitoring measurements for a buck-converter.

Recently, Siamese networks have drawn great attention in visual tracking community because of their balanced accuracy and speed. By comparing the target patch with the candidate windows in a search region, we can track the object to the location where the highest similarity score is obtained. However, in Siamese Network, pairs of training data come from different frames of the same video, and for each search area, the non-semantic background occupies the majority, while semantic entities and distractor occupy less. This imbalanced distribution makes the training model hard to learn instance-level representation, but tending to learn the differences between foreground and background. For the targets those with large differences in the background also achieve high scores, and even some extraneous objects get high scores. To overcome this limitation, we enhance Siamese Network by color histogram based Bayes classifier. This method allows us to identify potentially distracting regions in advance. The risk of drifting is significantly reduced. Experiment results show that our tracker achieves state-of-the-art performance.

In Bayesian learning, designs based on noninformative priors are appropriate when the user cannot confidently identify the data-generating distribution. While such learners cannot achieve the performance of those based on a well-matched subjective prior, they impart a robustness against poor prior selection. The uniform Dirichlet distribution is the true non-informative prior as it has full support over the space of candidate distributions; additionally, it leads to closed-form posteriors. This work applies such a prior to classification using the 0–1 loss, determines the optimal Bayes classifier and the corresponding minimum probability of error, and analyzes the results.

Parkinson's disease is a disorder in the central nervous system which affects in the movement functions of the body. It is a chronic disease with the symptoms growing with time. It generally affects the older people when their symptoms gradually increase to a maximum. The disease can affect the basic functions of the body such as hearing, walking, talking etc. The analysis of this disease can be done with the help of generic machine learning algorithms which produce varying accuracies. Thus, the best one is chosen which will provide the highest accuracy in predicting if the disease is present in the patient or not. The dataset is taken from the UCI machine learning repository namely-Parkinson disease dataset with replicated acoustic features. There are 48 features present in the dataset pertaining to the disease for 240 patients. Various machine learning techniques that are utilized compared their efficiency in the classification. Thus, the best one is chosen with the highest accuracy since the applications in healthcare generally requires more accuracy and efficiencies cannot be compromised. The significant models that are used in this process are naaive bayes classifier, gradient boosting, support vector machines. These techniques can be very powerful for the doctors in order to predict the disease by analysing the features present in the patients.

The analysis of the ECG signal provides important information useful for heart disease diagnosis. Moreover, it can also help in identifying other problems such as neonatal seizures. Considering the nature of ECG patterns, several features can be extracted and used for the purpose of heartbeat and rhythm classification. In this paper, we discuss a set of 13 ECG geometric features for the purpose of identifying five abnormal types of heart beats. The proposed algorithm for feature extraction is based on the Pan-Tompkins QRS model. The MIT-BIH arrhythmia database is used in this study to test the performance of the proposed algorithm. The results show that different types of ECG based features are optimal for different types of heartbeat abnormalities. More importantly, we show that using the developed 13 features, we can identify at least 5 types of abnormalities with an accuracy of more than 92%.

For a successful classification of jaw movement signals three different machine learning methods are implemented and evaluated: A Naive Bayes classifier, a Support Vector Machine (SVM) and a Convolutional Neuronal Network (CNN). In this case the present paper deals with the definition of intuitive jaw movements and their detection by a prototypical system for non-invasive measurement of pressure values in the closed external auditory canal. For this purpose, a database with 1013 recordings was created. The evaluation shows that all three learning methods provide a good recognition level. The Bayes classifier achieves with the used, small dataset with nearly 95% the best detection probability with realtime data. At the same time, the study shows that individual pressure profiles can severely impair recognition. Hence, the creation of a comprehensive database is required. Overall, this paper demonstrates that successful detection and differentiation of up to five jaw movement directions is possible. In total it could be shown using machine learning for jaw movement classification is a stable method with a high recognition level. However, the quality of classification increases with the size and diversity of the dataset.

Breast Cancer is one of the most common disease that is responsible for high number of women’s deaths every year. Despite the fact that cancer is treatable and healable in earliest stages, the huge number of patients are examined with cancer very late. Data mining process and classification are an efficient way to categorise the data particularly in medical fields, where those approaches are broadly used in diagnosis to make decision. This paper presents a performance comparison among the classifiers: Decision tree classifier (J4.8, Simple CART), Bayes classifier (NaiveBayes, Bayesian LogisticRegression). The Wisconsin Breast Cancer(original) dataset is used here and is taken from UCI Machine learning Repository. The main goal is to classify data of both the algorithms in terms of accuracy. Our experimental result shows that among all the classifiers, decision tree classifier i.e. Simple CART (98.13%) gives higher accuracy.

Malware is one of the biggest threats for the privacy and security of the smart-phone users. Android is currently the most popular operating system for smart-phones; consequently, many malwares are directed toward Android devices. Existing techniques for malware detection tend to compromise between accuracy and computational complexity. In this paper, we have proposed a novel technique to monitor the behavior of both malwares and benign applications using system calls and have developed a mathematical model that can detect mobile mal-wares. We have extracted features by sequencing the system calls of these applications. We have proposed a novel way of feature reduction using Gaussian dissimilarity and compared our feature selection technique with existing methods. Using the extracted set of features, we have implemented a machine learning classifier, namely Gaussian Bayes classifier, on two different malware data-sets (obtained from Malware Genome Project and Android Malware Dataset by Arguslab) and on non-malware samples (obtained from Google Play Store). We have found that our model is quite lightweight yet powerful to detect malwares with sianificant accuracy of 98%.

Supervised learning is an important component of several applications, such as speech recognition, handwritten symbol recognition, data mining, among others. Supervised classification algorithms aim at producing a learning model from a labeled training set. Different methods and approaches have been proposed to overcome the two-class classification problem. Among the existing techniques in literature, the classifier based on Straight Line Segments (SLS Classifier) is worthy of note. This technique is based on distances between points and two sets of straight line segments, whose initial positions are obtained by applying the K-Means algorithm. Then, the gradient descent method finds its optimal positions that minimize the Mean Squared Error. This paper aims to study the impact of the initial positions on the classifier accuracy. For this purpose, we performed two experiments to demonstrate the stability of the classifier performance when the initial positions are not optimal (close to the samples): (i) random initial positions and; (ii) k-means positions displaced by adding Gaussian and uniform noises. In addition, we perform a comparison with positions obtained using different clustering algorithms. As expected, the results suggest that with an increased noise level, the classification rate decreases, however, such reduction was not significant as compared when using the random initial positions. It is worth mentioning that in most of the experiments, the classification rate of the SLS and the Bayes classifier are comparable.

GPS and GLONASS are used worldwide to locate the devices using satellites but it cannot locate objects under the roof. Therefore different sensors are required to be deployed inside for indoor localization of the devices. Different techniques have been developed including angle of arrival technique, triangulation, trilateration, Artificial Neural Networks, KNN Classification techniques and Bayesian classification techniques. One of the most popular technique known as Naive Bayes Technique is mostly used for indoor localization of the objects and devices. Nave Bayes classifier assumes conditional independence between the attributes but in real world this is not the case. In order to overcome dependence and zero probability issue of Naive Bayes algorithm, different variants of Naive Bayes technique have been developed. In this paper we have done a comparative study of different Naive Bayes theorem based classification techniques and some other classification techniques for location estimation of device in indoor environment are done. The accuracy and efficiency of different techniques including SVM, SMO, Random Forest, Random Trees, Augmented Naive Bayes, Hidden Naive Bayes, Fine Grained Naive Bayes and Multinomial Naive Bayes technique are compared to find the best location estimation algorithm.

Various Convolutional Neural Network (CNN) architecture has been proposed for image classification and Object recognition. For the image based classification, it is a complex task for CNN to deal with hundreds of MRI Image slices, each of almost identical nature in a single patient. So, classifying a number of patients as an AD, MCI or NC based on 3D MRI becomes vague technique using 2D CNN architecture. Hence, to address this issue, we have simplified the idea of classifying patients on basis of 3D MRI but acknowledging the 2D features generated from the CNN framework. We present our idea regarding how to obtain 2D features from MRI and transform it to be applicable to classify using machine learning algorithm. Our experiment shows the result of classifying 3 class subjects patients. We employed scratched trained CNN or pretrained Alexnet CNN as generic feature extractor of 2D image which dimensions were reduced using PCA+TSNE, and finally classifying using simple Machine learning algorithm like KNN, Navies Bayes Classifier. Although the result is not so impressive but it definitely shows that this can be better than scratch trained CNN softmax classification based on probability score. The generated feature can be well manipulated and refined for better accuracy, sensitivity, and specificity.

Conventional approaches to acoustic source localization simply based on the received microphone signals, are often vulnerable to adverse acoustic conditions, such as low signal-to-noise ratio (SNR) or high reverberation. But, approaches based on Pattern Recognition and Machine Learning Technology can increase accuracy to locate source in adverse acoustic environment. The advantage of the algorithm is that it requires no calibration of microphone arrays. And Naive Bayes Classifier is simple, fast, and has a small error rate. This paper proposed an improved localization algorithm based on classification of cross-correlation functions (GCC). The weighted cross power spectrum of GCC is smoothed by a smooth filter to formed smooth generalized cross-correlation (SGCC). Then, the classifier model is obtained in each location and form the feature vector. Finally, acoustic source location is estimated by Naive- Bayes classifier. We also proposed in this study the source localization system that based on merely two microphones to input sound signals, combined with improved and optimal methods proposed above. Real-data experiments have demonstrated that algorithm with SGCC has higher localization accuracy than with GCC by 20% in the proposed system at least. The system has good ability to acoustic source localization.

In multiclass text classification, the performance of classifier is usually very low while the data distribution is very uneven. The reason may be that the majority class has a great influence on classifier, and the minority class ignored. This study attempts to solve this problem by improving CHI-square feature selection algorithm and by improving the Term Frequency–Inverse Document Frequency (TF-IDF) weight algorithm and combining with the K-means clustering method, generating minority class samples to reduce the influence of inclined class distribution. The experimental results demonstrate that the improved method has a significant improvement on the classification efficiency on the bias dataset.

This paper introduces a method of abnormal human activity recognition in surveillance video. The method uses Bayes Classifier and Convolutional Neural Network to detect four activities, including walking, running, punching and tripping. KTH dataset is used as the input of Bayes Classifier and Convolutional Neural Network. Moving human target in each frame is detected by Kalman Filter and three features of the target image are extracted. The features include length-width ratio, entropy, and Hu invariant moment. Meanwhile, convolutional neural network of abnormal human activity recognition is built and trained. Experiments show that recognition accuracy of Bayes Classifier reaches 88%, 92%, 92% and 100% for each activity, and Convolutional Neural Network reaches 92%, 96%, 100% and 100% for each activity.

This paper presents a new method to improve accuracy of supervised change detection in Synthetic Aperture Radar (SAR) imagery. The model is based on the idea to apply a low size labeled dataset to the input of an Ensemble of Self-Organizing Maps (ESOM) for training data generation (TDG). The resulted synthetic data set produced by ESOM substitutes the initial authentically labeled sample set and it is used to train a supervised change detection classifier. The proposed method is evaluated using a TerraSAR-X image of 400×400 pixels acquired in the Fukushima region, Japan, before and after tsunami. As change detection classifiers we have comparatively considered Support Vector Machine (SVM), Nearest Neighbor (NN), the three-Nearest Neighbors (3-NN), and Likelihood Bayes classifier. The experimental results have confirmed the effectiveness of the proposed approach using only 100 authentic labeled pixels.

Meta learning of optimal classifier error rates allows an experimenter to empirically estimate the intrinsic ability of any estimator to discriminate between two populations, circumventing the difficult problem of estimating the optimal Bayes classifier. To this end we propose a weighted nearest neighbor (WNN) graph estimator for a tight bound on the Bayes classification error; the Henze-Penrose (HP) divergence. Similar to recently proposed HP estimators [1], the proposed estimator is non-parametric and does not require density estimation. However, unlike previous approaches the proposed estimator is rate-optimal, i.e., its mean squared estimation error (MSEE) decays to zero at the fastest possible rate of $O(1/M+1/N)$ where M, N are the sample sizes of the respective populations. We illustrate the proposed WNN meta estimator for several simulated and real data sets.

A novel, fully automated segmentation framework that can extract cerebral blood vessels larger than 1 mm diameter from time-offlight magnetic resonance angiography (TOF-MRA) scans was developed. The segmentation framework consists of the following main steps: 1) Following bias correction, the TOF-MRA image is modeled as a a three-dimensional (3-D) generalized Gauss-Markov random field (GGMRF) where pairwise interactions between the 26-neighborhood voxels enhance the homogeneity of brain and vascular tissues while preserving the 3-D edges between them. 2) To automatically extract the region of interest, a novel approach was developed to remove the skull from TOF-MRA images based on the integration of the Markov-Gibbs random field (MGRF) model of TOF-MRA visual appearance with a geometric deformable model (brain isosurface) that preserves brain topology during the extraction process. 3) The initial vascular system is extracted using Bayes classifier following the estimation of the marginal probability density of TOF-MRA voxel values for both blood vessels and other brain tissues, modeled as a linear combination of discrete Gaussians (LCDG). 4) To enhance the extraction of small vessels, a seed-generation refinement algorithm was used to identify points within regions likely to contain small vessels overlooked by the previous step. 5) Finally, a 3-D region-growing connected components algorithm was applied to extract the final, connected vascular tree. The proposed approach was tested on in-vivo data (200 TOF-MRA data sets), and qualitatively and quantitatively validated by an MRA expert (ground truth). The proposed approach resulted in high segmentation accuracy with 86% Dice similarity coefficient, and 10% absolute volume difference.

Learning from fast data stream is one of the most challenging tasks in data stream mining. The fact that data streams are unbounded sequences, highlights exclusive challenges in contrast to classifiers from batch data. Most of methods aren't naturally parallel and thus their scalability is limited. This paper proposes a scalable data stream Bayes classifier utilizing a new estimation(DIB). The new estimation takes conjugate Dirichlet prior as parameter's prior distribution and thus improves the predictive accuracy. Meanwhile, this paper proposes a new distributed implementation of DIB on Flink. Experiments show that DIB classifier significantly outperforms Naive Bayes in terms of accuracy. Also, the experiment proves parallel DIB running on Flink enhances the throughput and reduces execution time.

Several studies indicate that roughly 70% of the projects based on the software development have resulted in failure, thereby researchers and practitioners have been tried to develop solutions that will improve project success rates. It is insisted that to raise success rates, support should be provided by the organization to which the projects belong. With the aid of predictions that incorporate project outcomes for various information technology (IT) vendors, this study aims at identifying projects that should be preferentially supported by an organization. The data of 332 projects of various Japanese IT vendors were collected using an Internet survey, and a success/failure prediction algorithm is created by employing the Bayes classifier technique on the collected data. A resultant algorithm with 77.3% prediction capability was obtained. It is expected that the success/failure prediction procedure, including the prediction algorithm, help significantly to specify projects that an organization needs to participate in as priority.

This work proposes a method that allows the entry of text using the Google Cardboard, for this, uses gestures based on geometric forms and smartwatches. Thus, it is proposed the development of a prototype with the purpose of inserting a letter with no more than two user interactions. Gesture recognition is performed using the incremental recognition algorithm. After recognizing the gestures, they are sent to a classifier Naïve Bayes which is responsible for predicting the letter that will be inserted. The Naïve Bayes classifier was trained with a user gesture base that drew all the letters of the alphabet using only the gestures available in the set presented to them. Using the gesture base and the Naïve Bayes classifier a prototype for smartwatches was developed that works in conjunction with the Google Cardboard. The user performs the interactions on smartwatch and views them on Google Cardboard. This prototype automatically suggests the most likely letters to be entered and allows the user to enter text without having to look at smartwatch thus allowing the user to continue performing his other tasks.

Data classification is an important task to label the class of data. Attributes or feature is a portion of information which is applicable to the task of computation. Our task is to predict and prevent cardiac arrest which is one of the biggest challenges of cardiology using a machine learning classifier. Since a particular classifier may or may not work well for such datasets so it is important to do a comparative study of classifiers in order to achieve maximum performance in such critical predictions of cardiac arrest. The UCI dataset is chosen for the purpose of comparison, and a comparative study of various classifiers is provided on the same dataset. Results are given as accuracy of different classifiers. The various classifier methods include the KNN classifier, Nave Bayes classifier, Support Vector Machine, Neural Network, Gaussian Mixture Model and Decision Tree classifier.