Bayesian Classification Technique Research Articles

Étude comparative d'algorithmes d'apprentissage artificiel pour la reconnaissance faciale

Background: The fundamental need for authentication and identification of humans using their physiological, behavioral or biological characteristics, continues to be applied extensively to secure localities, property, financial transactions, etc. Biometric systems based on face characteristics, continue to attract the attention of researchers, major public and private services. In the literature, many methods have been deployed by different authors. The best performance must be found in order to be able to recommend the most effective method. So, the main objective of thisarticle is to make a comparative study of different existing techniques.Methods: A biometric system is generally composed of four stages: acquisition of facial images, preprocessing, extraction of characteristics and finally classification. In this work, the focus is on machine learning algorithms for classification. These algorithms are: Support Vector Machines (SVM), Artificial Neural Networks (ANN), K-Nearest Neighbors (KNN), Random Forests (RF), Logistic Regression (LR), Naive Bayesian Classification (NB: Naive Bayes’ Classifiers) and deep learning techniques such as Convolutional Neural Networks (CNN). The comparison criterion is the average performance, calculated using three performance measures: recognition rate, confusion matrix, and the Area Under Receiver Operating Characteristic (ROC) curve.Results: Based on this criterion, the performance comparison of selected machine learning algorithms, shows that CNN is the best, with an average performance of 100.00% On ORL face database. However, on the YALE database, classical algorithms such as artificial neural networks have obtained the best performances, the highest being a rate of 100%.Discussion: Deep learning techniques are very efficient in image classification as proven by the results on the ORL database. However, their inefficiency on YALE face database is due to the small size of this database which is inappropriate for some deep learning algorithms. But this weakness can be corrected by image augmentation techniques. The comparison of these results with existing state-of-the-art methods is nearly the same. Authors achieved performances of 94.82%, 95.79%, 96.15%, 96.44%, 97.27%, 98.52% and 98.95% for NB, KNN, RF, LR, ANN, SVM and CNN classifiers, respectively. Finally, in depth discussion, it is concluded that between all these approaches which are useful in face recognition, the CNN is the best classification algorithm.

CHOLERA PREDICTION MODEL USING FEATURE CLUSTERING BAYESIAN TECHNIQUE

Cholera is one the most deadly disease that is mostly caused due by poor sanitation or and drinking contaminated water or food with a bacterium called Vibrio Cholera. Many researchers have provided a solution to prevent the outbreak of cholera using various methods, the recent ones are using machine learning techniques and some mathematical methods such as mathematical epidemiological, spatial statistics, and based on association rule mining using the nonstandard distribution dataset to mention a few. These few methods are mostly used in predicting cholera outbreaks but have some limitations, such as using fewer features for prediction, waiting until certain cases are reported before getting data, based on Rainfall, based on the flowing speed of rivers, wind direction, and flood, etc. in this research a more comprehensive cholera features would be used in predicting an outbreak before it occurs based on the existing secondary dataset using The Naïve Bayesian Classification technique. The proposed model has more features and is not dependent on certain events to occur before predicting any outbreak. Python programming was used in implementing the proposed model. An accuracy of 99% was achieved and it shows it is better than the previous models used in predicting cholera outbreaks.

Practical workflows for monitoring saturation and pressure changes from 4D seismic data: A case study of Malay Basin

Discrimination between fluid saturation and pore pressure-related changes using 4D seismic response is a challenging issue in reservoir studies. The main objectives of this study are the estimation of pore pressure and water saturation for baseline and monitor seismic surveys as well as determination of bypassed hydrocarbons location. To achieve this goal, we propose two practical workflows. The workflows originate from the fact that changes in pressure and saturation result in different trends on a Vp/Vs ratio versus P-Impedance cross-plot. The fact that an increase in hydrocarbon saturation results in a decrease in the Vp/Vs ratio and that an increase in pore pressure causes an increase in the Vp/Vs ratio plays a key role in this study. In this approach, after rock physics modeling and seismic data conditioning, elastic seismic inversion is performed for baseline and monitor surveys. The Bayesian probability classification technique is used to differentiate pay zones from non-pay zones which is then incorporated for prediction of fluid saturation changes. In the pore pressure prediction workflow, the Bowers equation is replaced with an empirical relationship in which a consistent match is attained between the calculated effective pressure and Vp/Vs ratio at injected water zones. Pore pressure prediction reveals noticeable anomalies around the injector wells but pressure depletion around the producers remains uncertain due to the insignificant pore pressure drop and the poor quality of seismic data near the platform area. However, at most of the producers' locations, the predicted pore pressure shows a depletion trend confirmed by the field-measured pressure. In this case study, the unswept oil and injected water areas are clearly outlined, which helps to improve sweep efficiency and update EOR scenarios.

Water Level Prediction In Water Shed Management Utilizing Machine Learning

Due to uneven rainfall, nowadays the amount of rain to be showered in a month is getting showered in few days. The massive wastage of water occurs due to irregular heavy rainfall and water released from dams. To avoid this, the proposal suggests an idea to develop a watershed and to predict the water level measurement by Bayesian classification, clustering, and optimization techniques. Artificial Neural Network is one of the previous techniques used to predict water level which gives approximate result only. To overcome the disadvantage, this proposal suggests an idea to develop the watershed by using different machine learning techniques. The level of water that can be stored is calculated using Bayes Network which will classify the data into labels according to the condition of the capacity of the minimum and maximum storage level of the watershed. The standardized data considered for the classification are normalized using the z-score normalization. Classification will represent the result by means of the instances that are correctly classified. The output of the classified data is fed into clustering algorithm where the labels are grouped into different clusters. The K-Mean algorithm is utilized for clustering which iteratively assign data point to one of the k group according to the given attribute. The clustered output gives the result of how many instances are correctly clustered. The clustered output will be refined for further process such that the data will be extracted as ordered dataset of year wise and month wise data. For the extracted data gradient descent algorithm is applied for reducing the error and predicting the amount of water stored in watershed for upcoming years by means of calculating the actual and prediction value. Later the result will be visualized in the form of graph. The obtained output is considered as an input for posterior probability that uses J48 algorithm which gives the result of probability of event happened after all the evidence is taken for consideration and gives the accurate result. The above methodology provides high performance and efficient result.

Water Level Prediction In Water Shed Management Utilizing Machine Learning

Due to uneven rainfall, nowadays the amount of rain to be showered in a month is getting showered in few days. The massive wastage of water occurs due to irregular heavy rainfall and water released from dams. To avoid this, the proposal suggests an idea to develop a watershed and to predict the water level measurement by Bayesian classification, clustering, and optimization techniques. Artificial Neural Network is one of the previous techniques used to predict water level which gives approximate result only. To overcome the disadvantage, this proposal suggests an idea to develop the watershed by using different machine learning techniques. The level of water that can be stored is calculated using Bayes Network which will classify the data into labels according to the condition of the capacity of the minimum and maximum storage level of the watershed. The standardized data considered for the classification are normalized using the z-score normalization. Classification will represent the result by means of the instances that are correctly classified. The output of the classified data is fed into clustering algorithm where the labels are grouped into different clusters. The K-Mean algorithm is utilized for clustering which iteratively assign data point to one of the k group according to the given attribute. The clustered output gives the result of how many instances are correctly clustered. The clustered output will be refined for further process such that the data will be extracted as ordered dataset of year wise and month wise data. For the extracted data gradient descent algorithm is applied for reducing the error and predicting the amount of water stored in watershed for upcoming years by means of calculating the actual and prediction value. Later the result will be visualized in the form of graph. The obtained output is considered as an input for posterior probability that uses J48 algorithm which gives the result of probability of event happened after all the evidence is taken for consideration and gives the accurate result. The above methodology provides high performance and efficient result.

Malware Classification using Machine Learning Algorithms and Tools

The explosive growth of malware variants poses a major threat to information security. Malware is the one which frequently growing day by day and becomes major threats to the Internet Security. According to numerous increasing of worm malware in the networks nowadays, it became a serious danger that threatens our computers. Networks attackers did these attacks by designing the worms. A designed system model is needed to defy these threats, prevent it from multiplying and spreading through the network, and harm our computers. In this paper, we designed a classification on system model for this issue. The designed system detects the worm malware that depends on the information of the dataset that is taken from website, the system will receive the input package and then analyze it, the Naïve Bayesian classification technique will start to work and begin to classify the package, by using the data mining Naïve Bayesian classification technique, the system worked fast and gained great results in detecting the worm. By applying the Naïve Bayesian classification technique using its probability mathematical equations for both threat data and benign data, the technique will detect the malware and classify data whether it was threat or benign.

Machine Learning Technique for Crop Recommendation in Agriculture Sector

Numerous efforts have been demonstrated through various innovations to lead towards the betterment of agriculture sector till now. Effective and innovative use of Science and Technology can help to improve crop quality and production, yield prediction and crop disease analysis. Agriculture sector provides various productions such as food, raw material for industry and has a significant impact on economy and employment of a country. The agriculture sector contains huge data with respect to factors affecting its input and output. With advances in technology, various data mining techniques have been introduced accordingly. Also different analytical models like Decision Tree, Random Forest, Support Vector Machine and Bayesian Neural Network are available for required analysis. These data mining techniques can be used to analyze the multidimensional, time specific data of agriculture sector to produce effective knowledge in support of the efforts to boost the economy. These methods can be further used to analyze soil, climate, moisture, humidity, temperature. In this paper, we used Naive Bayesian Classification technique to identify the class of crop and used the food grain dataset to analyze the technique over different attribute. Further we used Machine Learning approach for the accurate crop recommendation. Crop recommendation will help for effective decision making to the end user to take proper decision according to the output of the system.

Automatic Folder Allocation System for Electronic Text Document Repositories Using Enhanced Bayesian Classification Approach

This article proposes a system equipped with the enhanced Bayesian classification techniques to automatically assign folders to store electronic text documents. Despite computer technology advancements in the information age where electronic text files are so pervasive in information exchange, almost every single document created or downloaded from the Internet requires manual classification by the users before being deposited into a folder in a computer. Not only does such a tedious task cause inconvenience to users, the time taken to repeatedly classify and allocate a folder for each text document impedes productivity, especially when dealing with a huge number of files and deep layers of folders. In order to overcome this, a prototype system is built to evaluate the performance of the enhanced Bayesian text classifier for automatic folder allocation, by categorizing text documents based on the existing types of text documents and folders present in user's hard drive. In this article, the authors deploy a High Relevance Keyword Extraction (HRKE) technique and an Automatic Computed Document Dependent (ACDD) Weighting Factor technique to a Bayesian classifier in order to obtain better classification accuracy, while maintaining the low training cost and simple classifying processes using the conventional Bayesian approach.

NBC-MAIDS: Naïve Bayesian classification technique in multi-agent system-enriched IDS for securing IoT against DDoS attacks

Internet of Things (IoT) makes physical objects and devices interact with each other through wireless technologies. IoT is expected to deliver a significant role in our lives in near future. However, at the current stage, IoT is vulnerable to various kinds of security threats just like other wired and wireless networks. Our work mainly focuses on protecting an IoT infrastructure from distributed denial-of-service attacks generated by the intruders. We present a new approach of using Naive Bayes classification algorithm applied in intrusion detection systems (IDSs). IDSs are deployed in the form of multi-agents throughout the network to sense the misbehaving or irregular traffic and actions of nodes. In the paper, we also discuss the fundamental concepts related to our work and recent research done in similar area.

Inferring the most probable maps of underground utilities using Bayesian mapping model

Mapping the Underworld (MTU), a major initiative in the UK, is focused on addressing social, environmental and economic consequences raised from the inability to locate buried underground utilities (such as pipes and cables) by developing a multi-sensor mobile device. The aim of MTU device is to locate different types of buried assets in real time with the use of automated data processing techniques and statutory records. The statutory records, even though typically being inaccurate and incomplete, provide useful prior information on what is buried under the ground and where. However, the integration of information from multiple sensors (raw data) with these qualitative maps and their visualization is challenging and requires the implementation of robust machine learning/data fusion approaches. An approach for automated creation of revised maps was developed as a Bayesian Mapping model in this paper by integrating the knowledge extracted from sensors raw data and available statutory records. The combination of statutory records with the hypotheses from sensors was for initial estimation of what might be found underground and roughly where. The maps were (re)constructed using automated image segmentation techniques for hypotheses extraction and Bayesian classification techniques for segment-manhole connections. The model consisting of image segmentation algorithm and various Bayesian classification techniques (segment recognition and expectation maximization (EM) algorithm) provided robust performance on various simulated as well as real sites in terms of predicting linear/non-linear segments and constructing refined 2D/3D maps.

SLMBC: spiral life cycle model-based Bayesian classification technique for efficient software fault prediction and classification

Software fault prediction and classification plays a vital role in the software development process for assuring high quality and reliability of the software product. Earlier prediction of the fault-prone software modules enables timely correction of the faults and delivery of reliable product. Generally, the fuzzy logic, decision tree and neural networks are deployed for fault prediction. But these techniques suffer due to low accuracy and inconsistency. To overcome these issues, this paper proposes a spiral life cycle model-based Bayesian classification technique for efficient software fault prediction and classification. In this process, initially the dependent and independent software modules are identified. The spiral life cycle model is used for testing the software modules in each life cycle of the software development process. Bayesian classification is applied to classify the software modules as faulty module and non-faulty module, by using the probability distribution models. Robust similarity-aware clustering algorithm performs clustering of the faulty and non-faulty software modules based on the similarity measure of the features in the dataset. From the experimental results, it is observed that the proposed method enables accurate prediction and classification of the faulty modules. The proposed technique achieves higher accuracy, precision, recall, probability of detection, F-measure and lower error rate than the existing techniques. The misclassification rate of the proposed technique is found to be lower than the existing techniques. Hence, the reliability of the software development process can be improved.

Oculomotor inhibition covaries with conscious detection.

Saccadic eye movements occur frequently even during attempted fixation, but they halt momentarily when a new stimulus appears. Here, we demonstrate that this rapid, involuntary "oculomotor freezing" reflex is yoked to fluctuations in explicit visual perception. Human observers reported the presence or absence of a brief visual stimulus while we recorded microsaccades, small spontaneous eye movements. We found that microsaccades were reflexively inhibited if and only if the observer reported seeing the stimulus, even when none was present. By applying a novel Bayesian classification technique to patterns of microsaccades on individual trials, we were able to decode the reported state of perception more accurately than the state of the stimulus (present vs. absent). Moreover, explicit perceptual sensitivity and the oculomotor reflex were both susceptible to orientation-specific adaptation. The adaptation effects suggest that the freezing reflex is mediated by signals processed in the visual cortex before reaching oculomotor control centers rather than relying on a direct subcortical route, as some previous research has suggested. We conclude that the reflexive inhibition of microsaccades immediately and inadvertently reveals when the observer becomes aware of a change in the environment. By providing an objective measure of conscious perceptual detection that does not require explicit reports, this finding opens doors to clinical applications and further investigations of perceptual awareness.

Service composition with quality of service management in environmental sensor networks

Environmental sensor networks consist of geographically distributed resource-constrained computers that use sensors to measure environmental parameters. They typically collect and distribute environmental data in near real-time using different types of web services. These services are consumed by users and applications that expect appropriate quality of service QoS levels. However, acceptable and stable QoS levels are difficult to maintain in environmental sensor networks due to limited computing resources and hostile deployment conditions. We developed a new approach to efficiently manage QoS levels in these networks using two separate tasks: online QoS estimation and QoS-aware service composition. This paper presents our first prototype of a web service composition system that leverages Bayesian classification techniques to perform online QoS estimation by constantly monitoring key environmental and network performance parameters. This system has been tested in a live noise pollution environmental sensor network deployed in the Flanders region in Belgium.

Sentiment classification of Roman-Urdu opinions using Naïve Bayesian, Decision Tree and KNN classification techniques

Sentiment mining is a field of text mining to determine the attitude of people about a particular product, topic, politician in newsgroup posts, review sites, comments on facebook posts twitter, etc. There are many issues involved in opinion mining. One important issue is that opinions could be in different languages (English, Urdu, Arabic, etc.). To tackle each language according to its orientation is a challenging task. Most of the research work in sentiment mining has been done in English language. Currently, limited research is being carried out on sentiment classification of other languages like Arabic, Italian, Urdu and Hindi. In this paper, three classification models are used for text classification using Waikato Environment for Knowledge Analysis (WEKA). Opinions written in Roman-Urdu and English are extracted from a blog. These extracted opinions are documented in text files to prepare a training dataset containing 150 positive and 150 negative opinions, as labeled examples. Testing data set is supplied to three different models and the results in each case are analyzed. The results show that Naïve Bayesian outperformed Decision Tree and KNN in terms of more accuracy, precision, recall and F-measure.

Landslide hazards mapping using uncertain Naïve Bayesian classification method

Landslide hazard mapping is a fundamental tool for disaster management activities in Loess terrains. Aiming at major issues with these landslide hazard assessment methods based on Naive Bayesian classification technique, which is difficult in quantifying those uncertain triggering factors, the main purpose of this work is to evaluate the predictive power of landslide spatial models based on uncertain Naive Bayesian classification method in Baota district of Yan’an city in Shaanxi province, China. Firstly, thematic maps representing various factors that are related to landslide activity were generated. Secondly, by using field data and GIS techniques, a landslide hazard map was performed. To improve the accuracy of the resulting landslide hazard map, the strategies were designed, which quantified the uncertain triggering factor to design landslide spatial models based on uncertain Naive Bayesian classification method named NBU algorithm. The accuracies of the area under relative operating characteristics curves (AUC) in NBU and Naive Bayesian algorithm are 87.29% and 82.47% respectively. Thus, NBU algorithm can be used efficiently for landslide hazard analysis and might be widely used for the prediction of various spatial events based on uncertain classification technique.

Enhancing Proxy Server cache Management using Log Analysis and Recommendations

Now a day’s web based applications are growing rapidly due to this the network performance is affected significantly. Thus a performance improvement technique is required by which the application speed is maintained and delivers the high performance web pages. Thus pre-fetching techniques are applied. There are various kinds of pre-fetching techniques are available among them a promising data model is find in [1]. In this technique the proxy log data is consumed for performing navigation pattern analysis. Thus this model incorporates a K-mean algorithm for cluster log data and then the Apriori algorithm is applied to find the frequent pattern rules. Using these rules the system recommends the possible user pages for prefetching. In order to enhance the performance of the traditional model two different techniques are implemented and compared with the traditional model. First utilize the Bayesian classification technique for analyzing pattern and in second method the ID3 decision tree algorithm for analyzing patterns. The comparison of the both the techniques are performed in terms of memory used, time consumption, accuracy and error rate. According to the obtained results the proposed predictive system offers high performance results as compared to the traditional data model.

Absorption, Distribution, Metabolism, Excretion, and Toxicity Evaluation in Drug Discovery. 14. Prediction of Human Pregnane X Receptor Activators by Using Naive Bayesian Classification Technique.

The activation of pregnane X receptor (PXR), a member of the nuclear receptor (NR) superfamily, can mediate potential drug-drug interactions, and therefore, prediction of PXR activation is of great importance for evaluating drug metabolism and toxicity. In this study, based on 532 structurally diverse compounds, we present a comprehensive analysis with the aim to build accurate classification models for distinguishing PXR activators from nonactivators by using a naive Bayesian classification technique. First, the distributions of eight important molecular physicochemical properties of PXR activators versus nonactivators were compared, illustrating that the hydrophobicity-related molecular descriptors (AlogP and log D) show slightly better capability to discriminate PXR activators from nonactivators than the others. Then, based on molecular physicochemical properties, VolSurf descriptors, and molecular fingerprints, naive Bayesian classifiers were developed to separate PXR activators from nonactivators. The results demonstrate that the introduction of molecular fingerprints is quite essential to enhance the prediction accuracy of the classifiers. The best Bayesian classifier based on the 21 physicochemical properties, VolSurf descriptors, and LCFC_10 fingerprints descriptors yields a prediction accuracy of 92.7% for the training set based on leave-one-out (LOO) cross-validation and of 85.2% for the test set. Moreover, by exploring the important structural fragments derived from the best Bayesian classifier, we observed that flexibility is an important structural pattern for PXR activation. In addition, chemical compounds containing more halogen atoms, unsaturated alkanes chains relevant to π-π stacking, and fewer nitrogen atoms tend to be PXR activators. We believe that the naive Bayesian classifier can be used as a reliable virtual screening tool to predict PXR activation in the drug design and discovery pipeline.

Adaptive Bayesian Diagnosis of Intermittent Faults

With increasing transient error rates, distinguishing intermittent and transient faults is especially challenging. In addition to particle strikes relatively high transient error rates are observed in architectures for opportunistic computing and in technologies under high variations. This paper presents a method to classify faults into permanent, intermittent and transient faults based on some intermediate signatures during embedded test or built-in self-test. Permanent faults are easily determined by repeating test sessions. Intermittent and transient faults can be identified by the amount of failing test sessions in many cases. For the remaining faults, a Bayesian classification technique has been developed which is applicable to large digital circuits. The combination of these methods is able to identify intermittent faults with a probability of more than 98 %.

Assessing an ensemble docking-based virtual screening strategy for kinase targets by considering protein flexibility.

In this study, to accommodate receptor flexibility, based on multiple receptor conformations, a novel ensemble docking protocol was developed by using the naïve Bayesian classification technique, and it was evaluated in terms of the prediction accuracy of docking-based virtual screening (VS) of three important targets in the kinase family: ALK, CDK2, and VEGFR2. First, for each target, the representative crystal structures were selected by structural clustering, and the capability of molecular docking based on each representative structure to discriminate inhibitors from non-inhibitors was examined. Then, for each target, 50 ns molecular dynamics (MD) simulations were carried out to generate an ensemble of the conformations, and multiple representative structures/snapshots were extracted from each MD trajectory by structural clustering. On average, the representative crystal structures outperform the representative structures extracted from MD simulations in terms of the capabilities to separate inhibitors from non-inhibitors. Finally, by using the naïve Bayesian classification technique, an integrated VS strategy was developed to combine the prediction results of molecular docking based on different representative conformations chosen from crystal structures and MD trajectories. It was encouraging to observe that the integrated VS strategy yields better performance than the docking-based VS based on any single rigid conformation. This novel protocol may provide an improvement over existing strategies to search for more diverse and promising active compounds for a target of interest.

Modeling of tomato fruits into nine shape categories using elliptic fourier shape modeling and Bayesian classification of contour morphometric data

Classification and characterization of the shape of plant organs are important tools for plant biologists, breeders and growers. Here we use boundary measurements, i.e. contour morphometric data, of scanned tomato fruits in conjunction with elliptic Fourier shape modeling and Bayesian classification techniques to find the optimum number of shape categories. Our findings show that there are nine computationally and visually distinct tomato shape categories: ellipsoid, flat, heart, long, long rectangular, rectangular, round, obovoid, and oxheart. Analyses of fruits from a diverse set of tomato accessions demonstrate that some varieties carry fruits that conform to predominantly one shape category while others carry fruits that conform to multiple shape categories. In particular the categories oxheart and long rectangular feature fruit that tend to equivalently fit several categories of shape, while the flat and obovoid categories contain fruit that consistently conform exclusively to a single category. The findings show that elliptic Fourier shape modeling and Bayesian classification provide an excellent tool for further in depth analyses of fruit shape variation that may occur across varieties and/or result from growth under different environmental conditions.