Using Cross Entropy as a Performance Metric for Quantifying Uncertainty in DNN Image Classifiers: An Application to Classification of Lung Cancer on CT Images

Cross entropy is a measure in machine learning and deep learning that assesses the difference between predicted and actual probability distributions. In this study, we propose cross entropy as a performance evaluation metric for image classifier models and apply it to the CT image classification of lung cancer. A convolutional neural network is employed as the deep neural network (DNN) image classifier, with the residual network (ResNet) 50 chosen as the DNN archi-tecture. The image data used comprise a lung CT image set. Two classification models are built from datasets with varying amounts of data, and lung cancer is categorized into four classes using 10-fold cross-validation. Furthermore, we employ t-distributed stochastic neighbor embedding to visually explain the data distribution after classification. Experimental results demonstrate that cross en-tropy is a highly useful metric for evaluating the reliability of image classifier models. It is noted that for a more comprehensive evaluation of model perfor-mance, combining with other evaluation metrics is considered essential. .

Real and Altered Fingerprint Classification Based on Various Features and Classifiers

Biometric recognition refers to the identification of individuals through their unique behavioral features(e.g.,fingerprint,face,and iris).We need distinguishing characteristics to identify people,such as fingerprints,which are world-renowned as the most reliablemethod to identify people.The recognition of fingerprints has become a standard procedure in forensics,and different techniques are available for this purpose.Most current techniques lack interest in image enhancement and rely on high-dimensional features to generate classification models.Therefore,we proposed an effective fingerprint classification method for classifying the fingerprint image as authentic or altered since criminals and hackers routinely change their fingerprints to generate fake ones.In order to improve fingerprint classification accuracy,our proposed method used the most effective texture features and classifiers.Discriminant Analysis(DCA)and Gaussian Discriminant Analysis(GDA)are employed as classifiers,along with Histogram of Oriented Gradient(HOG)and Segmentation-based Feature Texture Analysis(SFTA)feature vectors as inputs.The performance of the classifiers is determined by assessing a range of feature sets,and the most accurate results are obtained.The proposed method is tested using a Sokoto Coventry Fingerprint Dataset(SOCOFing).The SOCOFing project includes 6,000 fingerprint images collected from 600 African people whose fingerprints were taken ten times.Three distinct degrees of obliteration,central rotation,and z-cut have been performed to obtain synthetically altered replicas of the genuine fingerprints.The proposal achieved massive success with a classification accuracy reaching 99%.The experimental results indicate that the proposed method for fingerprint classification is feasible and effective.The experiments also showed that the proposed SFTA-based GDA method outperformed state-of-art approaches in feature dimension and classification accuracy.

Clustering-Aided Supervised Malware Detection with Specialized Classifiers and Early Consensus

One of the most common types of threats to the digital world is malicious software.It is of great importance to detect and prevent existing and new malware before it damages information assets.Machine learning approaches are used effectively for this purpose.In this study,we present a model in which supervised and unsupervised learning algorithms are used together.Clustering is used to enhance the prediction performance of the supervised classifiers.The aim of the proposed model is to make predictions in the shortest possible time with high accuracy and f1 score.In the first stage of the model,the data are clustered with the k-means algorithm.In the second stage,the prediction is made with the combination of the classifier with the best prediction performance for the related cluster.While choosing the best classifiers for the given clusters,triple combinations of ten machine learning algorithms(kernel support vector machine,k-nearest neighbor,naive Bayes,decision tree,random forest,extra gradient boosting,categorical boosting,adaptive boosting,extra trees,and gradient boosting)are used.The selected triple classifier combination is positioned in two stages.The prediction time of the model is improved by positioning the classifier with the slowest prediction time in the second stage.The selected triple classifier combination is positioned in two tiers.The prediction time of the model is improved by positioning the classifier with the highest prediction time in the second tier.It is seen that clustering before classification improves prediction performance,which is presented using Blue Hexagon Open Dataset for Malware Analysis(BODMAS),Elastic Malware Benchmark for Empowering Researchers(EMBER)2018 and Kaggle malware detection datasets.The model has 99.74%accuracy and 99.77%f1 score for the BODMAS dataset,99.04%accuracy and 98.63%f1 score for the Kaggle malware detection dataset,and 96.77%accuracy and 96.77%f1 score for the EMBER 2018 dataset.In addition,the tiered positioning of classifiers shortened the average prediction time by 76.13%for the BODMAS dataset and 95.95%for the EMBER 2018 dataset.The proposed method’s prediction performance is better than the rest of the studies in the literature in which BODMAS and EMBER 2018 datasets are used.

Predicting Stock Prices Using Polynomial Classifiers: The Case of Dubai Financial Market

Predicting stock price movements is a challenging task for academicians and practitioners. In particular, forecasting price movements in emerging markets seems to be more elusive because they are usually more volatile often accompa-nied by thin trading-volumes and they are susceptible to more manipulation compared to mature markets. Technical analysis of stocks and commodities has become a science on its own;quantitative methods and techniques have been applied by many practitioners to forecast price movements. Lagging and sometimes leading technical indicators pro-vide rich quantitative tools for traders and investors in their attempt to gain advantage when making investment or trading decisions. Artificial Neural Networks (ANN) have been used widely in predicting stock prices because of their capability in capturing the non-linearity that often exists in price movements. Recently, Polynomial Classifiers (PC) have been applied to various recognition and classification application and showed favorable results in terms of recog-nition rates and computational complexity as compared to ANN. In this paper, we present two prediction models for predicting securities’ prices. The first model was developed using back propagation feed forward neural networks. The second model was developed using polynomial classifiers (PC), as a first time application for PC to be used in stock prices prediction. The inputs to both models were identical, and both models were trained and tested on the same data. The study was conducted on Dubai Financial Market as an emerging market and applied to two of the market’s leading stocks. In general, both models achieved very good results in terms of mean absolute error percentage. Both models show an average error around 1.5% predicting the next day price, an average error of 2.5% when predicting second day price, and an average error of 4% when predicted the third day price.

The Diversity of Classifiers and Its Applications to Combination

In various application areas of pattern recognition, combing multiple classifiers is regarded as a new method for achieving a substantial gain in performance of systems. This paper discusses the properties of the diversity of classifiers and its applications. At the same time, the paper presents a novel method for combining multiple classifiers based on the diversity. Fusion strategies are discussed for providing a basis for combing classifiers. These combination strategies are experimentally tested on online handwritten Chinese character recognition system and their effectiveness is considered.

Comparative Study on Tree Classifiers for Application to Condition Monitoring ofWind Turbine Blade through Histogram Features Using Vibration Signals: A Data-Mining Approach

Wind energy is considered as a alternative renewable energy source due to its low operating cost when compared with other sources.The wind turbine is an essential system used to change kinetic energy into electrical energy.Wind turbine blades,in particular,require a competitive condition inspection approach as it is a significant component of the wind turbine system that costs around 20-25 percent of the total turbine cost.The main objective of this study is to differentiate between various blade faults which affect the wind turbine blade under operating conditions using a machine learning approach through histogram features.In this study,blade bend,hub-blade loose connection,blade erosion,pitch angle twist,and blade cracks were simulated on the blade.This problem is formulated as a machine learning problem which consists of three phases,namely feature extraction,feature selection and feature classification.Histogram features are extracted from vibration signals and feature selection was carried out using the J48 decision tree algorithm.Feature classification was performed using 15 tree classifiers.The results of the machine learning classifiers were compared with respect to their accuracy percentage and a better model is suggested for real-time monitoring of a wind turbine blade.

Performance Evaluation of Multiple Classifiers for Predicting Fake News

The rise of fake news on social media has had a detrimental effect on society. Numerous performance evaluations on classifiers that can detect fake news have previously been undertaken by researchers in this area. To assess their performance, we used 14 different classifiers in this study. Secondly, we looked at how soft voting and hard voting classifiers performed in a mixture of distinct individual classifiers. Finally, heuristics are used to create 9 models of stacking classifiers. The F1 score, prediction, recall, and accuracy have all been used to assess performance. Models 6 and 7 achieved the best accuracy of 96.13 while having a larger computational complexity. For benchmarking purposes, other individual classifiers are also tested.

Fault Detection of Fuel Injectors Based on One-Class Classifiers

Fuel injectors are considered as an important component of combustion engines. Operational weakness can possibly lead to the complete machine malfunction, decreasing reliability and leading to loss of production. To overcome these circumstances, various condition monitoring techniques can be applied. The application of acoustic signals is common in the field of fault diagnosis of rotating machinery. Advanced signal processing is utilized for the construction of features that are specialized in detecting fuel injector faults. A performance comparison between novelty detection algorithms in the form of one-class classifiers is presented. The one-class classifiers that were tested included One-Class Support Vector Machine (OCSVM) and One-Class Self Organizing Map (OCSOM). The acoustic signals of fuel injectors in different operational conditions were processed for feature extraction. Features from all the signals were used as input to the one-class classifiers. The one-class classifiers were trained only with healthy fuel injector conditions and compared with new experimental data which belonged to different operational conditions that were not included in the training set so as to contribute to generalization. The results present the effectiveness of one-class classifiers for detecting faults in fuel injectors.

Design of Polynomial Fuzzy Neural Network Classifiers Based on Density Fuzzy C-Means and L2-Norm Regularization

In this paper, polynomial fuzzy neural network classifiers (PFNNCs) is proposed by means of density fuzzy c-means and L2-norm regularization. The overall design of PFNNCs was realized by means of fuzzy rules that come in form of three parts, namely premise part, consequence part and aggregation part. The premise part was developed by density fuzzy c-means that helps determine the apex parameters of membership functions, while the consequence part was realized by means of two types of polynomials including linear and quadratic. L2-norm regularization that can alleviate the overfitting problem was exploited to estimate the parameters of polynomials, which constructed the aggregation part. Experimental results of several data sets demonstrate that the proposed classifiers show higher classification accuracy in comparison with some other classifiers reported in the literature.

Emotional dialog generation via multiple classifiers based on a generative adversarial network

Background Human-machine dialog generation is an essential topic of research in the field of natural language processing.Generating high-quality,diverse,fluent,and emotional conversation is a challenging task.Based on continuing advancements in artificial intelligence and deep learning,new methods have come to the forefront in recent times.In particular,the end-to-end neural network model provides an extensible conversation generation framework that has the potential to enable machines to understand semantics and automatically generate responses.However,neural network models come with their own set of questions and challenges.The basic conversational model framework tends to produce universal,meaningless,and relatively"safe"answers.Methods Based on generative adversarial networks(GANs),a new emotional dialog generation framework called EMC-GAN is proposed in this study to address the task of emotional dialog generation.The proposed model comprises a generative and three discriminative models.The generator is based on the basic sequence-to-sequence(Seq2Seq)dialog generation model,and the aggregate discriminative model for the overall framework consists of a basic discriminative model,an emotion discriminative model,and a fluency discriminative model.The basic discriminative model distinguishes generated fake sentences from real sentences in the training corpus.The emotion discriminative model evaluates whether the emotion conveyed via the generated dialog agrees with a pre-specified emotion,and directs the generative model to generate dialogs that correspond to the category of the pre-specified emotion.Finally,the fluency discriminative model assigns a score to the fluency of the generated dialog and guides the generator to produce more fluent sentences.Results Based on the experimental results,this study confirms the superiority of the proposed model over similar existing models with respect to emotional accuracy,fluency,and consistency.Conclusions The proposed EMC-GAN model is capable of generating consistent,smooth,and fluent dialog that conveys pre-specified emotions,and exhibits better performance with respect to emotional accuracy,consistency,and fluency compared to its competitors.

Recognition of Characters by Adaptive Combination of Classifiers

In this paper, the visual feature space based on the long Horizontals, the long Verticals, and the radicals are given. An adaptive combination of classifiers, whose coefficients vary with the input pattern, is also proposed. Experiments show that the approach is promising for character recognition in video sequences.

Gait Based Human Recognition with Various Classifiers Using Exhaustive Angle Calculations in Model Free Approach

Human Gait recognition is emerging as a supportive biometric technique in recent years that identifies the people through the way they walk. The gait recognition in model free approaches faces the challenges like speed variation, cloth variation, illumination changes and view angle variations which result in the reduced recognition rate. The proposed algorithm selected the exhaustive angles from head to toe of a person, and also height and width of the same subject. The experiments were conducted using silhouettes with view angle variation, and cloth variation. The recognition rate is improved to the extent of 91% using Support vector machine classifier. The proposed method is evaluated using CASIA Gait Dataset B (The institute of Automation, ChineseAcademy of Sciences), China. Experimental results demonstrate that the proposed technique shows promising results using state of the art classifiers.

Flow Field Characteristics of the Rotor Cage in Turbo Air Classifiers

The turbo air classifier is widely used powder classification equipment in a variety of fields. The flow field characteristics of the turbo air classifier are important basis for the improvement of the turbo air classifier＇s structural design. The flow field characteristics of the rotor cage in turbo air classifiers were investigated trader different operating conditions by laser Doppler velocimeter（LDV）, and a measure diminishing the axial velocity is proposed. The investigation results show that the tangential velocity of the air flow inside the rotor cage is different from the rotary speed of the rotor cage on the same measurement point due to the influences of both the negative pressure at the exit and the rotation of the rotor cage. The tangential velocity of the air flow likewise decreases as the radius decreases in the case of the rotor cage＇s low rotary speed. In contrast, the tangential velocity of the air flow increases as the radius decreases in the case of the rotor cage＇s high rotary speed. Meanwhile, the vortex inside the rotor cage is found to occur near the pressure side of the blade when the rotor cage＇s rotary speed is less than the tangential velocity of air flow. On the contrary, the vortex is found to occur near the blade suction side once the rotor cage＇s rotary speed is higher than the tangential velocity of air flow. Inside the rotor cage, the axial velocity could not be disregarded and is largely determined by the distances between the measurement point and the exit.

Classification performance of model coal mill classifiers with swirling and non-swirling inlets

The classification performance of model coal mill classifiers with different bottom incoming flow inlets was experimentally and numerically studied.The flow field adjacent to two neighboring impeller blades was measured using the particle image velocimetry technique.The results showed that the flow field adjacent to two neighboring blades with the swirling inlet was significantly different from that with the non-swirling inlet.With the swirling inlet,there was a vortex located between two neighboring blades,while with the nonswirling inlet,the vortex was attached to the blade tip.The vorticity of the vortex with the non-swirling inlet was much lower than that with the swirling inlet.The classifier with the non-swirling inlet demonstrated a larger cut size than that with the swirling inlet when the impeller was stationary(~0 r·min-1).As the impeller rotational speed increased,the cut size of the cases with non-swirling and swirling inlets both decreased,and the one with the non-swirling inlet decreased more dramatically.The values of the cut size of the two classifiers were close to each other at a high impeller rotational speed(≥120 r·min-1).The overall separation efficiency of the classifier with the non-swirling inlet was lower than that with the swirling inlet,and monotonically increased as the impeller rotational speed increased.With the swirling inlet,the overall separation efficiency first increased with the impeller rotational speed and then decreased when the rotational speed was above 120 r·min-1,and the variation trend of the separation efficiency was more moderate.As the initial particle concentration increased,the cut sizes of both swirling and non-swirling inlet cases decreased first and then barely changed.At a low initial particle concentration(b 0.04 kg·m-3),the classifier with the swirling inlet had a larger cut size than that with the non-swirling inlet.

Integrating RFID Technology with Intelligent Classifiers for Meaningful Prediction Knowledge

Radio Frequency Identification (RFID) is wireless technology that has been designed to automatically identify tagged objects using a reader. Several applications of this technology have been introduced in past literature such as pet identification and luggage tracking which have increased the efficiency and effectiveness of each environment into which it was integrated. However, due to the ambiguous nature of the captured information with the existence of missing, wrong and duplicate readings, the wide-scale adoption of the architecture is limited to commercial sectors where the integrity of the observations can tolerate ambiguity. In this work, we propose an application of RFID to take the reporting of class attendance and to integrate a predictive classifier to extract high level meaningful information that can be used in diverse areas such as scheduling and low student retention. We conclude by providing an analysis of the core strengths and opportunities that exist for this concept and how we might extend it in future research.

Belief Combination of Classifiers for Incomplete Data

Data with missing values,or incomplete information,brings some challenges to the development of classification,as the incompleteness may significantly affect the performance of classifiers.In this paper,we handle missing values in both training and test sets with uncertainty and imprecision reasoning by proposing a new belief combination of classifier(BCC)method based on the evidence theory.The proposed BCC method aims to improve the classification performance of incomplete data by characterizing the uncertainty and imprecision brought by incompleteness.In BCC,different attributes are regarded as independent sources,and the collection of each attribute is considered as a subset.Then,multiple classifiers are trained with each subset independently and allow each observed attribute to provide a sub-classification result for the query pattern.Finally,these sub-classification results with different weights(discounting factors)are used to provide supplementary information to jointly determine the final classes of query patterns.The weights consist of two aspects:global and local.The global weight calculated by an optimization function is employed to represent the reliability of each classifier,and the local weight obtained by mining attribute distribution characteristics is used to quantify the importance of observed attributes to the pattern classification.Abundant comparative experiments including seven methods on twelve datasets are executed,demonstrating the out-performance of BCC over all baseline methods in terms of accuracy,precision,recall,F1 measure,with pertinent computational costs.

Assessment of Supervised Classifiers for Land Cover Categorization Based on Integration of ALOS PALSAR and Landsat Data

Many supervised classification algorithms have been proposed, however, they are rarely evaluated for specific application. This research examines the performance of machine learning classifiers support vector machine (SVM), neural network (NN), Random Forest (RF) against maximum classifier (MLC) (traditional supervised classifier) in forest resources and land cover categorization, based on combination of Advanced Land Observing Satellite (ALOS) Phased Array type L-band Synthetic Aperture Radar (PALSAR) and Landsat Thematic Mapper (TM) data, in Northern Tanzania. Various data categories based on Landsat TM surface reflectance, ALOS PALSAR backscattering and their derivatives were generated for various classification scenarios. Then a separate and joint processing of Landsat and ALOS PALSAR data were executed using SVM, NN, RF and ML classifiers. The overall classification accuracy (OA), kappa coefficient (KC) and F1 score index values were computed. The result proves the robustness of SVM and RF in classification of forest resource and land cover using mere Landsat data and integration of Landsat and PALSAR (average OA = 92% and F1 = 0.7 to 1). A two sample t-statistics was utilized to evaluate the performance of the classifiers using different data categories. SVM and RF indicate there is no significance difference at 5% significance level. SVM and RF show a significant difference when compared to NN and ML. Generally, the study suggests that parametric classifiers indicate better performance compared to parametric classifier.

Video-Based Face Recognition with New Classifiers

An exhaustive study has been conducted on face videos from YouTube video dataset for real time face recognition using the features from deep learning architectures and also the information set features. Our objective is to cash in on a plethora of deep learning architectures and information set features. The deep learning architectures dig in features from several layers of convolution and max-pooling layers though a placement of these layers is architecture dependent. On the other hand, the information set features depend on the entropy function for the generation of features. A comparative study of deep learning and information set features is made using the well-known classifiers in addition to developing Constrained Hanman Transform (CHT) and Weighted Hanman Transform (WHT) classifiers. It is demonstrated that information set features and deep learning features have comparable performance. However, sigmoid-based information set features using the new classifiers are found to outperform MobileNet features.

Ad Hoc Network Hybrid Management Protocol Based on Genetic Classifiers

The purpose of this paper is to solve the problem of Ad Hoc network routing protocol using a Genetic Algorithm based approach. In particular, the greater reliability and efficiency, in term of duration of communication paths, due to the introduction of Genetic Classifier is demonstrated.

Robust model for tunnel squeezing using Bayesian optimized classifiers with partially missing database

Accurately predicting and estimating the squeezing and ground response to tunneling remains challenging.Moreover,tunnel-squeezing hazards are much more likely to occur in deeply buried long tunnels with complex engineering-geological environments.There-fore,a high-performance predictive model for tunnel squeezing is necessary.A superior ensemble classifier is put forward in this study,which is composed of four individual classifiers(gradient boosting classifier,extra-trees classifier,AdaBoost classifier,and Logistic regression classifier)and two optimization algorithms(Bayesian optimization(BO)and sparrow search algorithm(SSA)).The training database covers five parameters:tunnel depth(H),rock tunneling quality index(Q),tunnel diameter(D),support stiffness(K),and strength stress ratio(SSR),about which the basic information is accessible at the early design phases.However,the dataset compiled from the literature is insufficient.Thus,the ten proposed methods are used to replace the missing values.During the model training pro-cess,BO shows its strong ability to optimize seventeen hyperparameters.When applied to tune the classifiers’weights,SSA achieves a fast and efficient performance.The novel Shapley Additive Explanations–LightGBM method indicates that the K is the most important input feature,followed by SSR,Q,H,and D,respectively.The ensemble classifier is then validated using the test set and additional his-torical case projects.The validation shows that the model can achieve an accuracy of 98%(i.e.,the error rate is 2%)on the test set,higher than those achieved by previous prediction models.Moreover,the predicted probability could provide warning information for timely support measures.Finally,the application results are illustrated through tests on the tunnel sections that have not yet been excavated in the line of the Sichuan–Tibet railway project.The applied predictive tendencies and laws are in line with the practical experience.In sum-mary,the proposed model’s prediction results are reasonable,and its prediction will be more accurate as more data is collected and trained for prewarning the tunnel squeezing hazard.