Search

One of the main concerns of cloud-based services based on machine and deep learning algorithms is the privacy of users' data. This is particularly relevant when companies want to leverage such services because they have to outsource potentially sensible data to be processed. In this work, the problem of privacy-preserving anomaly detection on industrial vibrational data with machine learning is tackled. It consists in the detection of irregularities or deviations from expected patterns in the vibration signals generated by industrial machinery and equipment. Such anomalies can be indicative of potential equipment failures, maintenance needs, or process deviations, making their timely detection critical for ensuring the smooth operation and reliability of industrial systems. We combine this industrial need with the ability to guarantee data privacy by proposing encrypted vibrational anomaly detection (EVAD). EVAD allows the detection of anomalies on vibrational data in a privacy-preserving manner by integrating, for the first time in the literature, one-class support vector machines and homomorphic encryption, the latter being a particular kind of encryption that allows the computation of some operations directly on encrypted data. Experimental results show that, on two publicly available datasets for vibrational anomaly detection, EVAD is able to distinguish, in a privacy-preserving manner, between nominal and anomaly situations, in an effective and efficient way. To the best of our knowledge, EVAD represents the first privacy-preserving solution for the detection of anomalies in vibrational data present in the literature. [ABSTRACT FROM AUTHOR]

Text classification has critical applications, including healthcare, where it can assist patients in locating specialized clinics based on their symptom descriptions. This can enhance healthcare services by reducing misdiagnoses and efficiently guiding patients to appropriate specialists. This research focuses on building a model for multi-class text classification in Arabic healthcare. Two evaluation schemes were employed using health data from the Altibbi dataset. Various feature extraction techniques were employed, including term frequency-inverse document frequency (TF-IDF) and word-to-vector (Word2Vec). Classical machine learning models such as Logistic Regression (LR), Random Forest (RF), Multinomial Naïve Bayes (MNB), Stochastic Gradient Descent, and Support Vector Machine (SVM), along with ensemble classifications, were implemented. Evaluation measures were used to compare model performance, including precision, recall, and F1-score. Results indicated that TF-IDF generally outperformed Word2Vec regarding the F1-score across most classes. LR and RF consistently demonstrated strong performance, while MNB exhibited lower precision. Ensemble methods like stacking and bagging showed promise, with LR and RF performing well across multiple classes. The SVM model displayed lower precision in certain classes. Preliminary experiments with the transformer-based models for Arabic language AraBERT and MarBERT as feature extraction and classification models demonstrated impressive precision, recall, and F1 scores across various classes. The findings show the effectiveness of ensemble models using TF-IDF and Arabic transformer models in accurately classifying Arabic health texts. However, challenges related to standardization, semantic variations, data availability, and resources should be considered. [ABSTRACT FROM AUTHOR]

Categorical features appear in datasets from almost every practice area, including real estate datasets. One of the most critical handicaps of machine learning algorithms is that they are not designed to capture the qualitative nature of the categorical features, leading to sub-optimal predictions for the datasets with categorical observations. This study focuses on a new fuzzy regression functions framework, namely hierarchical fuzzy regression functions, that can handle categorical features properly for the regression task. The proposed framework is benchmarked with linear regression, support vector machines, deep neural networks, and adaptive neuro-fuzzy inference systems with real estate data having categorical features from six markets. It is observed that the proposed method produces better prediction performance for real estate price prediction than the benchmark methods in a wide variety of real estate markets. Since we provide all the required software codes to implement the proposed hierarchical fuzzy regression functions framework, our approach offers practitioners a readily applicable, high-performing tool for real estate price prediction and other regression problems involving categorical independent features. [ABSTRACT FROM AUTHOR]

The catastrophic forgetting problem is one of the hotspots in the field of deep learning. At present, there is no doubt that storing samples of previous tasks in fixed-size memory is the best way to solve this problem. However, the number of samples stored in fixed-size memory is limited. With the increase in tasks, the number of samples stored in memory for a single task will decrease sharply. It is also difficult to balance the memory capacity and number of samples. To solve this problem, some methods use a fixed size of memory to store dimensionality reduction images. However, this will create new problems. 1) The quality of dimensionality reduction images is poor, and they are significantly different from original images. 2) How to choose the dimensionality reduction method of images. To address these problem, we put forward a new method. Firstly, we employ a simple and reliable scheme to solve the domain difference between dimensionality reduction images and original images. And we theoretically analyzed which image dimensionality reduction method is better. Secondly, to increase the generalization ability of our method and further mitigate the catastrophic forgetting phenomenon, we utilize a self-supervised image augmentation method and the output features similarity loss. Thirdly, we make use of the neural kernel mapping support vector machine theory to improve the interpretability of our method. Experimental results demonstrated that the top-1 average accuracy of our method is much higher than other methods when using the same size of memory. [ABSTRACT FROM AUTHOR]

This paper presents an intelligent healthcare framework by incorporating modern computing technologies like machine learning and deep learning. The sole motivation of this paper is to predict diabetes status intelligently so that patients can be made aware of their diabetes status regularly. A novel ensemble machine learning-based diabetes prediction framework, namely "iDP", has been proposed in this paper. Six different machine learning approaches have been used for the iDP framework, namely random forest, decision tree, neural network, AdaBoost, support vector machine, and XGBoost. Training and testing of ensemble modeling have been performed based on these learning techniques. A comprehensive preliminary data analysis and screening of statistical analysis have been systematically demonstrated in this paper. The recursive feature elimination and fivefold cross-validation methods have been used to select the best model. All experiments have been performed using R-Studio 1.2.5 with the help of the Rattle library 5.4.0. The preliminary data analytic and Pearson correlation coefficient values have been computed to verify the linearity relationship among the data. Results have been reported in eight performance metrics, including accuracy, specificity, sensitivity, area under curve, area under convex hull, minimum error rate, minimum weighted loss, and precision. The proposed iDP framework improves the results on different aspects in order to compare them with individual learning approaches. Despite this, a comparative performance analysis of the iDP framework has also been performed with state-of-the-art prevailing models. As per experiment, the iDP framework gives better results in most cases, like accuracy: 95.26%, AUC: 91.15%, sensitivity: 96.81%, specificity: 97.72% and precision: 90.72%. [ABSTRACT FROM AUTHOR]

Breast cancer is a significant global health concern, with millions of cases and deaths each year. Accurate diagnosis is critical for timely treatment and medication. Machine learning techniques have shown promising results in detecting breast cancer. Previous studies have primarily used single-modality data for breast cancer diagnosis. Hence, this work aims to mobilize the benefits of multimodal data over unimodality samples. This study proposes a custom deep learning-based model pipeline that works over this multimodal data. This work has been separated into three phases. Phase 1 and Phase 2 under the unimodal category examine gene expression data and histopathological images separately. The Cancer Genome Atlas makes these datasets available. In Phase 3, the proposed pipeline operates on both data types' samples for each patient in the multimodal category. This study investigates how data pre-processing (cleaning, transformation, reduction) and cascaded filtering affect model performance. Precision, recall, f1-score, and accuracy assessed the models, whereas L2 regularization, exponentially weighted moving average, and transfer learning minimized over-fitting. A custom deep neural network and support vector machine obtained 86% accuracy in Phase 1, whereas the VGG16 model reached 80.21% accuracy in Phase 2. In Phase 3, the curated multimodal dataset was applied to a custom deep learning pipeline (VGG16 backbone with hyper-tuned machine learning models as head classifiers) to achieve 94% accuracy, demonstrating the importance of multimodal data over unimodal in breast cancer subtype classification. These findings highlight the importance of multimodal data for breast cancer diagnosis and subtype prediction. [ABSTRACT FROM AUTHOR]

Accurate short-term electricity load forecasts are critical for the secure and economic operation of power systems. This paper presents a computationally efficient and powerful three-stage model to accurately forecast short-term electricity load. Variational mode decomposition (VMD) was used in the first stage to extract features from the historical load signal. The stacked kernel extreme learning machine (KELM)-based auto-encoders were utilized in the unsupervised feature learning-based second stage. In the third stage, the high-order learned features were used as the inputs for the KELM-based regression model. The proposed deep KELM architecture combines stacked KELM-based auto-encoders and a KELM-based regression model to forecast short-term electricity load effectively. In order to examine the performance improvement of the proposed forecasting model, several performance comparison tests were realized using publicly available electricity load and day-ahead forecast data from the Turkish transmission system operator (TSO). The proposed model results were compared with state-of-the-art deep extreme learning machine (ELM) architectures as well as the benchmark forecasting models based on original ELM, KELM, artificial neural network (ANN), support vector machine (SVM), and regression tree (RT). The comparison results indicated that the proposed model outperformed state-of-the-art architectures and was significantly more successful than the TSO model. [ABSTRACT FROM AUTHOR]

Recently, various techniques have been implemented to investigate the cavity beneath the pavement using ground-penetrating radar (GPR) data. These techniques usually focus on the determination of the height of the cavity and the depth at which it is located, which play a vital role in the agencies' decision-making on the appropriate rehabilitation method for the pavement. Image processing, a relatively advanced technique that utilizes the image generated from the GPR data, has been used for this purpose. However, it is still time-consuming and yields low accuracy. As a solution, this study proposes a machine learning (ML)-based method to estimate cavity properties, such as depth and height, from GPR images with high precision in a significantly reduced time. In this study, a deep convolutional generative adversarial networks-based hyperbolic fitting algorithm, called GAN–cavity, was developed based on the Pix2Pix model to identify multi-shape signature segmentation from the GPR data. The GAN–cavity was trained and tested using 1920 and 480 images, respectively, and the model performed well with an average F1–score of 83.1 per cent. Consequently, an extreme gradient boosting (XGBoost) algorithm was used to predict the cavity depth and height based on the GAN–cavity results. Moreover, the developed model (XGB–cavity classifier) can classify the cavity's height into three classes: small, medium, and large. The XGBoost model XGB–cavity regressor used 341 and 87 cavities for training and testing, respectively, yielding regression scores (R2) and root mean squared error (RMSE) of 0.88 and 3.28 cm in cavity depth prediction, respectively. By applying the support vector machine synthetic minority oversampling technique (SVM–SMOTE) algorithm to overcome the data imbalance, the XGB–cavity classifier model obtained a high accuracy of 0.89 with precision–avg, recall–avg, and F1–score–avg of 0.83, 0.83, and 0.83, respectively. Based on the results, the proposed method (GAN–XGB–cavity) of combining the GAN–cavity and XGB–cavity can successfully identify and fit multi-shape segmentation for depth estimation and height classification of cavities. [ABSTRACT FROM AUTHOR]

With the increased use of social network sites, such as Twitter, attackers exploit these platforms to spread counterfeit content. Such content can be fake advertisements or illegal content. Classifying such content is a challenging task, especially in Arabic. The Arabic language has a complex structure and makes classification tasks more difficult. This paper presents an approach to classifying Arabic tweets using classical machine learning (non-deep machine learning) and deep learning techniques. Tweets corpus were collected through Twitter API and labelled manually to get a reliable dataset. For an efficient classifier, feature extraction is applied to the corpus dataset. Then, two learning techniques are used for each feature extraction technique on the created dataset using N-gram models (uni-gram, bi-gram, and char-gram). The applied classical machine learning algorithms are support vector machines, neural networks, logistics regression, and naïve Bayes. Global vector (GloVe) and fastText learning models are utilised for the deep learning approaches. The Precision, Recall, and F1-score are the suggested performance measures calculated in this paper. Afterwards, the dataset is increased using the synthetic minority oversampling technique class to create a balanced dataset. After applying the classical machine learning models, the experimental results show that the neural network algorithm outperforms the other algorithms. Moreover, the GloVe outperforms the fastText model for the deep learning approach. [ABSTRACT FROM AUTHOR]

With each passing day, the number of smart vehicles is increasing manifold, hence, automatic/automated parking lot detection is gaining a lot of importance among Smart City applications. A robust approach is desired to identify parking spaces effectively and efficiently. This work presents a deep learning classifier based on convolutional neural network (CNN) and extreme learning machine (ELM), i.e., CNN-ELM to classify the parking space as vacant or occupied. CNN is well-known for efficient image classification, but its training time is highly influenced by backpropagation of errors in the fully connected layer. Thus, ELM is plugged into CNN to replace the fully connected layer and perform classification whereas, feature extraction is performed using CNN. The performance of CNN-ELM is validated on the publicly available PKLot dataset, which contains approximately 700,000 images categorized into sunny, overcast, and rainy weather conditions. The experimental results indicate that CNN-ELM approach outperforms other hybrid CNN models using different classifiers such as support vector machine, Xgboost, and Extra Trees in terms of sensitivity, specificity, and accuracy. The comparison of results with other state-of-the-art approaches based on accuracy and Area under the curve (AUC) score further justifies the effectiveness of the proposed approach in real-time parking space detection. [ABSTRACT FROM AUTHOR]

Deep convolutional neural networks (DCNNs) are one of the most advanced techniques for classifying images in a range of applications. One of the most prevalent cancers that cause death in women is breast cancer. For survival rates to increase, early detection and treatment of breast cancer is essential. Deep learning (DL) can help radiologists diagnose and classify breast cancer lesions. This paper proposes a computer-aided system based on DL techniques for automatically classify breast cancer tumors in histopathological images. There are nine DCNN architectures used in this work. Four schemes are performed in the proposed framework to find the best approach. The first scheme consists of pre-trained DCNNs based on the transfer learning concept. The second scheme performs feature extraction of the DCNN architectures and uses a support vector machine (SVM) classifier for evaluation. The third one performs feature integration to show how the integrated deep features may enhance the SVM classifiers' accuracy. Finally, in the fourth scheme, the Chi-square (χ2) feature selection method is applied to reduce the large feature size in the feature integration step. The results of the proposed system present a promising performance for breast cancer classification with an accuracy of 99.24%. The system performance shows that the proposed tool is suitable to assist radiologists in diagnosing breast cancer tumors. [ABSTRACT FROM AUTHOR]

A computerized coconut tree detection system can help dendrologists and laypersons in identifying coconut trees based on three morphological parameters including height, inclination, and orientation. These three parameters help to determine the health and the nature of growth of coconut trees which influences the design and use of robots for harvesting coconuts. Deep learning is a powerful tool used for feature extraction as it is better in extracting deeper details (features) in an image. In this research work, a new Modified Inception Net based Hyper Tuning Support Vector Machine classification method named MIN-SVM is proposed for coconut tree classification based on three morphological parameters including height, inclination and orientation. The features from the pre-processed coconut tree images were extracted using four distinct Convolutional Neural Network models including Visual Geometry Group, Inception Net, ResNet, and MIN-SVM. These extracted features were then classified using a Machine Learning model named Support Vector Machine (SVM). The MIN-SVM have achieved a remarkable accuracy of 95.35 percent as contrasted to Visual Geometry Group (91.90%), Inception Net (81.66%), and ResNet (71.95%). The features extracted from Modified Inception Net fitted good with SVM classifier. Experimental results show that MIN-SVM can be powerful computerized automated system to identify coconut trees based on height, inclination, and orientation. [ABSTRACT FROM AUTHOR]

In this manuscript, imbalanced and small sample space (IB-SSS) dataset problems for pedestrian gender classification using fusion of selected deep and traditional features (PGC-FSDTF) are considered. In this regard, data preparation is first done through data augmentation and preprocessing steps to handle imbalanced classification problem and environmental effects, respectively. The proposed approach follows different types of feature extraction schemes, for instance, pyramid histogram of oriented gradients, hue saturation value histogram, deep visual features of DenseNet201 and InceptionResNetV2-based convolutional neural network architectures. The parallel fusion method computes the maximum and average values-based features from the learned features of both deep networks. Features are selected through features selection methods such as entropy and principal component analysis (PCA). The subsets of features are serially fused and provided to multiple classifiers to perform gender classification on IB-SSS datasets. Resultantly, the proposed PGC-FSDTF method shows better results in terms of different accuracies (overall, mean, and balanced), and area under curve on selected datasets. Further, improved results are achieved on applied datasets using PCA-based selected features and medium Gaussian support vector machine (M-SVM) classifier. These results on different datasets confirm that the selected feature combination provides a way to handle IB-SSS issues for PGC effectively. [ABSTRACT FROM AUTHOR]

In this study, a new deep learning-based approach has been developed that detects and classifies surface defects that occur in the steel production process. The proposed methodology was created in four steps. In the first step, a deep learning model is designed that trains the residual and attention structures in parallel, thus increasing the classification performance. In the second step, deep features were extracted from the Parallel Attention Residual-Convolutional Neural Network model. The extracted features in the third step were selected by a new and simple algorithm (NCA-ReliefF Matched Index) based on matching the indexes obtained from the Neighborhood Component Analysis and Relief algorithms. In the last process, classification was done with the support vector machine algorithm. The proposed methodology was used for dual and multi-class classification tasks and evaluated on a dataset in the Kaggle database named Severstal: Steel Defect Detection. [ABSTRACT FROM AUTHOR]

Hyperspectral images (HSIs) contain significant number of contiguous dense spectral bands which often have large redundancy and high correlation that subsequently results into "curse of dimensionality" in HSI analysis. Therefore, efficient band selection techniques are crucial for dimensionality reduction of HSIs without any significant loss of spectral information contained in it. In this paper, deep learning autoencoders and genetic algorithm (GA) are used for efficient selection of the most revealing bands from a remotely sensed HSI. The proposed method formulates the HSI band selection process as a GA-based evolutionary optimization that minimizes the reconstruction error of an autoencoder which uses a few informative bands for HSI reconstruction. The proposed approach starts with spectral segmentation of the bands in an HSI into a number of spectral regions, and then, different autoencoders are trained on each segment with the original input band vectors contained in the segmented region. Finally, GA-based search heuristics is applied on each region in order to find out sparse sub-combination of spectral bands in such a way that the trained autoencoders would reconstruct the original segmented spectral vectors from the resulting band sub-combinations with least reconstruction errors. The final band selection is carried out by aggregating all the band sub-combinations returned from the segmented regions. Finally, the effectiveness of the proposed method is verified through selected bands validation by a support vector machine classifier. Experimental results on three publicly available HSI datasets depict the consistently superior effectiveness of the proposed band selection method over other state-of-the-art methods in land cover classification of remotely sensed HSIs. [ABSTRACT FROM AUTHOR]

The accurate diagnosis of Alzheimer's disease (AD) in the early stages, such as significant memory concern (SMC) and mild cognitive impairment (MCI), is essential in order to slow its progression through timely treatment. Recent achievements have shown that fusing multimodal neuroimaging data effectively facilitates AD diagnosis. However, most proposed fusion methods simply add or concatenate multimodal features and do not make full use of nonlinear features and texture features across the range of modalities. This paper proposes a diagnostic model that effectively diagnoses AD in different stages by fusing functional magnetic resonance imaging (fMRI) and structural MRI (sMRI) information. First, fMRI and sMRI scans are preprocessed, and mean regional homogeneity (mReHo) transformation is performed for the preprocessed fMRI scans. Then, 3DMR-PCANet extracts features of mReHo images. The basic ResNet module is stacked to build a 3DResNet-10 model for feature extraction of sMRI scans. Next, two image features are fused by kernel canonical correlation analysis. Finally, a support vector machine (SVM) is utilized for the classification of fused features. Experimental results on the Alzheimer's Disease Neuroimaging dataset demonstrate the effectiveness of the proposed method. Specifically, this method improves on the accuracy, specificity, sensitivity, F1 value and area under the curve (AUC) of existing methods in comparisons of the normal control (NC) versus SMC, NC versus MCI, NC versus AD, SMC versus MCI, SMC versus AD, and MCI versus AD groups, which confirms that the proposed method can mine information on the correlation between fMRI and sMRI data of the same subject and can effectively classify AD patients in different stages. [ABSTRACT FROM AUTHOR]

Hyperspectral imaging is highly important with respect to the detection, identification and classification of various natural resources—minerals, earth's natural eruptions, vegetation and related man-made materials and other existing backgrounds. In this study, a novel deep learning-based fuzzy-twin proximal support vector machine (DL-FTPSVM) kernel neural network model is proposed to perform an effective hyperspectral image (HSI) classification. The modelled new DL-FTPSVM is designed to handle the irregularities and existing training complexities of the present hyperspectral image classifier models. A mapreduce framework is introduced for analysing the large volumes of hyperspectral images by partitioning these images into few portions based on mapper and reducer functions. The fuzzy-twin proximal support vector machine neural model is designed with the fuzzy twin hyperplanes comprising of Gaussian kernel, Bessel kernel, ANOVA radial basis function (RBF) kernel and linear spline kernel and a simple triangular fuzzy function is employed to construct the twin hyperplanes for HSI classification. Deep learning (DL) framework supports the new fuzzy-twin proximal SVM kernels to learn more effectively with the auto-encoder and decoder mechanism. For testing and validating the proposed DL-FTPSVM neural network model, the test beds are the Kennedy Space Centre datasets, University of Pavia datasets, Salinas datasets and Indian Pine datasets and performance metrics are evaluated. Results evaluated prove the superiority of the DL-FTPSVM model incurring better classification accuracy compared to techniques from previous works. The randomness of the proposed classifier is handled with the statistical analysis for the HSI classification. [ABSTRACT FROM AUTHOR]

Machine learning (ML) is a fast-evolving field of artificial intelligence that has been applied in many domains due to the increasing availability of computerized databases, including materials science; for instance, validating crystal descriptors for energy prediction poses difficult problems. This work investigates machine learning models to substitute the laboratory crystal energy prediction using two- and three-body distribution functions as structural and atomic descriptors. To achieve this, ML algorithms were used notably ElasticNet, Bayesian Ridge, Random Forest, Support Vector Machine, and Deep Neural Networks to model structural descriptors. Moreover, a non-conventional Deep Neural Networks topology was developed and implemented to model atomic descriptors. Five-fold cross-validation procedure was performed on each model; quality assessment metrics were else used for testing and evaluation in order to identify the most robust descriptors. Finally, the best result of energy prediction was achieved by combining both two- and three-body atomic distribution functions. [ABSTRACT FROM AUTHOR]

In this study, the extreme learning machine, support vector machine, and deep learning models were used to predict the effect on the voting behavior of election music's used in the elections in Turkey. A questionnaire was conducted to measure the effect of election music on the voting behavior in Turkey. This questionnaire was applied face to face to 412 people. The twelve questions were used to determine the effect of the election music on the voters. The first eleven of these questions in the questionnaire was used as the input variables of the models. The last question was determined whether the election music affects the voting behavior or not. The last question was selected as the output variable of the models. The extreme learning machine, support vector machine, and deep learning models estimated the voting behavior with 98.32%, 92.02% and 99.08% accuracy, respectively. Moreover, the sensitivity analysis and t-test were performed for the input variables in the models. These analyses obtained that 7th question was the most important question on the questionnaire. This study found that politicians can be used the deep learning algorithm to estimate the voting behavior in elections. [ABSTRACT FROM AUTHOR]

Alzheimer's disease (AD) is an irreversible neurodegenerative disease characterized by thinking, behavioral and memory impairments. Early prediction of conversion from mild cognitive impairment (MCI) to AD is still a challenging task. No study has been able to predict the exact conversion time of MCI patients. In addition, most studies have achieved poor performance making this prediction using only a small number of features (e.g., using only MRI images). Therefore, previous approaches have not gained the trust of medical experts. This study proposes a novel two-stage deep learning AD progression detection framework based on information fusion of several patient longitudinal multivariate modalities, including neuroimaging data, cognitive scores, cerebrospinal fluid biomarkers, neuropsychological battery markers, and demographics. The first stage of the progression detection framework employs a multiclass classification task that predicts a patient's diagnosis (i.e., cognitively normal, MCI, or AD). In the second stage, a regression task that predicts the exact conversion time of MCI patients is used. The study is based on data of 1,371 subjects collected by the Alzheimer's Disease Neuroimaging Initiative (ADNI). Comprehensive experiments were carried out to evaluate the framework stages and find the optimal model for each stage. Proposed model was compared with various machine learning models, including decision tree (DT), random forest (RF), support vector machine (SVM), logistic regression (LR), and K-nearest neighbor (KNN). In the classification stage, the proposed long-short term memory (LSTM) model achieved an accuracy of 93.87%, precision of 94.070%, recall of 94.07%, and F1-score of 94.07%. The results showed that the LSTM model outperformed other machine learning models (i.e., decision tree by 2.48%, random forest by 1.27%, support vector machine by 1.86%, logistic regression by 1.59%, and K-nearest neighbor by 14.77%). In the regression stage, the proposed LSTM model achieved the best results (i.e., mean absolute error of 0.1375). Compared to other regular regressors, this LSTM model achieved less errors (i.e., 0.0064, 0.0152, 0.0338, 0.0118, 0.0198, and 0.0066, compared to DT, RF, SVM, LR, and KNN, respectively). By learning deep representation from patient high-dimensional longitudinal time-series data, the proposed LSTM model was more stable and medically acceptable. The framework may have a clinical impact as a predictive tool for AD progression detection due to its accurate results to predict the exact conversion time of MCI cases using patient time-series multimodalities data. [ABSTRACT FROM AUTHOR]

In Genome-Wide Association Studies (GWAS), detection of T2D-related variants in genome sequences and accurate modeling of the complex structure of the relevant gene are of great importance for the diagnosis of diabetes. For this purpose, this paper presents a novel strong algorithm to accurately and effectively identify Type 2 Diabetes (T2D) risk variants at high-performance rates. The proposed algorithm consists of five important phases. The first stage is to collect T2D-associated DNA sequences and to digitize them by the Entropy-based technique. The second stage is to transform these digitized DNA sequences into 224 × 224 pixels size spectrum images. The third is to extract a distinctive feature set from these spectrum images using the ResNet and VGG19 architectures. The fourth is to classify the effective feature set using SVM and k-NN methods. The last stage is to evaluate the system with k-fold cross-validation. As a result of the developed algorithm, the performances of the used Convolutional Neural Network (CNN) methods, the Entropy-based technique, and the classifiers were compared in relation. As a result of the study a combination model of the proposed Entropy-based technique, ResNet and Support Vector Machine (SVM) achieved the highest accuracy rate with 99.09%. With this study, the performance of the system in the extraction of epigenetic features and prediction of T2D from spectrogram images was investigated. The results show that the system will contribute to the identification of all genes in diabetes-related tissue and studies on new drug targets. [ABSTRACT FROM AUTHOR]

In ancient times, there was no system to record or document music. A basic notation system to write European music was formulated around 14th century in the Baroque period which slowly evolved into the standard notation system that we have today. Later, the musical pieces from the classical and post-classical period of European music were documented as scores using this standard European staff notations. These notations are used by most of the modern genres of music due to their versatility. Hence, it is very important to develop a method that can store such music sheets containing handwritten music scores digitally. Optical music recognition (OMR) is a system that automatically interprets the scanned handwritten music scores. In this work, we have proposed a classifier ensemble of deep transfer learning models with support vector machine (SVM) as the aggregator for handwritten music symbol recognition. We have applied three pre-trained deep learning models, namely ResNet50, GoogleNet and DenseNet161 (each trained on ImageNet), and fine-tuned on our target datasets i.e., music symbol image datasets. The proposed ensemble technique can capture a more complex association of the base classifiers, thus improving the overall performance. We have evaluated the proposed model on five publicly available standard datasets, namely Handwritten Online Music Symbols (HOMUS), Capitan_Score_Uniform, Capitan_Score_Non-uniform, Rebelo_real and Fornés, and achieved state-of-the-art results for all these datasets. Additionally, we have evaluated our model on publicly available two non-music symbols datasets, namely CMATERdb 2.1.2 containing 120 handwritten Bangla city names and CMATERdb 3.1.1 dataset containing handwritten Bangla numerals to validate its effectiveness on diversified datasets. The source code of this present work is available at https://github.com/ashis0013/Music-Symbol-Recognition. [ABSTRACT FROM AUTHOR]

Cancer is a severe condition of uncontrolled cell division that results in a tumor formation that spreads to other tissues of the body. Therefore, the development of new medication and treatment methods for this is in demand. Classification of microarray data plays a vital role in handling such situations. The relevant gene selection is an important step for the classification of microarray data. This work presents gene encoder, an unsupervised two-stage feature selection technique for the cancer samples' classification. The first stage aggregates three filter methods, namely principal component analysis, correlation, and spectral-based feature selection techniques. Next, the genetic algorithm is used, which evaluates the chromosome utilizing the autoencoder-based clustering. The resultant feature subset is used for the classification task. Three classifiers, namely support vector machine, k-nearest neighbors, and random forest, are used in this work to avoid the dependency on any one classifier. Six benchmark gene expression datasets are used for the performance evaluation, and a comparison is made with four state-of-the-art related algorithms. Three sets of experiments are carried out to evaluate the proposed method. These experiments are for the evaluation of the selected features based on sample-based clustering, adjusting optimal parameters, and for selecting better performing classifier. The comparison is based on accuracy, recall, false positive rate, precision, F-measure, and entropy. The obtained results suggest better performance of the current proposal. [ABSTRACT FROM AUTHOR]

The recognition of stomatological disorders and the classification of dental caries are important areas of biomedicine that can hugely benefit from machine learning tools for the construction of relevant mathematical models. This paper explores the possibility of using reflectivity data to distinguish between healthy tissues and caries by deep learning and multilayer convolutional neural networks. The experimental data set includes more than 700 observations recorded in the stomatology laboratory. For rigor, the results obtained from the deep learning systems are compared with those evaluated for selected sets of features estimated for each observation and classified by a decision tree, support vector machine (SVM), k-nearest neighbor, Bayesian methods, and two-layer neural networks. The classification accuracy obtained for the deep learning systems was 98.1% and 94.4% for data in the signal and spectral domains, respectively, in comparison with an accuracy of 97.2% and 87.2% evaluated by the SVM method. The proposed method conclusively demonstrates how the artificial intelligence and deep learning methodology can contribute to improved diagnosis of dental problem in stomatology. [ABSTRACT FROM AUTHOR]

In smartphone-based crowd/participatory sensing systems, it is necessary to identify the actual sensor data provider. In this context, this paper attempts to recognize the users' identity based on their gait patterns (i.e. unique walking patterns). More specifically, a deep convolution neural network (CNN) model is proposed for the user identification with accelerometer data generated from users smartphone sensors. The proposed model is evaluated based on the real-world benchmark dataset (accelerometer biometric competition data) having a total of 387 users accelerometer sensor readings (60 million data samples). The performance of the proposed CNN-based approach is also compared with five baseline methods namely Support Vector Machine (SVM), Decision Tree (DT), Random Forest (RF), Logistic Regression (LR), and K-Nearest Neighbours (KNN). It is observed that the proposed model achieves better results (accuracy = 98.8%, precision = 0.94, recall = 0.97, and F1-score = 0.95) as compared to the baseline methods. [ABSTRACT FROM AUTHOR]

In this work, an effective classification of hyperspectral images is modelled and simulated with the proximal support vector machine (PSVM) by integrating them with the deep learning approach. The modelled new deep proximal support vector machines are designed in a manner to handle the existing complexity, discrepancies and irregularities in the traditional hyperspectral image classifiers. This paper investigates the applicability of the new deep linear and nonlinear proximal support vector machines as applied for hyperspectral image classification. In respect of the new deep PSVM classifier, it is modelled for deep linear PSVM and deep nonlinear PSVM to perform classification of spectral images so as to bring out the best classifier model. To test and validate the proposed deep PSVM classifiers University of Pavia datasets, Indian Pine datasets and Kennedy Space Centre datasets are employed as test beds and results are attained. The developed new deep PSVM classifiers are developed with varied kernel functions to do the classification process. The deep learning technique enhances the linear and nonlinear PSVM classifier models to perform more effectively during the learning process and carry out the classification using auto-encoders and decoders. Results attained during the process infer that the developed new deep PSVM (linear and nonlinear) has come out with better classification accuracy in comparison with that of the other techniques from literature for the same datasets. Statistical analysis validates the randomness that occurs in the proposed deep learning techniques as applied for spectral image classification. [ABSTRACT FROM AUTHOR]

This work presents an efficient hybridized approach for the classification of electrocardiogram (ECG) samples into crucial arrhythmia classes to detect heartbeat abnormalities. The physiological detection using electrocardiogram (ECG) signals has been the most popular means and widely accepted automated detection system to monitor heart health. Additionally, arrhythmia beat classification plays a prominent role in electrocardiogram (ECG) analysis dedicated elucidate cardiac health status while analyzing heart rhythm. The authors aim to classify ECG samples into major arrhythmia classes precisely by removing the inherent noise of ECG signals in preprocessing phase using discrete wavelet transformation (DWT). The QRS complex plays a crucial role in ECG signal identification. Therefore, the position and amplitude of R-peaks are determined to detect the QRS complex. The feature vectors of the QRS complex are further optimized with cuckoo search (CS) optimization algorithm in addition to denoising signals using DWT to select the most relevant set of features. The Support vector machine (SVM)-trained support vector contains the best training information used to train feed-forward back-propagation neural network (FFBPNN) to propose the variant DWT + CS + SVM-FFBPNN to classify signals among five classes. MIT-BIH arrhythmia database is utilized for different types of heartbeats. The classification analysis based on a variant with optimized feature vector using cuckoo search algorithm and SVM-FFBPNN determines heart rate with an accuracy of 98.319%. In contrast, the variant FFBPNN without optimization obtains 97.95% accuracy. The improved performance of the novel combination of classifiers resulted in overall classification accuracy of 98.53% with precision and recall of 98.247% and 95.68%, respectively. The simulation analysis comprising 3600 samples and 1160 heartbeats also outperformed the existing arrhythmia classifications performed based on neural networks. This illustrates the success of the proposed ECG classification model in accurately categorizing ECG signals for arrhythmia classification. [ABSTRACT FROM AUTHOR]

Alzheimer's disease (AD), which occurs as a result of the loss of cognitive functions in the brain, causes near-forgetfulness in the case and dementia in subsequent processes. Dataset consists of MR images containing four phases of AD. The dataset was re-enhanced separately with DeepDream, fuzzy color image enhancement, hypercolumn techniques. Visual Geometry Group-16 (VGG-16) deep learning model is used in the enhancing process and deep features are combined. Linear Regression is used for the selection of efficient features. The Support Vector Machine is preferred as a classifier. With the proposed approach, the classification achievement was obtained as 100% in Mild Dementia, 99.94% in Moderate Dementia, 100% in non-Dementia, 99.94% in Very Mild Dementia. The overall accuracy was 99.94%. The proposed approach increased the prediction success in detecting Alzheimer's stages by re-enhancing MR images. Thus, an efficient early diagnosis model was realized at an affordable cost for individuals likely to progress with dementia. [ABSTRACT FROM AUTHOR]

One of the most captivating problems being faced nowadays in Physics are ultra-high energy cosmic rays. They are high-energy radiations coming mainly from outside the Solar System, and when they enter Earth's atmosphere, they produce a cascade of particles. This cascade of particles, named as extensive air shower, can be recorded by means of photomultiplier tubes in surface detectors, obtaining different recordings of the energy signal (since the air shower can hit one or more detectors). Based on these signals, different features can be obtained that might give an insight into which particle has caused the extensive air shower, which is of utmost importance for physicists. Therefore, this work presents a supervised learning algorithm to determine that the particle is a photon or a hadron. Convolutional neural networks and feed forward neural networks are compared in order to analyze the importance of spatial information for the classification. Then, a comparison between using the information of each surface detector separately and integrating the information from them for the classification is studied, showing that the integration improves the results compared to using each surface detector's trace independently. Furthermore, a comparison between manually extracted features from the signal and the automatically extracted features by the convolutional neural network is done, showing the classification potential of the latter. Finally, the addition of particle shower features which are external to the surface detector measurements is assessed, showing that the combination of automatically extracted features and external variables is able to predict the particle that has produced the air shower with an accuracy of 98.87%. [ABSTRACT FROM AUTHOR]

The monitoring of physical activities and recognition of motion disorders belong to important diagnostical tools in neurology and rehabilitation. The goal of the present paper is in the contribution to this topic by (1) analysis of accelerometric signals recorded by wearable sensors located at specific body positions and by (2) implementation of deep learning methods to classify signal features. This paper uses the general methodology to analysis of accelerometric signals acquired during cycling at different routes followed by the global positioning system. The experimental dataset includes 850 observations that were recorded by a mobile device in the spine area (L3 verterbra) for cycling routes with the different slope. The proposed methodology includes the use of deep learning convolutional neural networks with five layers applied to signal values transformed into the frequency domain without specification of any signal features. The accuracy of discrimination between different motion patterns for the uphill and downhill cycling and recognition of 4 classes associated with different route slopes was 96.6% with the loss criterion of 0.275 for sigmoidal activation functions. These results were compared with those evaluated for selected sets of features estimated for each observation and classified by the support vector machine, Bayesian methods, and the two-layer neural network. The best cross-validation error of 0.361 was achieved for the two-layer neural network model with the sigmoidal and softmax transfer functions. Our methodology suggests that deep learning neural networks are efficient in the assessment of motion activities for automated data processing and have a wide range of applications, including rehabilitation, early diagnosis of neurological problems, and possible use in engineering as well. [ABSTRACT FROM AUTHOR]

In order to provide benchmark performance for Urdu text document classification, the contribution of this paper is manifold. First, it provides a publicly available benchmark dataset manually tagged against 6 classes. Second, it investigates the performance impact of traditional machine learning-based Urdu text document classification methodologies by embedding 10 filter-based feature selection algorithms which have been widely used for other languages. Third, for the very first time, it assesses the performance of various deep learning-based methodologies for Urdu text document classification. In this regard, for experimentation, we adapt 10 deep learning classification methodologies which have produced best performance figures for English text classification. Fourth, it also investigates the performance impact of transfer learning by utilizing Bidirectional Encoder Representations from Transformers approach for Urdu language. Fifth, it evaluates the integrity of a hybrid approach which combines traditional machine learning-based feature engineering and deep learning-based automated feature engineering. Experimental results show that feature selection approach named as normalized difference measure along with support vector machine outshines state-of-the-art performance on two closed source benchmark datasets CLE Urdu Digest 1000k, and CLE Urdu Digest 1Million with a significant margin of 32% and 13%, respectively. Across all three datasets, normalized difference measure outperforms other filter-based feature selection algorithms as it significantly uplifts the performance of all adopted machine learning, deep learning, and hybrid approaches. The source code and presented dataset are available at Github repository https://github.com/minixain/Urdu-Text-Classification. [ABSTRACT FROM AUTHOR]

Growth stage determination plays an important role in yield prediction and cereal husbandry decision-making. Conventionally, crop growth stage determination is performed manually by means of visual inspection. This paper investigates wheat and barley growth stage estimation by classification of proximal images using convolutional neural networks (ConvNets). A dataset consisting of 138,000 images captured prior to the crop canopy closure stage was acquired from 4 sites (7 different fields) in Ireland. The dataset includes images of 12 growth stages of wheat and 11 growth stages of barley captured for a number of crop varieties, seed rates and brightness levels. A camera was held at 2 m from the ground and two camera poses were used—downward-looking and declined to 45 ∘ below the horizon. Classification was carried out using three different machine learning approaches: (1) a 5-layer ConvNet model, including three convolutional layers, which was trained from scratch on our crop dataset; (2) transfer learning based on a VGG19 network pre-trained on ImageNet with an additional four fully connected layers, and (3) a support vector machine with conventional feature extraction. The classification accuracies of the aforementioned models were found to be (1) 91.1–94.2% for the ConvNet model, (2) 99.7–100% for the transfer learning model and (3) 63.6–65.1% for the SVM. For both crops, the best accuracy was obtained using the 45 ∘ camera pose and the transfer learning ConvNet model. For the growth stage classification task, the transfer learning ConvNet has the advantage of significantly reduced training time when compared with the built-from-scratch ConvNet model. [ABSTRACT FROM AUTHOR]

Alzheimer's disease is characterized by a progressive and irreversible cognitive deterioration. In a previous stage, the so-called Mild Cognitive Impairment or cognitive loss appears. Nevertheless, this previous stage does not seem sufficiently severe to interfere in independent abilities of daily life, so it is usually diagnosed inappropriately. Thus, its detection is a crucial challenge to be addressed by medical specialists. This paper presents a novel proposal for such early diagnosis based on automatic analysis of speech and disfluencies, and Deep Learning methodologies. The proposed tools could be useful for supporting Mild Cognitive Impairment diagnosis. The Deep Learning approach includes Convolutional Neural Networks and nonlinear multifeature modeling. Additionally, an automatic hybrid methodology is used in order to select the most relevant features by means of nonparametric Mann–Whitney U test and Support Vector Machine Attribute evaluation. [ABSTRACT FROM AUTHOR]

With the advances in computing powers and increasing volumes of data, deep learning's emergence has helped revitalize artificial intelligence research. There is a growing trend of applying deep learning techniques to image processing, speech recognition, self-driving cars, and even health-care. Recently, several deep learning models have been employed to detect a cyber threat such as network attack, malware infiltration, or phishing website; nevertheless, they suffer from not being explainable to security experts. Security experts not only do need to detect the incoming threat but also need to know the incorporating features that cause that particular security incident. To address this issue, in this paper, we propose a deep embedded neural network expert system (DeNNeS) that extracts refined rules from a trained deep neural network (DNN) architecture to substitute the knowledge base of an expert system. The knowledge base later is used to classify an unseen security incident and inform the final user of the corresponding rule that made that inference. We consider different rule extraction scenarios, and to prove the robustness of DeNNeS, we evaluate it on two cybersecurity datasets including UCI phishing websites dataset and Android malware dataset comprising more than 4000 Android APKs from several sources. The comparison results of DeNNeS with standalone DNN, JRip, and common machine learning algorithms show that DeNNeS with the retraining uncovered samples scenario outperforms other algorithms on both datasets. Furthermore, the extracted rules approximately reproduce the accuracy of the neural network from which they are derived. DeNNeS achieves an outstanding accuracy of 97.5 % and a negligible false positive rate of 1.8 % about 2.4 % higher and 3.5 % lower than the rule learner JRip on the phishing dataset. Moreover, DeNNeS outperforms random forest (RF), which produces the highest results among decision tree (DT), support vector machine, k-nearest neighbor, and Gaussian naive Bayes. Despite smaller training data in the malware dataset, DeNNeS achieves an accuracy of 95.8 % and F 1 s c o r e of 91.1 % , much higher than JRip and RF. [ABSTRACT FROM AUTHOR]

Human action recognition has received significant attention because of its wide applications in human–machine interaction, visual surveillance, and video indexing. Recent progress in deep learning algorithms and convolutional neural networks (CNNs) significantly affects the performance of many action recognition systems. However, CNNs are designed to learn features from 2D spatial images, while action videos are considered as 3D spatiotemporal signals. In addition, the complex structure of most deep networks and their dependency on backpropagation learning algorithm have a negative impact on the efficiency of real-time human action recognition systems. To avoid these limitations, we propose a new human action recognition method based on principal component analysis network (PCANet) which is a simple architecture of CNN exploiting unsupervised learning instead of the commonly used supervised algorithms. However, PCANet is originally designed to solve 2D image classification problems. In order to make it suitable for solving action recognition problem in videos, the temporal information of the input video is appropriately represented using motion energy templates. Multiple short-time motion energy image (ST-MEI) templates are computed to capture human motion information. The deep structure of PCANet helps to learn hierarchical local motion features from the input ST-MEI templates. The dimension of the feature vector learned from PCANet is reduced using whitening principal component analysis algorithm and finally fed to linear support vector machines classifier for classification. The proposed method is evaluated using three different benchmark datasets, namely KTH, Weizmann, and UCF sports action. Experimental results using leave-one-out strategy demonstrate the effectiveness of the proposed method compared with other state-of-the-art methods. [ABSTRACT FROM AUTHOR]

Person re-identification (ReID) is an imperative area of pedestrian analysis and has practical applications in visual surveillance. In the person ReID, the robust feature representation is a key issue because of inconsistent visual appearances of a person. Also, an exhaustive gallery search is required to find the target image against each probe image. To answer such issues, this manuscript presents a framework named features-based clustering and deep features in person ReID. The proposed framework initially extracts three types of handcrafted features on the input images including shape, color, and texture for feature representation. To acquire optimal features, a feature fusion and selection technique is applied to these handcrafted features. Afterward, to optimize the gallery search, features-based clustering is performed for splitting the whole gallery into k consensus clusters. For relationship learning of gallery features and related labels of the chosen clusters, radial basis kernel is employed. Later on, cluster-wise, images are selected and provided to the deep convolution neural network model to obtain deep features. Then, a cluster-wise feature vector is obtained by fusing the deep and handcrafted features. It follows the feature matching process where multi-class support vector machine is applied to choose the related cluster. Finally, to find accurate matching pair from the classified cluster(s) instead of the whole gallery search, a cross-bin histogram-based distance similarity measure is used. The recognition rate at rank 1 is attained as 46.82%, 48.12%, and 40.67% on selected datasets VIPeR, CUHK01, and iLIDS-VID, respectively. It confirms the proposed framework outperforms the existing ReID approaches. [ABSTRACT FROM AUTHOR]

This paper highlights the ability of convolutional neural networks (CNNs) at classifying EEG data listening to different kinds of music without the requirement for handcrafted features. Deep learning architectures presented in this paper include CNN of different depths and different convolutional kernels. Support vector machine (SVM) taking in EEG features describing the frequency spectrum, signal regularity, and cross-channel correlation has been applied for performance comparison with CNN. The best performing CNN model presented in this paper achieves the tenfold cross-validation (CV) binary classification average accuracy of 98.94% (validation) and 97.46% (test), and the tenfold CV three-class classification accuracy of 97.68% (validation) and 95.71% (test). In comparison, the SVM classifier achieves tenfold CV binary classification accuracy of 80.23% (validation). The CNN model presented is able to not only differentiate EEG of subjects listening to music from that of subjects without auditory input, but it is also capable of accurately differentiating the EEG of subjects listening to different music. In the context of designing neural computing models for EEG analysis, this paper shows that decomposing two-dimensional spatiotemporal convolutional kernels into separate one-dimensional spatial and one-dimensional temporal kernels significantly reduces the number of trainable parameters (size) of the model while retaining the classification performance. This finding is useful, especially in designing CNN for memory-critical embedded systems for EEG processing. In neurological aspect, auditory stimulus is found to have altered the EEG pattern of the frontal lobe and the left cerebral hemisphere more than the other brain regions. [ABSTRACT FROM AUTHOR]

The paper introduces a novel modality to efficiently tune the convolutional layers of a deep neural network (CNN) and an approach to also rank the importance of the involved hyperparameters. Evolutionary algorithms (EA) offer a flexible solution to this twofold issue, while the expensive simulations of the deep learner with the generated configurations are resolved by surrogate modelling. Three models have been used and evaluated as surrogates: random forests (RF), support vector machines (SVM) and Kriging. Sample convolutional configurations are generated by Latin hypercube sampling and have attached computed accuracy outcomes from real CNN runs. For the hyperparameter estimation task, the fitness of an individual from the EA associated with a surrogate model is subsequently derived from the CNN accuracy estimation on those variable values. With respect to the ranking and variable selection task, RF includes implicit variable selection, the SVM can be straightforwardly supported by a second EA, and Kriging offers a ranking based on the corresponding θ values. The estimated accuracy of the found hyperparameter values is compared with the true validation accuracy, and they are next used for the prediction on the test cases. The ranking of the variables for each of the three surrogate models is compared, and their influence is also revealed by response surface methodology. The experimental testing of the proposed EA–surrogate approaches is conducted on a real-world scenario of histopathological image interpretation in colorectal cancer diagnosis. [ABSTRACT FROM AUTHOR]

In this paper, we propose a real-time online shopper behavior analysis system consisting of two modules which simultaneously predicts the visitor's shopping intent and Web site abandonment likelihood. In the first module, we predict the purchasing intention of the visitor using aggregated pageview data kept track during the visit along with some session and user information. The extracted features are fed to random forest (RF), support vector machines (SVMs), and multilayer perceptron (MLP) classifiers as input. We use oversampling and feature selection preprocessing steps to improve the performance and scalability of the classifiers. The results show that MLP that is calculated using resilient backpropagation algorithm with weight backtracking produces significantly higher accuracy and F1 Score than RF and SVM. Another finding is that although clickstream data obtained from the navigation path followed during the online visit convey important information about the purchasing intention of the visitor, combining them with session information-based features that possess unique information about the purchasing interest improves the success rate of the system. In the second module, using only sequential clickstream data, we train a long short-term memory-based recurrent neural network that generates a sigmoid output showing the probability estimate of visitor's intention to leave the site without finalizing the transaction in a prediction horizon. The modules are used together to determine the visitors which have purchasing intention but are likely to leave the site in the prediction horizon and take actions accordingly to improve the Web site abandonment and purchase conversion rates. Our findings support the feasibility of accurate and scalable purchasing intention prediction for virtual shopping environment using clickstream and session information data. [ABSTRACT FROM AUTHOR]

Artificial intelligence methods have been applied to power system fault diagnosis based on the switch values such as relay protection actions and electrical component actions. These methods have their own problems including availability of data, state exponential explosion, difficulty modeling, etc. This paper deals with the application of stacked sparse autoencoder (SSAE) for power system line trip fault diagnosis based on the analog quantity of operation data. SSAE, a deep learning structure, is a more effective approach to solve problems mentioned above due to its network structure and layer-wise training mechanism. It also covers the detection of incipient faults based on the strong data mining capability. In the paper, an SSAE-based network with support vector machine (SVM) and principal component analysis (PCA) is proposed to improve the accuracy of fault diagnosis in power systems. The real-world simulation experiments prove the improvement and practical application of the proposed method. The process steps and parameter selection are elaborated in detail. [ABSTRACT FROM AUTHOR]

In this study we represent malware as opcode sequences and detect it using a deep belief network (DBN). Compared with traditional shallow neural networks, DBNs can use unlabeled data to pretrain a multi-layer generative model, which can better represent the characteristics of data samples. We compare the performance of DBNs with that of three baseline malware detection models, which use support vector machines, decision trees, and the k-nearest neighbor algorithm as classifiers. The experiments demonstrate that the DBN model provides more accurate detection than the baseline models. When additional unlabeled data are used for DBN pretraining, the DBNs perform better than the other detection models. We also use the DBNs as an autoencoder to extract the feature vectors of executables. The experiments indicate that the autoencoder can effectively model the underlying structure of input data and significantly reduce the dimensions of feature vectors. [ABSTRACT FROM AUTHOR]

Hand gesture for communication has proven effective for humans, and active research is ongoing in replicating the same success in computer vision systems. Human-computer interaction can be significantly improved from advances in systems that are capable of recognizing different hand gestures. In contrast to many earlier works, which consider the recognition of significantly differentiable hand gestures, and therefore often selecting a few gestures from the American Sign Language (ASL) for recognition, we propose applying deep learning to the problem of hand gesture recognition for the whole 24 hand gestures obtained from the Thomas Moeslund's gesture recognition database. We show that more biologically inspired and deep neural networks such as convolutional neural network and stacked denoising autoencoder are capable of learning the complex hand gesture classification task with lower error rates. The considered networks are trained and tested on data obtained from the above-mentioned public database; results comparison is then made against earlier works in which only small subsets of the ASL hand gestures are considered for recognition. [ABSTRACT FROM AUTHOR]