Search

Most insulation in the power transformers consists of oil and cellulose (paper/pressboard). Mineral oil has been used as an insulating liquid for over a hundred years. However, mineral oil is a non-biodegradable oil and has a serious impact on the environment and the non- renewable energy resources. Therefore, it is important to replace this oil with another alternative insulating liquid, and PFAE was introduced in the year 2006 in Japan. In addition, the ability of an insulating system also depends on the thickness of the monomolecular cellulose fiber on the kraft paper. Hence, it is necessary to determine the appropriate thickness of kraft paper for use as an insulator on the power transformer for this type of oil. This project presents paper-oil (PFAE) charging and discharging current characteristics. The study was carried out through the implementation of experiments in the laboratory. Polarization and Depolarization Current (PDC) test has been carried out to test the conductivity and response function of the insulating materials. It was found that the thickness of kraft paper affecting the capacitance value and the PDC measurement of the tested model. In both dry and wet conditions, the sample of two layers of kraft insulation paper shows the best insulation status as it has the lowest conductivity compared to the other samples. It can be concluded that it is necessary to determine the ideal thickness of insulating paper to ensure that the power transformer has the best insulation.

The purpose of this short position paper is to stimulate a community wide discussion on how situation awareness (SA) technologies may best support the decisions made during pandemics. Currently much of SA is directed towards the critical information needed to make epidemiological and public health decisions; as important as this is, an epidemic also requires doctors and front line medical professional to make numerous decisions regarding individualized patient care. The question is how can SA contribute to both the rapid collection, analysis, and sharing of critical patient information among medical personnel, and then across all levels of decision-making from individual patients to public health policy?

Drowsiness can be dangerous when performing tasks that require constant attention, such as driving a vehicle. Sleepiness is correlated with a variety of physiological variables, such as eye closing, head movements, pulse rate, eye twitch rate, etc. Also, the yawn can be considered as an accurate indicator of drowsiness and fatigue. Yawning detection is very important for the safety purpose of drivers as it will let the driver know if he/she is getting drowsy. Driving at that moment may not be safe. Several automatic yawning detection techniques have been developed for driver's drowsiness monitoring system. Nevertheless, correctly detecting the yawning of the driver and predicting exhaustion in real-time situations is still a crucial challenge. In this paper, we will be reviewing various existing machine learning approaches for driver's yawning detection. In previous approaches, various classical machine learning algorithms such as viola-Jones, contour activation algorithm and SVM have been used for yawning detection, but these approaches failed to predict yawning in realtime situations. Using Deep learning techniques, we can make a real-time yawn detection system with high accuracy. We find that some precious Deep learning algorithms like CNN, RNN, LSTM, Bi-LSTM can detect the patterns with high accuracy. After the comparison of various algorithms and techniques, we find that with the help of Deep learning algorithms the yawning can be detected in real time with high accuracy.

The aim of this work is to classify the dementia stage using microstructural white matter (WM) diffusion kurtosis indices. We develop various class imbalance machine learning methods in classifying the cognitive normals (CN), Alzheimer's disease (AD) and mild cognitive impairment (MCI) individuals. Diffusion-weighted images of 155 subjects were collected with 44 CN, 88 MCI and 23 AD individuals aged between 60 to 96 years. We first perform skull striping using FSL tool along with head motion and artifact correction. We calculate the mean diffusivity, fractional anisotropy from various brain regions of WM indices from diffusion kurtosis imaging (DKI) using tract based spatial statistics. Then we estimate the kurtosis fractional anisotropy of various WM regions. After estimation of all the regional KFA indices, we trained these indices using various class imbalance machine learning classification algorithms, as the sample size is different in all the three groups. We found that balanced random forest classifier was the best classifier with an accuracy of 74% when compared to other methods for 5-fold cross-validation. We conclude that class imbalance machine learning methods are potential in classifying cognitive normals, AD and cognitive impairment individuals.

This research aims to study the aging behavior of mineral oil, Gas-to-Liquid, and natural ester oil on oil paper insulation. Experiments were carried out on oil insulation, which contained paper insulation, namely kraft paper (TUK), kraft paper (KP), pressboard (PB). Then the oil-paper insulation was placed in a sealed glass bottle & accelerated thermal aging for 1344 hours at a temperature of 150 °C. Observation of oil behavior was carried out at 336 hours intervals during aging. The aging behavior of the oil was observed by testing the oil moisture content, total acid number, breakdown voltage, color scale, interfacial tension, flash point, and Fourier Transform Infrared (FTIR). It was found that natural ester oil had better electrical performance than mineral oil and GTL. Meanwhile, natural ester oils due to their polar nature have different trends in both moisture and acid content when compared to hydrocarbon-based oils. It can be observed there is a relationship between moisture content, acidity, and breakdown voltage in all oils. The color scale shows changes with aging of the three oils, as well as the interfacial tension. Meanwhile, flash point shows the superiority of natural ester oil and GTL has a greater value than mineral oil. FTIR shows the occurrence of transesterification in natural ester oil in the 1700-1800 cm-1 spectrum where in mineral oil and GTL this is not found.

In this paper three different types of architectures have been trained and tested to determine which one of them is more efficient in detecting exudates from retinal images. The database that was used is SUSTech_SYSU from which 100 images were extracted for image testing and another 100 images that were used in obtaining the training dataset. The three architectures that were studied in this paper are SqueezeNet, ResNet50 and GoogLeNet. The results showed that from the three architectures the most efficient in exudates detection was ResNet50 with an average accuracy of 0.928, followed by SqueezeNet with 0.910 and finally, GoogLeNet for which the average accuracy was 0.907.

Pneumonia is one of the most common infectious diseases in clinic. X-ray chest is an important basis for early diagnosis of pneumonia. With the development of computer vision technology, using convolutional neural network to train pneumonia image classification model has been gradually applied to the process of medical clinical diagnosis. However, there are many problems in the process of using convolutional neural network to train pneumonia image classification model, such as too long model training time, over fitting and low accuracy due to too small training dataset. To solve these problems, this paper proposes an optimization method of pneumonia image classification model based on transfer learning and feature fusion, which is called Transfer Fusion. The Transfer Fusion optimization method will transplant the trained source model parameters to the target model, and add a specific feature fusion classification layer, so as to significantly shorten the training time of the new model, improve the accuracy and prevent over fitting. In this paper, Transfer Fusion optimization method is applied to three common convolutional neural network models: Google InceptionNetV3, MobileNetV2 and ResNet50. Through a large number of experiments, the performance of the three models has been significantly improved and improved.

Every year, doctors diagnose skin cancer in around 3 million or more patients across the globe. Currently, it is one of the most widely recognized kinds of cancers for human health. Hence, we need an early diagnosis to prevail any critical condition of the infected patients. Apparently, it can treat with topical drugs, if it diagnoses in an early stage. Hence as an outcome, skin cancer is responsible for less than 1% of all cancer deaths. There are two types of tumors in the skin cancer diseases domain, such as benign and malignant. To develop a robust and early screening system to diagnose skin cancer, it requires an efficient algorithm for prediction, trained with a large dataset. The primary aim of this research is to develop an efficient skin cancer screening process using a robust deep neural network with a large dataset. In this paper, we intend to determine considerate and dangerous types of skin cancer tumors using dermoscopic images from a publicly available dataset. We proposed an efficient and fast scaled 2D-CNN based on EfficientNet-B7 deep neural architecture with image preprocessing. This paper also uses two different pre-trained deep neural architectures, such as VGG19, and ResNet-50 to compare the performance with the proposed architecture. The proposed architecture outperformed the other pre-trained CNN models whereas the proposed architecture achieved higher AUC and accuracy compared to other architectures.

This paper presents a novel transformerless common-ground AC/DC switching converter with integral fault protection utilizing Active Isolation (AI) techniques. The power converter employs a standard buck-boost topology but isolates the converter by the addition of lateral active switch devices on the low-side power path. Compared to transformer isolation, this active isolated converter has reduced size and cost with increased efficiency. The paper includes a detailed discussion of circuit operating principles, modeling, fault detection, and experimental results. A 300W, $120 V_{ac}-to-19 V_{dc}$ converter is prototyped and experimental measurements are correlated to calculated results. The proposed fault protection solution is found to properly protect the converter in the event of a switch short failure on and switch device.

The description of gene intonation can be exploited in the determination of acute disorder like carcinoma. The picking of biological genes is substantial as well as vital for cancer observation. This paper offers an outline suggesting two phases of gene selection for colon data categorization. Colorectal tumor positions third among extensive cancer and foremost reason of cancer demise globally. There is a necessity to realize exact process to recognize the tissues and to improve the diagnostic method for discovering medicines. Freely accessible binary class colon data that embraces 2,000 genes and 62 trials is used in this research. The two-phase feature selection includes Analysis of Variance test as first phase and in second phase, the Hilbert transform and Firefly algorithm are applied to get pertinent genes. Then finally, the obtained colon data are classified as normal and tumor classes using Linear Regression, Logistic Regression, Naive Bayes and Harmonic search classifiers. The results indicate that the Harmonic search classifier provides a greater accuracy of 95.06% for Hilbert transformed colon tumor features and an accuracy of 90.37% for Hilbert transformed colon normal features. For Firefly clusters, the naive bayes classifier attains a higher accuracy of 93.36% and 91.93% respectively for colon tumor and normal data. It is deduced that the performance of two-phase gene selection method with Harmonic search and Naive bayes classifier has accomplished a better accuracy in contrast with the other classifiers employed in this paper.

Prostate cancer is a common type and it ranks second among men. Certain kinds of prostate cancers remain lethargic and don't lead to cancer death, even though lacking treatment. Though, a substantial amount of people with this cancer has violent lumps that advances quickly to neoplasm disease and remain very risky and hence they should be accurately classified. Presently, classification results for prostate patients are directed by numerous machine learning algorithms. The foremost goal of this research is to classify the prostate cancer in a precise way. Publicly available two class prostate data that contains 12,600 genes and 136 samples is considered in this paper. Initially Kruskal-Wallis test along with Principal Component Analysis (PCA) is applied to discover the most informative genes. Then those genes are classified using Linear Discriminant Analysis (LDA), Support Vector Machine (SVM), eXtreme Gradient Boosting (XGB) and bagging ensemble KNN classifiers for classifying prostate patients as either normal or abnormal. The results indicate that a maximum overall accuracy of 96.46% is achieved for abnormal prostate data samples and 95.78% for normal prostate data samples using XGB classifier when compared with other types of classifiers considered in this paper.

This paper suggests the concepts of "facial recognition of physical challenges". The greatest difficulty for blind people is seeing obstacles behind and outside. In many cases they rely on other people for their blindness .therefore they need the support of someone else (people).In their daily life they are facing with many problems if they want to move starting with one spot then onto the next place, they need guidance help them. Also they need to see the person in front of them and hence to avoid these situations this research work proposes a face recognition system for people with visually challenged by using raspberry pi. In this case, there is a face recognition camera and voice recognition so these components can be combined into one and it is easy to use and easy to behave by people facing visually challenge. In this will help people to remove obstacles and in front of them and around them .so they do not make mistake of hitting the obstacles. This paper will explain the design and features involving the surroundings and surrounding objects.

This research paper presents the importance of cardiologists and their availability to the public by the usage of various smart devices. These new simplified ECGs may be as complex or sophisticated as those that represent the usage of the medical facility however they are accurately able to observe the heart rate to monitor their health. One of the most commonly overlooked problems in the modern world is Arrhythmia. This is a problem related to the rhythm of the heartbeat and it occurs due to the improper coordination between the electrical impulses with your heartbeat. Heartbeat occurs daily in the lives of several people as they aren’t properly focused upon the monitoring of heartbeat. This research paper attempts to focus on the issue of Heart Arrhythmia as well as the creation and classification model which is capable of identifying the type of heart arrhythmia by an individual how may be suffering from heart issues. To successfully create our model, this research work utilizes the convolution neural networks to train the proposed model with existing ECG data and properly identify the heart rate as well as the type of arrhythmia that they have so that it provides the ability to immediately provide the proper medical attention.

Breast cancer is becoming one of the leading causes of mortality in women all over the world. The advanced engineering of Artificial Intelligence techniques and natural image classification approaches is mostly employed to classify breast images. The image classification permits the physicians and the doctor the next chance, and this, in turn, saves the time of the physician and the doctor. Apart from several publications on classifying breast images, only some of the review papers are present that offer a clear definition of the “approaches to breast cancer image classification, feature extraction and selection methods, classification measurement parameterizations, and image classification results” with the help of 3D images. Machine learning (ML) seems to be an important portion of medical imaging research. These are developed over the years from manual seeded inputs to automatic initializations. The enhancements in the area of ML lead to various self-reliant and intelligent "Computer-Aided Diagnosis (CAD) systems" since the learning capability of the ML techniques is increasing in a constant manner. Several automated techniques are developing with deep feature representations and learning. The modern enhancements of ML with extensive and deeper representation techniques called Deep Learning (DL) techniques have created an important effect on enhancing the diagnosing ability of these systems. The main objective is to undergo a critical review on diverse contributions related to breast cancer detection that could be highly advantageous for dealing with 3D mammogram images. This paper plans to make a review on different algorithmic categorizations, datasets used, simulation platforms, and performance metrics. Furthermore, the research gaps and challenges of the breast cancer diagnosis using 3D images are elaborated, encouraging the upcoming researchers to give tremendous concern on new contributions.

Epilepsy is a central nervous system disease that affects loss of consciousness and convulsions. The testing of electrical activity in the brain is identified as EEG. Ability to predict the onset of epileptic seizures before their commence seems to be very useful for medication-assisted seizure protection. Electroencephalogram (EEG) signals are now used to predict epileptic seizures using machine learning techniques and statistical analysis. Pre- processing of EEG signals for dynamic range and extra functionality, but in the other hand, extraction are two real issues that have a negative effect on the system’s run time and correctness. Therefore, this paper proposes a model that provides a reliable method for both pre-processing and classification. It is implemented using FIR filter for preprocessing and machine learning classification methods. Among the variety of machine learning algorithms, this paper uses RF algorithm and BPNN for classification. After the classification, the results were compared, and it is revealed that Random Forest classifier gives 95% accuracy which is more efficient than Neural network via back propagation.

After a short history about the beginnings of the railway electric traction systems, in the paper are presented the elements that have an interest in being monitored for the railway electric traction power system in single-phase alternating current with the frequency of 50 Hz and voltage 25 kV, specific to the railway electric traction in Romania. The return traction current at the source is assure by the return circuit that incorporates the impedance bonds, called joint coils in Romania. Also, the paper highlights the benefit of monitoring the impedance bonds of the railway electric traction power supply system. The benefits of monitoring the joint coils from electric traction consist of currents real-time surveillance through the impedance bond’s primary windings, by knowing the temperatures of: terminals, the connection wires, on the impedance bond case and the ambient, respectively of the oil level (for classic impedance bonds). Knowing the technical condition of the impedance bond helps to make preventive maintenance work more efficient, reduce downtime and optimize the use of maintenance resources.

Micro Expressions are those involuntary muscular movements of the facial muscles produced in response to a stimulus. They are short-lived expressions that last for anywhere between 0.04 to 0.2 seconds and are extremely subtle in their amplitude. Given their fleeting and elusive nature, it becomes almost impossible to detect these expressions through the naked eye. Recent developments in Deep Learning models have shown great success in efficiently identifying and analyzing Micro Expressions. In this paper, various models have been implemented on the SAMM dataset. The models studied are namely– VGG16, ResNet50, MobileNet, InceptionV3, and Xception. The experimental results have helped us carefully analyze the various metrics related to the models and compare them with each other to ascertain which one outperformed the others and is best suited for real-world applications. The MobileNet model has surpassed all other models in terms of its efficiency with respect to the domain of this paper. It has been able to describe and understand all the information that can be found in the various Micro Expressions.

The dominant causes of visual impairment worldwide are Cataract, Glaucoma, and retinal diseases among patients. The alarming cases of these diseases call for an urgent intervention by early diagnosis. The proposed system is designed and developed to easily facilitate the detection of cataract, glaucoma and retinal diseases among patients. The Logistic Regression, Random Forest, Gradient Boosting and Support Vector Machine algorithms are used for detection. The proposed system will help people to get the proper treatment of the aforementioned diseases at an early stage thus reducing the percentage of blindness being caused. The proposed system evaluates the effectiveness and safety of cataract surgery in eyes with age-related degeneration along with glaucoma and retinal diseases detection. This paper shows the accuracy of algorithms and SVM classifiers based upon the glaucoma, retina, cataract and normal eye’s fundus images. The idea of classifying the images based on its fundus and extracting features is widely known now-a-days and also it plays a vital role in the final outcome. This paper talks about the multiclass built models of these classifiers and on the basis of the ROC curves plotted it predicts the output of the images. As far as the algorithms are concerned, the efficiency of algorithms helps it stand best out of many and in our case Gradient boosting proves to give best results for the eye with cataract with 90% accuracy. Then the supervised algorithms logistic regression and random forest gives the accuracy of 89% and 86% respectively.

Chest radiograph (CXR) interpretation in pediatric patients is error-prone and requires a high level of understanding of radiologic expertise. Recently, deep convolutional neural networks (D-CNNs) have shown remarkable performance in interpreting CXR in adults. However, there is a lack of evidence indicating that D-CNNs can recognize accurately multiple lung pathologies from pediatric CXR scans. In particular, the development of diagnostic models for the detection of pediatric chest diseases faces significant challenges such as (i) lack of physician-annotated datasets and (ii) class imbalance problems. In this paper, we retrospectively collect a large dataset of 5,017 pediatric CXR scans, for which each is manually labeled by an experienced radiologist for the presence of 10 common pathologies. A D-CNN model is then trained on 3,550 annotated scans to classify multiple pediatric lung pathologies automatically. To address the high-class imbalance issue, we propose to modify and apply "Distribution-Balanced loss" for training D-CNNs which reshapes the standard Binary-Cross Entropy loss (BCE) to efficiently learn harder samples by down-weighting the loss assigned to the majority classes. On an independent test set of 777 studies, the proposed approach yields an area under the receiver operating characteristic (AUC) of 0.709 (95% CI, 0.690-0.729). The sensitivity, specificity, and F1-score at the cutoff value are 0.722 (0.694-0.750), 0.579 (0.563-0.595), and 0.389 (0.373-0.405), respectively. These results significantly outperform previous state-of-the-art methods on most of the target diseases. Moreover, our ablation studies validate the effectiveness of the proposed loss function compared to other standard losses, e.g., BCE and Focal Loss, for this learning task. Overall, we demonstrate the potential of D-CNNs in interpreting pediatric CXRs., Comment: This is a preprint of our paper which was accepted for publication to ICCV Workshop 2021

Medical data mining is known as methods of extracting data from human services databases to help clinicians get the best diagnosis. In this field, and among cancer diseases, breast cancer is the highest deadly disease in the world in recent years. Therefore, data mining techniques will be the largest part used in this study. To complement the previous research related to breast cancer detection, this paper proposed a model to help solve the difficulty of determining the degree of risk of the disease and obtain the best results. Aiming of reducing the costs and time used in diagnosing the disease too. The experiment used a dataset computed from a digitized fine needle aspirate (FNA) image of a breast mass dataset; diagnostic (WBCD) available in UCI machine learning repository. The model is the application of classification techniques to the breast cancer data collected, which, in turn, predicts the severity of a patient's breast cancer. In addition, this paper classifies and diagnose cancers using Deep-learning algorithms only, hybrid Machine learning algorithms and feature selection methods (applied correlative rules to find out what traits related to breast cancer severity).

Liver diseases like fatty liver disease, chronic active hepatitis, and cirrhosis are the major cause of mortality in India. Alcohol consumption, inhalation of harmful toxic gases, improper consumption of contaminated pickles, drugs, and foods are the major cause of diseases in the liver. Diagnosis of liver disease needs high accuracy and precise results for predicting whether a person is suffering from liver disease or not. Major disastrous repercussions can be the result of minor errors in the diagnosis of liver diseases. The major goal of this paper is for the detection of liver disease at right time and helping the doctors and combating the increasing number of cases. In this paper, we implemented various machine learning techniques like logistic regression, KNN, XG-Boost, SVM, Gaussian NB, Random forest, Decision tree, Gradient Boosting, CatBoost, AdaBoost, and LightGBM on selected features from the dataset for predicting liver disease and it was found that Random Forest performed best among all the technique and gained high accuracy and performed outstandingly in all metric evaluations.

Lungs of people are susceptible to developing cancer; according to WHO, cancer is a leading cause of death worldwide, accounting for nearly 10 million deaths in 2020, and lung cancer alone has caused 1.8 million deaths. To tackle the lung cancer, this paper proposes a 3D U-Net model, which can be utilized to diagnose cancer on CT scans. In the preprocessing step, it converts lung nodule CT scans in DICOM format to PNG format for the purpose of training the model. After that, this study employs 3-dimensional U-Net model, segmentation algorithm, to detect all possible areas of lung nodules on the CT scan. Since lung nodules might be misclassified, this paper introduces a convolutional neural network to classify real and fake lung nodules. The experimental results demonstrate that ACC reached 93.8%. Furthermore, it has been noted that AUC has 0.9358, which means the model has good classification ability for positive and negative cases.

Cancer is a disease that attacks several organs in the human body and rapidly spreads to other parts of the body. For the last few decades, cancer has become the most vulnerable illness on the planet, with the highest mortality rate due to late diagnosis. Since the 1960s, researchers have been studying computational methods for learning from cancer data, with number of advanced techniques directly applied on medical images. Depending on the organ, cancer can manifest in a number of ways. Different treatment processes can be used to remove the contaminated organ, prevent its spread, and save the person's life when these symptoms are considered. As a result, early detection of cancer has emerged as one of the most pressing issues in medical science and technology. This paper presents a review of recent work on the methods of computational intelligence for prostate cancer prediction modelling, as well as the issues that must be addressed. The paper focuses on a broad concept of computational intelligence that includes Artificial Neural Networks and Deep Learning, as well as a survey of various deep learning application areas for prostate cancer diagnosis and detection. Initially we will start with an overview of architectural principles and the use of various common deep learning technologies such as Convolutional Neural Networks (CNN), Deep Belief Networks (DBN), and auto-encoders (AE). With that, we'll move on to the study of these various algorithms for early cancer detection using technology.

Electric cars are evolving into a clean and stable environment. It has lower maintenance costs than fossil fuels, supports emerging mobility technologies and emits less air pollution. When it is combined with renewable sources of energy such as wind power, it can reduce the carbondioxide(CO2) emissions from both the sector transport and power generation segments. The global adoption of electric vehicles is a challenge for the energy sector like as distribution companies and consumers. It is not performed properly. The increased of electric vehicles will adversely affect the load-bearing network, increased demand and reduced voltage stability. By combining the EV with the smart grid supplemented by supporting technologies such as bidirectional converter. In this paper Vehicle-to-grid(V2G) and renewable energy source integration technology is used. This type of technology is capable of adapting distributed loads (DG), energy storage, and extremely fluctuating load requirements. This paper analyzes the importance of EV and renewable sources of energy into standard 13 nodes system. Combining EV and renewable energy sources into the 13 nodes system is analyzed different scenarios for stability using MATLAB / Simulink model.

Epilepsy is a common neurological disease at present. It is caused by abnormal discharge of brain neurons. The clinical manifestations are whole body convulsions, loss of consciousness and stopping breathing. By using electroencephalogram (EEG) to capture changes in brain electrical activity, many studies use EEG signals for epileptic seizure prediction. This paper proposes to predict epileptic seizure EEG signals based on a Gated Recurrent Unit (GRU) prediction model. First, the original data set is passed through a 50HZ notch filter to filter out the noise and artifacts that may be included in the data collection, and then it is sent Enter the GRU prediction model for prediction. This paper uses both LSTM and random forest model algorithms. Experiments show that the GRU model algorithm has the shortest training time and the accuracy rate is 12.48% higher than that of random forest, which proves that the method is feasible.

The paper considers passive magnetic bearings for small and medium-sized rotating rotors. It proposes several schemes of a magnetic support made using two permanent annular magnets with the axial magnetization. A distinctive feature of these magnetic bearing schemes is the possibility of short-term changes in the stiffness parameters. This is achieved through the use of bias windings. The technique and results of computational studies to determine the dependencies of the restoring magnetic forces and stiffnesses on the displacement of the rotor in the radial direction are presented. The validation of the calculation results was carried out according to the experimental data for an option of a controlled radial passive magnetic bearing used in a test rig. This rig implements a full rotor magnetic suspension, includes two radial passive magnetic bearings and one axial active magnetic bearing and is a prototype for small and medium-sized turbomachines. The paper also presents a comparative analysis of various design schemes of passive magnetic bearings with controlled stiffness.

Electroencephalography (EEG) signals are normally susceptible to various artifacts and noises from different sources. In this paper, firstly the existence of artifacts will be identified on the recorded EEG signals and then the origin of the detected artifact will be determined among 7 different sources. Due to the nature of EEG signals, almost no specialist can determine artifact source through eye inspection. This paper introduces the utilization of 1-D Convolutional Neural Network (CNN) in multi-class EEG artifact classification. Proposed CNN models were kept as simple as possible to have the best operation time but in the meantime, models were selected adequately deep to extract appropriate artifact features from applied EEG signals. Obtained results prove that proposed architectures are able to classify artifacts with high accuracy.

With bonding wire converted from Au to Cu, corroded Cu wedge defects observed with some packages. Epoxy Mold Compound (EMC) to lead frame delamination provide penetration path for corrosive chemicals to penetrate inside and react with Cu wedge. To improve delamination performance without changing exist production and Bill of Materials, study was carried out with molding process, Lead Frame strip's structure, EMC properties. Delamination performances were collected via C-SCAN. In depth study on delamination mapping vs mold flow influence been carried out through detail short shot analysis. Mold parameter influence on delamination was investigated [1]. EMC flowability vs lead frame surface wetting behavior was explored. Study result shows different LF strip location design, mold process optimization can influence delamination [2]. In this paper, wedge pad surface cleanness was checked via mechanical decapsulation on package, EMC viscosity curve were analyzed to explain mold parameter settings. These all helped well explained mechanism on support delamination improvement. This paper gives insight on material&process optimization approach to achieve minimum package delamination. Involves discussion on mold process factors, EMC material property that influence package delamination.

Our limited understanding of the nervous system forms a bottleneck which impedes the effective treatment of neurological disorders. In order to improve patient outcomes it is highly desirable to interact with the nervous tissue at the resolution of individual cells. As neurons number in the billions and transmit signals electrically, high-density, cellular-resolution microelectrode arrays will be a useful tool for both treatment and research.This paper investigates the advantages and versatility of laser-patterning technologies for the development of such high-density microelectrode arrays in flexible polymer substrates. In particular, it aims to elucidate the mechanisms involved in laser patterning of thin polymers on top of thin metal layers. For this comparative study, a pulsed picosecond laser (Schmoll Picodrill) with two separate wavelengths (1064 nm (infrared (IR)) and 355 nm (ultraviolet (UV))) was used. A 5 $\mu$ m thick electroplated layer of gold (Au) was used to form the microelectrodes. Laser-patterning was investigated to expose the Au electrodes when encapsulated by two different thermoplastic polymers: thermoplastic polyurethane (TPU), and Parylene-C, with thicknesses of maximum 25 $\mu$ m. The electrode diameter and the distance between electrodes were reduced down to 35 $\mu$ m and 30 $\mu$ m, respectively. The structures were evaluated using optical microscopy and white light interferometry and the results indicated that both laser wavelengths can be successfully used to create high-density microelectrode arrays in polymer substrates. However, due to the lower absorption coefficient of metals in the IR spectrum, a higher uniformity of the exposed Au layer was observed when IR-based lasers were used. This paper provides more insight into the mechanisms involved in laser-patterning of thin film polymers and demonstrates that it can be a reliable and cost-effective method for the rapid prototyping of thin-film neural interfaces.

The paper presents a concept of an air pollution monitoring system based on a wireless sensor network, in which particular sensors are able to predict future pollution levels within a specific time horizon. In the solutions reported in state-of-the art literature, forecasting the changes in pollution levels is usually based on the use of artificial neural networks (ANNs), globally trained for a given area, e.g. a city. In contrast, we propose to equip the sensors with hardware-implemented ANN s that could make the prediction based on data that the network obtains from the place where particular sensor is installed. Such networks require appropriate hardware optimization to maintain low energy consumption of the overall sensor. For this reason, in this paper we present a method of optimizing one of more complex components that is involved in determining the value of neuron weights in the network training process. In this work, we focus on the backpropagation (BP) learning algorithm, which results from the widespread use of this type of network in predicting air pollution levels.

Low Intensity Pulse Ultrasound (LIPUS) is widely used to treat various diseases because of its non-invasiveness and non-ionizing radiation. However, few studies have analyzed the specific mechanism and optimal parameters of LIPUS in treating the Achilles tendon injury. This paper investigated the LIPUS topography and the thermal properties with different parameters through simulation for ultrasound treatment of the Achilles tendon. First, a four-layer model of bone tissue, tendon, muscle, and soft tissue was established. Then, the finite element method was used to simulate the sound intensity distribution and thermal effects of ultrasound. Finally, the influence of ultrasonic parameters such as frequency, initial intensities, and transducer diameters was discussed. Experimental simulation results show that the frequency of 3 MHz may be a recommended choice to treat Achilles tendon disease. This paper theoretically explores the underlying physical mechanism of ultrasound treatment of Achilles tendinopathy, which has guiding significance for in vivo and in vitro experiments.

This paper deals with the task of classification of accounts registered in e-commerce systems. The authors consider solving this problem with different machine learning algorithms and LSTM recurrent neural network. A brief description of machine learning algorithms and LSTM neural network used during the research is given in the paper. The dataset and algorithm of its preprocessing are described, additional features are introduced. The paper presents the results of numerical experiments: confusion matrices and classification metrics received with considered algorithms. Comparing of the models has been conducted on the basis of received metrics. The algorithm that provides the best metrics is selected. The SMOTE and ADASYN resampling algorithms are applied to the dataset, the received classification metrics are provided. The methods of improving classification results are proposed.

High mortality rate among women in India is mainly due to breast cancer. As mammography is a standard imaging tool used in screening the breast, it sometimes tends to miss certain abnormalities that lie beneath the breast regions. Thermography being a temperature-based functional imaging modality has enough potential to identify the abnormal changes in the breast through asymmetries at an early stage. This paper considers breast thermograms of patients that are normal and abnormal wherein pre-processing is applied to extract the ROI. The pre-processing includes breast region segmentation using horizontal and vertical projection profile approach along with the Otsu method. The features are then extracted from this ROI for which they are also fed as input to the non-linear classifiers for classification of normal and abnormal. The outcome of the paper results in a good accuracy rate that can be efficiently utilized as a Computer-Aided Detection tool for a second opinion in the breast cancer diagnosis.

With the ascent of data innovation, and it has proceeded with coming into the human services division, the instances of diabetes just as their side effects are reported. Examination in the distinction of regression strategies depends on specific parameters discovering answers to analyze the illness by investigating the examples found in the information through regression models. The paper throws light on the diabetic records of pregnant women. In this paper, linear regression and logistic regression calculations have been utilized on a pre-existential dataset to anticipate whether diabetes is recorded or not in a patient. Results from both the calculations have been analysed and introduced.

There has been tremendous distress over the globe since 2019 due to the novel Corona Virus. The COVID-19 disease has caused the medical services a great deal of pressure and a medical emergency all over the world. Very quick dissemination of coronavirus has muzzled as well as endangered human civilization. Various processes have brought attention to the medical propensity for the diagnosis of Corona Virus disease. Antigen and Reverse Transcription Polymerase Chain Reaction (RT-PCR) tests have been used since the brisk emergence of COVID-19. The emanation of various mutants developed in these two years has put some questions on the precision and accurate result of these methods especially during the second wave of the pervasive coronavirus. So, various radiological methods like Computed Tomography (CT scan) and X-ray, etc. have been utilized for the spot because of the more meticulous result. The investigatory study in the given paper is about the application of Artificial Intelligence in speculating the coronavirus disease using a suitable model. The research carried out in the given paper gave an accuracy of 99%. Various machine learning-based classifiers are being adapted for the prediction of the disease.

Cervical Cancer (CC) is the fourth most frequent disease among females. According to a 2020 report from WHO, approximately 6,04,127 females identified with cervical cancer worldwide and about 3,41,831 females estimated under the death cases from this cancer. The main purpose behind this paper aims to provide a review and comprehensive overview of the automatic segmentation and classification of DL algorithms, and also offers a description of their architectures in the cervical cancer image analysis from Pap smear images. The research articles are considered from publically available scholarly sites such as Google Scholar, IEEE, Science Direct, PubMed, Springer, and Elsevier on cervical cancer medical image analysis. Only those papers are included that are related to the detection of cervical cancer with DL algorithms. Each review articles are considered using three sets namely: Set1- Segmentation, classification, and cervical Cancer; Set2-Medical imaging, Machine learning, and pap-smear, and Set3- Pap Smear images, Automated System, and Classification. Most of the existing techniques achieve an accuracy of 97.89% on publicly available pap smear datasets (Herlev Dataset). In the conclusion, research gaps and limitations are mentioned that reduced the classification accuracy. For reducing the limitation, also discussed the recent advance and future perspective concerning cervical cancer detection.

The tissues that are existing inside the skull becomes spontaneous called Brain tumor, and this tumor has gotten one of the significant worries in the medical care area prompting the passing of individual explicitly kids and men. This paper fundamentally outlines those processes that are vital to examine a patient's condition to ensure that brain tumors can be perceived as soon as possible alongside the certainty that the odds of errors are diminished as much as possible. The requirement for exactness and better determination can likewise assist tolerance with beating certain issues, for example, diminishing in cash, the danger of various types, for example, early demise, and so forth. In this paper, there is an explanation about the collection of datasets, improvement in the quality of image, image separation, feature mining, feature reduction as well as selection., detection of cancer cells, and performance estimation methodologies. This paper additionally gives an outline of the technique or methods needed to deal with the MRI (Magnetic Resonance Images) brain images in those above steps of processes.

Unawareness of a disease can be very dangerous and when it comes to an abnormality in blood pressure, it can be life-threatening. Hypertension or blood pressure higher than normal that is 140 /90 can give birth to diseases like cardiac arrest, dementia, or even deaths. Although, the good news is that abnormal blood pressure can be treated if predicted before it's too late. Managing hypertension is easier with medications and treatment if detected timely. This paper focuses on the prediction of the likeliness of abnormality in blood pressure in people in the future. In simpler terms, it predicts the chances of people prone to develop hypertension or hypotension. The prediction would be based on 13 different attributes like smoking habits, genetic history, Body Mass Index, salt content in diet and so on. This paper can provide medical foresight using classification algorithms that can help people to understand the factors that increase blood pressure. Therefore, by making certain possible amendments in their lifestyle like salt content in food, alcohol consumption, etc. people can keep their blood pressure in range. The paper provides insight and comparison of several supervised learning algorithms and gives an overall summary of the existing work.

The respiratory sound classification paper focuses on analyzing sounds collected through the stethoscope at various chest positions to understand critical concerns of human breathing. Most doctors used a stethoscope for auscultation of lung sounds, along with heart sound, as a basic checkup and utilized later in predicting chest pathology. Using deep learning-based sound classification can build intelligent digital stethoscopes or understand overall health conditions later used along with other Lung organ analyses. The solution enables affordable and practical tools for remote parts of India, where doctors or specialists are not keen to work. Such solutions can be used as first analysis before further investigation by traveling to far-off specialist hospitals. The lung sound was prepared and hosted by a team from the University of Coimbra, Portugal; the University de Aveiro, Portugal; and the Aristotle University of Thessaloniki, Greece and the collaboration at Kaggle website lung sounds of patients who are Normal lung functioning and with some anomaly collected. The collected lung sounds are labeled as Normal, seven classes - Abnormal. Overall, eight levels are considered to classify using a deep learning model. The paper proposes a multipath convolution neural network (CNN) model for analyzing sound after exploring all approaches, like Recurrent Neural network (RNN), Machine learning models like SVM, and other CNN models as per paper [3]. Lung sound is processed using signal processing methods and converts wav files to Mel-frequency cepstral coefficients (MFCCs), Mel spectrogram, and chromatogram from a waveform or power spectrogram. Instead of using one feature among three, the paper built a Multipath CNN model to feed all three components to model and concatenate feature maps to classify the classes. The model was evaluated with various parameters like accuracy and F1 score. The proposed model gives 96.17% for the dataset used. The result implies data, and a multipath approach helps to fuse features extracted from different images and allows models to classify well.

The rise of music industry across the globe can be seen with the new type of genre being created, and more artist and musicians joining this profession. A lot of music is created and launched every day. A major task for various music streaming platform is to classify these songs based on the genres and recommend music to the users. To overcome this many artificial intelligence algorithms are developed. One of the major problems in designing machine learning models is inadequate data for training. Certain dataset contains lot of redundant features that could cause the models to overfit. This problem could be resolved by data filtering. This paper has developed the multiple Artificial Intelligence (AI) models and applied data filtering method on the GTZAN dataset for music genre classification. A comparative analysis is done and discussed in this paper.

Emotion recognition and analysis is the process of identifying emotions and feelings of a person. Emotion analysis process is accurate in identifying expression in normal people in a single attempt. Emotion analysis is difficult in case of Autism Spectrum Disorder (ASD) children which are suffering with communication problems and speech problems. This paper proposed an optimized deep learning model with multi label classification for predicting ASD and NoASD with emotion analysis in children of age group 1 to 10 years. The kaggle dataset [1] of 1857 ASD children and 1850 Typically Developed (TD) children are used in this paper. Proposed model performance is tested on Yale Expression Dataset [2], CAFE children dataset [3] and also tested on social media dataset of autism parents group. The model is implemented by extracting face landmarks and is used to predict ASD and NoASD as first classification label and emotion is detected based on landmarks by computing internal and external distances by feature wise. Convolutional Neural Networks (CNN) is used to work with extracted face landmarks by using optimization methods, dropout, batch normalization and parameter updating. The proposed model is applied to predict 6 emotions irrespective of 4 general emotions with better accuracy.

The paper is focused on AC railway traction system 25kV / 50Hz (respectively 60Hz). Currents progress at fields of power electronic allows to use semiconductors converter as a part of traction substations. It brings more advantages for operation modes of TSS (symmetrization, compensation, filtration, TSS cooperation, recuperation, etc.). The most promising topologies are described in the second half of the paper. The main benefit of the paper is comparison of TSS selected topologies.

In the world of online marketing, Consent Management Providers (CMPs) are on the rise. CMPs collect online users’ consent to the processing of their personal data by publishers and ad-tech vendors. At the same time, there are reasonable doubts about the compliance of the existing market standard with the General Data Protection Regulation (GDPR). This could potentially create compliance risks for the companies adopting the standard, in particular for ad-tech vendors that – in the pre-CMP era – were invisible to users.This paper reveals drivers and obstacles for the adoption of the Transparency & Consent Framework (TCF) by ad-tech-vendors, gained in semi-structured interviews with representatives of Global Vendors List (GVL) members. Presenting the first qualitative study of ad-tech vendors’ perspectives, the paper provides novel insights into the CMP ecosystem. It particularly shows existing market pressure and reveals ad-tech vendors’ confusion and doubts about the TCF’s compliance with the GDPR.

Many techniques for medical diagnosis have been used in the medical world. For medical image processing of massive datasets, artificial intelligence is the most reliable and efficient process. Alzheimer's detection is also a popular diagnosis that can be simplified with the help of machine learning algorithms. Alzheimer's disease is caused by a malfunction of brain cells, and is also the cause of dementia. This paper investigated various machine learning methods such as RCNN, Quick RCNN, Faster RCNN, and help vector machines in depth (SVM). We used machine learning methods to analyse training and testing times in this article. Deep neural networks are a relatively recent machine learning method that has shown interest in a number of classification tasks. On a number of image recognition activities, the convolutional neural network outperforms the others. Medical image datasets, on the other side, are hard to come by since labelling them necessitates a high level of technological expertise. As a consequence, utilising a convolutional neural network (CNN)-based algorithm on a chest X-ray dataset, this paper explores how to detect pneumonia. Three techniques are tested using experiments.

In this paper, an application is developed to detect melanoma and non-melanoma type skin cancer using deep learning algorithm such as Convolutional Neural Networks (CNN). In addition, the paper discusses different models for detection of various types of skin diseases and skin cancer detection as well[3]–[6], [10], [11], [13]. And it also proposes to improve the accuracy of detection upto 95% using the basic CNN model along with various preprocessing algorithms on image and CSV data unlike models mentioned in the literature[3], [15]. Also implemented the detection of melanoma and non-melanoma type of skin cancer using ResNet-50.

The globe has been dominated today by Online Social Networks (OSN) like Facebook, Instagram, and Twitter. However, the rise of popularity and attractiveness of these web-based services has also led to the rise of malicious accounts known as bots. Not only do they cause Internet traffic, but also a group of these accounts working together can spread disinformation, incorrect, or manipulated news. The detection of bots, as well as their interactions with their communities and the rest of the world, is essential. This paper focuses on plotting a network of Twitter users, based on a particular hashtag, and detecting the communities in it, followed by detecting and locating the bots in these communities. In addition to this, sentiment analysis is conducted on normal users' tweets as well as the bots in these communities. This paper also aims to identify the overall sentiment of the communities, and thus provide promising conclusions relating to bot behavior in the overall network.

This paper investigates the potential of detecting stroke gaits by Deep Neural Network (DNN) models based on data from different sensor systems. Stroke survivors usually experience partial disability and need long-term rehabilitation. Because correct evaluation of stroke gaits is crucial for clinics to apply suitable medical treatments and rehabilitation strategies, we propose a method to detect stroke gaits by generalized gait data that might be obtained from different measurement systems. In this paper, we measure the subjects’ gait data by two equipment: an IMU system and a motion capture system. Then, we apply the IMU data to develop DNN models, which achieve an accuracy of 99.04% in detecting the stroke gaits by k-fold validation. Last, we applied the data measured by the motion capture system to test the DNN models. The results showed that the model could successfully detect stroke gaits with an average accuracy of 98.68%. In addition, we also applied a public dataset, where the gait data was measured by different IMU systems, to evaluate the model performance. The results show that the DNN models achieve an average accuracy of 98.25% in distinguishing normal gaits. These results confirm the possibility of detecting abnormal gaits based on data measured at different locations by assorted sensor systems.

By various hip joint disorders, human proximal femurs are deformed and the natural pure shape of the human proximal femurs is lost. Automatic classification of the proximal femur deformities emerging as a result of the hip diseases is important for the computer-aided diagnosis. In this paper, a study on automatic classification of Waldenstrom stages of Legg-Calve-Perthes disease (LCPD) patients from the 2D total deformity occurring on the proximal femurs is presented. Within the scope of the study, total deformities of the pathological proximal femurs were quantified initially by using a novel and hip joint symmetry based method that we reported in our previous research studies. To this end, total deformities of the pathological proximal femurs were represented as rational values. To classify the Waldenstrom stages of LCPD automatically, the Support Vector Machine (SVM) classifier was employed. Within the experimental studies performed on a dataset consisting of 13 magnetic resonance (MR) images of 13 patients with LCPD, an accuracy about 62% was observed in automatic classification of the Waldenstrom stages of the LCPD patients. The main contribution of the paper is that it is the first study so far, to the best of our knowledge, for the computer vision and machine learning based automatic classification of Waldenstrom stages of LCPD.

In this paper, a sum capacity maximization of the downlink non orthogonal multiple access (NOMA) based wireless network is studied in the presence of ambient backscattering (ABS). By incorporating a downlink NOMA with ABS, the mobile user can also be utilized as a Internet of Things (IoT) devices. Further, this paper employs the well known physical layer network coding (PLNC) scheme to support a pair of nodes in a single time slot. To ensure the quality of services (QoS) of each user, optimal power is allocated at the downlink level with total transmit power and minimum data rate requirement per user constraints using Lagrangian’s function. In addition, end-to-end outage probability of the proposed NOMA-PLNC based wireless network with ABS is derived. It is noted that the outage performance of the proposed wireless network with ABS is significantly improved. Moreover, considerable improvement in the far user capacity is achieved using ABS. Simulation results are presented to validate the analytical expressions.

Covid-19 pandemic is ravaging the world and humankind is facing one of the toughest challenges of this century. The main requirement is to stay safe. According to the Covid-19 protocol, a healthy person who is adjacent to an infected person for more than 15 minutes has a very high chance to get infected. Body temperature more than the normal temperature is one of the symptoms of the symbiotic Covid-19 infected person. This paper presents an idea to design a low-cost affordable wrist band that alerts the user if a person with higher body temperature is in his/her proximity. A simple component like thermopile temperature sensor, amplifier, monostable multivibrator, and LEDs are used to design this wrist band. The paper discusses the outline circuit to achieve this, which can be used by all people.