A novel memetic algorithm for discovering knowledge in binary and multi class predictions based on support vector machine

A Novel Tuberculosis Prediction Model by Extracting Radiological Features Present in Chest X-ray Images Using Modified Discrete Grey Wolf Optimizer Based Segmentation

In 2018, an invariant numbers ranging from 10 million people suffered from Tuberculosis (TB) approximately that has remained quite stable in recent years, based on the WHO 2019 survey report. This infection rate differs invariable among countries, from less than 5 to more than 500 new infections per 1,00,000 people each year, with a global average of around 130. Around 1.2 million HIV negative deaths existed in 2018. If this prevailing disease were diagnosed earlier, the death rate would have been under control, however sophisticated testing techniques tend to be cost prohibitive of wider acceptance. Some of the most important methods for TB diagnosis include thoracic X-ray image interpretation through image processing by the identification of various structures on thoracic X-rays and anomaly assessment is an important stage in computer-aided diagnosis systems. Chest form and size may contain indications for serious disorders such as pneumothorax, pneumoconiosis, tuberculosis and emphysema. Substantial work might have contributed to simplify diagnosis through implementing various statistical strategies to medical images, minimizing overtime and dramatically lowering overhead costs. In addition, recent advances in deep learning have provided magnificent results in the detection of images in different fields, but their use in diagnose TB remains limited. Thus, this work focuses on the development of a novel approach in disease detection. The concepts presented in this work are placed into practice and linked to current literature. We also proposed an automatic approach in conventional poster anterior chest X-rays for TB identification and diagnosis. We use the chest X-ray image with modified discrete grey wolf optimizer for segmentation techniques to eradicate abnormal areas and shape abnormality. We extract various features from the X-ray image with a shear let extraction that allows the image to be classified as normal or abnormal, based on a deep learning classifier, via the improved residual VGG net CNN with big data. Using Shenzhen Hospital Chest X-ray data set we test the efficiency of our system. The suggested technique has competitive results with comparatively shorter training period and greater precision depending on Masientropy based discrete gray wolf optimizer segmentation with an improved residual VGG net CNN. All the simulations are carried out in a mat lab environment.

EEG-based single-channel authentication systems with optimum electrode placement for different mental activities

BACKGROUND

Electroencephalogram (EEG) signals of a brain contain a unique pattern for each person and the potential for biometric applications. Authentication and security is a very important issue in our life and brainwave-based authentication is an addition to biometric authentication systems, which has many advantages over others. In this paper, we study the performance of a single channel brainwave-based authentication systems and select optimum channels based on mental activities.

METHODS

In this study, we used a dataset with five mental activities with seven subjects (325 samples). The EEG based authentication system includes three pre-processing steps, feature extraction, and classification. Features for Subject Authentication, are obtained from discrete Fourier transform, discrete wavelet transform, autoregressive modeling, and entropy features. Then these features are classified using the Neural Network, Bayesian network and Support Vector Machine.

RESULTS

We achieved accuracy in the range of 97-98% mean accuracy with Neural Network classifier for single-channel authentication system with optimum electrode placement for mental activity. We also analyzed the authentication system independently from the type of mental activity and chose channel O₂ as the optimum channel with an accuracy of 95%.

CONCLUSIONS

Channel optimization can obtain higher performance by reducing the number of EEG channels and defined the optimum electrode placement for different mental activities.

A Sensor Dynamic Measurement Error Prediction Model Based on NAPSO-SVM

Dynamic measurement error correction is an effective way to improve sensor precision. Dynamic measurement error prediction is an important part of error correction, and support vector machine (SVM) is often used for predicting the dynamic measurement errors of sensors. Traditionally, the SVM parameters were always set manually, which cannot ensure the model’s performance. In this paper, a SVM method based on an improved particle swarm optimization (NAPSO) is proposed to predict the dynamic measurement errors of sensors. Natural selection and simulated annealing are added in the PSO to raise the ability to avoid local optima. To verify the performance of NAPSO-SVM, three types of algorithms are selected to optimize the SVM’s parameters: the particle swarm optimization algorithm (PSO), the improved PSO optimization algorithm (NAPSO), and the glowworm swarm optimization (GSO). The dynamic measurement error data of two sensors are applied as the test data. The root mean squared error and mean absolute percentage error are employed to evaluate the prediction models’ performances. The experimental results show that among the three tested algorithms the NAPSO-SVM method has a better prediction precision and a less prediction errors, and it is an effective method for predicting the dynamic measurement errors of sensors.