Recent Advances in Electrical & Electronic Engineering (Formerly Recent Patents on Electrical & Electronic Engineering) - Volume 16, Issue 8, 2023

In recent years, the extraction of retinal blood vessels from low contrast retinal images has become a challenging task for diagnosing retinal diseases such as Diabetic Retinopathy, Agerelated Macular Degeneration (AMD), Retinopathy of Prematurity (ROP), cataract, and glaucoma. Another challenge is screening the retinal image to identify the disease early on. However, data analysis from a large population-based study of retinal diseases is required to help resolve the uncertainty in identifying the retinal disease based on retinal image classification using deep learning approaches from the retinal diseases dataset. Therefore, we proposed the survey on the deep learning approach for screening the retinal image to identify the early stages of the disease and discussed retinal disease analysis based on deep learning approaches to detect Diabetic Retinopathy, AMD ROP, and Glaucoma. We also discuss deep learning applications in the segmentation of retinal blood vessels, extraction of the optic disc, optic cup, and fovea, and OCT segmentation to detect retinal disease for diagnosis of diseases. Finally, discuss the classification of arteries/veins using a deep learning approach.

Introduction: Cloud has gained a huge number of consumers in today's environment due to its broad range of services. On the cloud server, data owners can store any type of data, and users can access it whenever they need it. Methods: Numerous challenges come with the ease of use of outsourced data. The accuracy and secrecy of data outsourced come at stake. Users have a low level of trust in cloud service providers since they can be deceptive at times. A cloud audit is an examination conducted by a data owner to assess and document the performance of their cloud vendor. Results: The goal of a cloud vendor audit is to determine how well they follow a set of predetermined controls. Conclusion: In the article, the authors present various schemes of cloud auditing, their categorizations, and merits and demerits along with the future directions of research in the domain.

Background: The novelty of the work lies in the formulation of these frequency-based generators, which reflects the lowest level of information loss in the intermediate calculations. The core idea behind the approach presented in this work is that a module with complex logic involved may have more probability of bugs. Software defect prediction is the area of research that enables the development and operations team to have the probability of bug proneness of the software. Many researchers have deployed multiple variations of machine learning and deep learning algorithms to achieve better accuracy and more insights into predictions. Objective: To prevent this fractional data loss from different derived metrics generations, a few optimal transformational engines capable of carrying forward formulations based on lossless computations have been deployed. Methods: A model Sodprhym has been developed to model refined metrics. Then, using some classical machine learning algorithms, accuracy measures have been observed and compared with the recently published results, which used the same datasets and prediction techniques. Results: The teams could establish watchdogs thanks to the automated detection, but it also gave them time to reflect on any potentially troublesome modules. For quality assurance teams, it has therefore become a crucial step. Software defect prediction looks forward to evaluating error-prone modules likely to contain bugs. Conclusion: Prior information can definitely align the teams with deploying more and more quality assurance checks on predicted modules. Software metrics are the most important component for defect prediction if we consider the different underlying aspects that define the defective module. Later we deployed our refined approach in which we targeted the metrics to be considered.

Background: The necessity for data security is demanding in the digital age. The main technique for achieving data security is Steganography. The technique of camouflaging the secret object behind another cover object is known as Steganography. A novel, robust, high payload, and imperceptible image steganography approach using matrix inverse in the spatial domain are proposed in this manuscript. The basic idea is to devise a robust novel approach against various image processing attacks, like cropping, compression, filters, and noise. Objectives: The study's objective is to develop a novel data-hiding approach that increases imperceptibility and payload capacity and is robust against various image processing attacks, like filters, compression, cropping, and noise. Methods: The matrix inverse procedure is used for the insertion and extraction of data. The symmetry feature of the matrix inverse makes the task of insertion of data simple and efficient. It also increases the hiding capacity while maintaining a finer level of imperceptibility and robustness. MATLAB is used for the implementation of the new technique and results analysis. Results: The proposed method's robustness has been analyzed against image processing assaults such as the inclusion of various noises, cropping, a variety of filters, and compression assaults. The imperceptibility of the approach has been tested successfully using PSNR, BER, and NCC metrics. The proposed method has been compared with the other two techniques. The experimental and comparison results depict that the proposed approach provides high hiding capacity, finer robustness, and imperceptibility. Conclusion: A novel, robust and imperceptible approach has been developed in this manuscript. The proposed method has been compared with the methods developed by Jung & Yoo and Joshi & Gill. The experimental findings show that the proposed technique offers better resilience, payload capacity, and finer imperceptibility.

Background: Steganography is the approach of camouflaging the covert object within another cover object. This manuscript suggested a novel steganography approach to conceal the covert data presence. The basic idea behind this is to generate an information-hiding approach that increases the payload capacity and good PSNR value without sacrificing much distortion of the image. Objective: To develop a novel data-hiding approach that increases imperceptibility, robustness, and payload capacity. Methods: The Neighbour Mean Interpolation technique is used to scale up the original image to generate Interpolated pixels of the given image. An even-odd scheme on the interpolated stego pixel is used to camouflage the obscure code. MATLAB is used for the implementation of the new approach and results calculation. Results: The Experimental analysis reveals that our suggested approach has a finer embedding capacity for camouflaging the secret data as the original image of size (MxN) is scaled up to size (2M- 1 x2N-1) and also manages the good visuality of the cover or graven image. The proposed method is compared with Jung and Yoo, and Selvrani's method. The result of this comparison shows that the proposed method has finer imperceptibility than these two previously existing techniques. Conclusion: A novel approach towards image steganography using neighbor mean interpolation has been proposed and implemented. A new steganography method is used for camouflaging the confidential code into the cover object using NMI without producing any major differences in the input image. The new approach provides better imperceptibility, robustness, and payload capacity.

Background: For the efficient and secure running of the power industry, accurate monthly electricity projections are crucial. Due to coupling variations and a variety of data resolutions, current approaches are still unable to accurately extract multidimensional time-series data. Objective: For monthly electricity consumption forecasting, a multi-time-scale transformation and temporal attention neural network for a temporal convolutional network is proposed. Methods: First, a multi-time-scale compression model of temporal convolutional network is proposed, which compresses data on several time scales from different resolutions, such as the economy, weather, and historical load. Second, a multi-source temporal attention module is built to further dynamically extract crucial information. Finally, the decoding-encoding and residual connections' structure contributes to the prediction's improved resilience. Results: The proposed method was compared with the state-of-the-art monthly load forecasting based on two years of historical data in a certain region, demonstrating its effectiveness. Conclusion: Through the verification of local historical data, the proposed model was contrasted with cutting-edge monthly load forecasting techniques. The obtained results demonstrate the effectiveness.

Background: This paper is consecrated to the development of a new approach to control a bidirectional DC-DC converter dedicated to battery storage systems by applying an optimal control based on a linear quadratic regulator (LQR) combined with an artificial neural network (ANN) algorithm. A state representation of the Buck-boost converter is performed. Then the ANN-LQR control strategy is compared to a classical control based on the proportional-integral controller combined with an ANN algorithm. The ANN algorithm generates the reference charging or discharging current based on a comparison between the power generated and the power consumed. In order to obtain an accurate comparison, two identical systems are designed, each consisting of a photovoltaic system optimized by an incremental conductance algorithm (INC) that powers a dynamic load and a backup storage system consisting of a lithium-ion battery. A management and protection algorithm is developed to protect the batteries from overcharge and deep discharge and to manage the load availability on the DC bus. The simulation results show an improvement in the performances of the storage system by the ANN-LQR control compared to the ANN-PI method and an increase in the stability, accuracy, efficiency of the system is observed. Photovoltaic (PV) energy is one of the most promising technologies for combating climate change and meeting the urgent need for green renewable energy and long-term development. PV energy generation has a number of advantages: Solar energy is limitless and available anywhere on the planet. However, photovoltaic energy is intermittent and depends on meteorological conditions; also, the energy consumed is unpredictable. For this reason, a storage system is necessary to overcome these problems. Objective: The objective of this study is to develop an optimal control using a Linear Quadratic Regulator (LQR) combined with a neural network algorithm (ANN) to improve the performance of an electrical energy storage system and compare the results obtained with the classical control based on the PI regulator. Methods: The state representation of the bidirectional Buck-boost converter was performed in order to apply the optimal control and determine the gain K and the ANN algorithm allowed to determine the charge and discharge current according to a comparison between the power produced and consumed. Results: The simulation results obtained by two control methods can be used to compare and select the appropriate control method to achieve optimal efficiency of the storage system. Conclusion: The combined ANN-LQR technique offer better performances and stability of the installation compared to the ANN-PI controller.