International Research Journal of Advances in Computing Sciences

Comparative Analysis of Neural Network Architectures for Software Cost Estimation

2024-05-12T06:32:02+00:00

Software cost estimation is a critical aspect of project management, with significant implications for decision-making and resource allocation. In recent years, the use of neural networks (NNs) has shown promise in improving the accuracy of cost estimation models. This research, unique in its systematic investigation of various NN architectures, aims to determine their strengths and weaknesses in the context of cost estimation. The study focuses on three primary neural network architectures: feedforward neural networks (FNNs), recurrent neural networks (RNNs), and convolutional neural networks (CNNs). It scrutinizes each architecture's ability to capture and model the complex relationships within software development datasets, with a particular emphasis on their performance in terms of prediction accuracy, scalability, and adaptability to diverse project scenarios. By providing a comprehensive understanding of the nuanced dynamics between neural network structures and their effectiveness in software cost estimation, this review aims to offer practical insights that can guide practitioners and researchers in selecting the most suitable neural network architecture for specific cost estimation challenges. Additionally, it identifies potential areas for future research, thereby offering a roadmap for advancing the application of neural networks in software cost estimation.

An Innovative Methodology for Elicitation Technique Selection via Attribute Mapping

2024-05-12T06:46:59+00:00

The software development process relies entirely on stakeholders' requirements. The end product will likely be optimal and successful if stakeholders' requirements are integrated into the proposed system. Various Requirement Elicitation Techniques (RET) are employed to achieve a successful product. The selection of a suitable RET depends on the nature of the product under development. Therefore, a single RET cannot be universally applied to all products. This paper aims to distinguish between all RETs, facilitating analysts in selecting the most appropriate RET from the available options. Additionally, we have devised a novel mapping framework that identifies the most suitable RET for any software based on its attributes. We have also implemented this framework using an online vehicle booking system as an illustrative example.

Analyzing The Best Key Encryption Algorithm

2024-05-12T07:00:58+00:00

With technology now firmly established and this being the decade of tech professionalism, data security stands out as a paramount concern. In an era where data is just a click away for users, ensuring its security from unauthorized access becomes imperative. Various fields dedicated to data security have emerged to address this challenge and uphold confidentiality. This paper delves into the spectrum of data encryption techniques to identify the optimal algorithm for safeguarding data. Our examination compares symmetric and asymmetric encryption methods and evaluates algorithms within these categories. Factors such as time, efficiency, memory usage, latency, key size, and number of rounds are scrutinized to determine the most suitable algorithm for a given dataset.

Retinal Fundus Image Segmentation using U-Net Deep Neural Network

2024-05-16T13:56:42+00:00

Deep learning algorithms have become popular in machine learning and computer vision due to their ability to achieve state-of-the-art results. This paper proposes a method for accurately classifying retinal vessel segmentation using deep learning algorithms and state-of-the-art techniques. Specifically, the approach uses a U-Net convolutional neural network to extract vessels from retinal fundus images taken from the DRIVE dataset. Pre-processing is applied to the images to extract features such as shape, size, and arterio-venous crossing types, which can provide valuable insights into various eye diseases. The model is trained on five epochs and 300 layers, achieving an accuracy of 0.9691 and producing clear and accurate vessel segmentation. These results have the potential to aid computer-aided detection tasks in retinal images.

A Threshold-based Steganographic Approach for Improved Security with Payload and PSNR

2024-05-12T07:06:19+00:00

It is essential to ensure data security in the modern world where all our important information is online. Security is the primary concern when sharing information over a communication network to avoid third parties. Steganography is the method of hiding secret data into a cover medium so that the data that will be transmitted cannot be detected by a third party. Steganography plays a vital in securing data over any network. Many steganographic methods have been introduced to ensure the transmission of data. Where image steganography is the well-known method. The main challenges in image steganography are the payload capacity of hidden data, imperceptibility, and robustness against unauthorized access. The Peak Signal to Noise Ratio (PSNR) can be low by hiding large amounts of secret data inside the image. Steganography and cryptography are interrelated, but cryptography converts plain text into cipher text to make it unreadable.In contrast, steganography is the method of hiding secret data in a cover image so that it is even unable to detect or guess the existence of any secret information. In this research, a method of threshold-based steganography is introduced. Secret data is encoded in the cover picture depending on a threshold of the cover image's colour levels after the cover image is scanned for all colour levels. We used the cover image's least significant bits (LSB) to embed the secret data. To improve payload, we encrypt-ensure security; we encrypted to ensure the encode it. The proposed method is implemented and analyzed on 512 × 512 sizes of coloured and greyscale images as a dataset using MATLAB tool. Our work outperforms the existing methods by attaining a 99% performance rate to achieve the best stego image that securely carries maximum secret data, i.e., Payload and security, along with other parameters.

Influenza Pneumonia an Efficient Deep Learning Based Classification to Differentiate Coronavirus and Influenza Pneumonia

2024-05-12T07:26:21+00:00

In the context of increased numbers of coronavirus-affected people, the availability of highly accurate detection of COVID-19 is a crucial task worldwide due to its structural changes and feature similarities with influenza pneumonia. RT-PCR (Reverse Transcription Polymerase Chain Reaction), a molecular test frequently used for virus detection, is a diagnostic tool for coronavirus and influenza pneumonia. Two subtypes of influenza pneumonia (IP), known as types A and B, share biological characteristics with coronaviruses. The medical findings might not be able to distinguish between the novel coronavirus and numerous types of influenza pneumonia due to coronavirus variants. Coronavirus is detected using real-time reverse transcription polymerase chain reaction, but this method has a significant negative false positive rate. A medical specialist's help is required for the early and correct detection of coronavirus and influenza pneumonia to treat and recover patients from infections following their modality. Coronavirus and influenza pneumonia have similar illness patterns, and no diagnostics have a high positive rate. The precise diagnosis of these infections is difficult for medical professionals. For the disease to be identified and treated promptly to prevent further health problems, precise COVID-19 and influenza pneumonia detection is necessary. Therefore, our proposed work in this thesis focuses on the high-quality evaluation and detection of both viruses and answers the following questions:

How can early recognition of COVID-19 and influenza pneumonia be evaluated through medical images?
How can the quality of results regarding the early recognition of COVID-19 and influenza pneumonia be enhanced?

The above questions will be answered by deep learning state-of-the-art models and evaluation metrics for early detection of the viruses. This thesis presents precise virus detection through the proposed model with enhanced results. The medical images are evaluated using the deep learning-based model VGG-8. Our research used the database of X-ray medical images. X-ray images comprise the coronavirus, pneumonia and normal ordinary people, which are used for the classification through deep learning model. Evaluation of X-ray medical data shows the model’s accuracy to up to 98.99% for classification. Therefore, the proposed evaluation system and classification model successfully detected the coronavirus and influenza pneumonia with improved results.

A Comprehensive Survey of Feature Modeling in Augmented/Virtual Reality: Current Trends and Future Directions

2024-05-16T15:11:17+00:00

Recent advances in Augmented Reality and Virtual Reality have led to significant changes across various industries and businesses. The feature modelling method is used to handle the diversity of software products. This has been a strategy that can be used to create AR/VR apps. The article provides a comprehensive summary of the current research in feature modelling for AR/VR. Over 100 research articles are examined to identify this topic's patterns, challenges, and future paths. The survey is designed to provide researchers and practitioners with valuable insight into advanced feature modelling for AR/VR and suggest relevant areas for future study.