Research Articles

Browse

Now showing 1 - 20 of 67

Develop a System to Analyze Logs of a Given System Using Machine Learning
(Independent University, Bangladesh (IUB), 2023) Hasan, Md. Tarek; Sadia, Farzana; Hasan, Mahady; Rokonuzzaman, M.
Software error detection is a critical aspect of software development. However, due to the lack of time, budget, and workforce, testing applications can be challenging, and in some cases, bug reports may not make it to the final stage. Additionally, a lack of product domain knowledge can lead to misinterpretation of calculations, resulting in errors. To address these challenges, early bug prediction is necessary to develop error-free and efficient applications. In this study, the author proposed a system that uses machine learning to analyze system error logs and detect errors in real time. The proposed system leverages imbalanced data sets from live servers running applications developed using PHP and Codeigniter. The system uses classification algorithms to identify errors and suggests steps to overcome them, thus improving the software’s quality, reliability, and efficiency. Our approach addresses the challenges associated with large and complex software where it can be difficult to identify bugs in the early stages. By analyzing system logs, we demonstrate how machine learning classification algorithms can be used to detect errors and improve system performance. Our work contributes to a better understanding of how machine learning can be used in real-world applications and highlights the practical benefits of early bug prediction in software development.
Machine Learning Modeling for Reconditioned Car Selling Price Prediction
(Independent University, Bangladesh (IUB), 2023) Abdullaha, Fatema; Rahmana, Md. Ataur; Shidujaman, Mohammad; Hasan, Mahady; Habibb, Md. Tarek
Almost 80% of the vehicles required for Bangladesh's road transportation industry are supplied by reconditioned cars. Using machine learning (ML) to predict car prices refers to using ML algorithms and techniques to make assumption about future car prices. This can be useful for a variety of purposes, such as helping car buyers and sellers make informed decisions, assisting car dealerships with inventory management, or providing insights for car manufacturers and other industry stakeholders. To predict car prices using ML, data is collected on a variety of factors that can affect the ongoing cost of a car, such as its make and model, age, mileage, condition, and location. This data is then fed into the XGBoost ML model, which uses statistical techniques to analyze the data and identify patterns and trends. The model performs 98% accurately in the tested portion of the data set and ensures that the model can then be used to make predictions on the future cost of an automobile based on these patterns and trends.
Tissue Artifact Segmentation and Severity Assessment for Automatic Analysis using WSI
(Independent University, Bangladesh, 2023-02) HOSSAIN, MD SHAKHAWAT; SHAHRIAR, GALIB MUHAMMAD; SYEED, M M MAHBUBUL; FAISAL UDDIN, MOHAMMAD; HASAN, MAHADY; HOSSAIN, MD SAKIR; BARI, RUBINA
Traditionally, pathological analysis and diagnosis are performed by manually eyeballing glass-slide specimens under a microscope by an expert. The whole slide image (WSI) is the digital specimen produced from the glass slide. WSI enabled specimens to be observed on a computer screen and led to computational pathology where computer vision and artificial intelligence are utilized for automated analysis and diagnosis. With the current computational advancement, the entire WSI can be analyzed autonomously without human supervision. However, the analysis could fail or lead to wrong diagnosis if the WSI is affected by tissue artifacts such as tissue fold or air bubbles depending on the severity. Existing artifact detection methods rely on experts for severity assessment to eliminate artifact-affected regions from the analysis. This process is time-consuming, exhausting and undermines the goal of automated analysis or removal of artifacts without evaluating their severity, which could result in the loss of diagnostically important data. Therefore, it is necessary to detect artifacts and then assess their severity automatically. In this paper, we propose a system that incorporates severity evaluation with artifact detection utilizing convolutional neural networks (CNN). The proposed system uses DoubleUNet to segment artifacts and an ensemble network of six fine-tuned CNN models to determine severity. This method outperformed current state-of-the-art in accuracy by 9% for artifact segmentation and achieved a strong correlation of 97% with the pathologist’s evaluation for severity assessment. The robustness of the system was demonstrated using our proposed heterogeneous dataset and practical usability was ensured by integrating it with an automated analysis system.
Educational Certification on Ethereum blockchain: Analysis on How BCert and UZHBC differ
(Independent University, Bangladesh, 2023-03) Aadeef, Tanvir Ahmed; Monir, Md. Fahad; Alam, Sanzar Adnan
Counterfeit certificates and diplomas are a common issue. Blockchain technology offers a potential solution through its decentralized ledger system, which uses cryptographic techniques and incentives to verify and store transactions in blocks that are added to the public blockchain. This secure and transparent method of storing data makes it ideal for Industry 4.0 and beyond as it is difficult for information to be stolen or for corrupt institutions to sell unauthenticated certificates. In this research paper, we compare and analyze existing two different methods of certification issuance and management which is BCert, and University of Zurich Blockchain. We also discuss the differences in algorithms for hashing and how we can improve the existing code for each method. Our aim is to thoroughly examine these approaches and contribute to the field of certification through our analysis and conclusions.
Impact of COVID-19 on the Factors Influencing On-Time Software Project Delivery: An Empirical Study
(18th International Conference on Evaluation of Novel Approaches to Software Engineering (ENASE), April 24 - 25, 2023 Prague, Czech Republic, 2023-03) Islam, Mahmudul; Khan, Farhan; Hasan, Mehedi; Sadia, Farzana; Hasan, Mahady
The objective of this research paper is to investigate the impact of COVID-19 on the factors influencing ontime software project delivery in different Software Development Life Cycle (SDLC) models such as Agile, Incremental, Waterfall, and Prototype models. Also to identify the change of crucial factors with respect to different demographic information that influences on-time software project delivery. This study has been conducted using a quantitative approach. We surveyed Software Developers, Project Managers, Software Architect, QA Engineer and other roles using a Google form. Python has been used for data analysis purposes. We received 72 responses from 11 different software companies of Bangladesh, based on that we find that Attentional Focus, Team Stability, Communication, Team Maturity, and User Involvement are the most important factors for on-time software project delivery in different SDLC models during COVID-19. On the contrary, before COVID-19 Team Capabilities, Infrastructure, Team Commitment, Team Stability and Team Maturity are found as the most crucial factors. Team Maturity and Team Stability are found as common important factors for both before and during the COVID-19 scenario. We also identified the change in the impact level of factors with respect to demographic information such as experience, company size, and different SDLC models used by participants. Attentional focus is the most important factor for experienced developers while for freshers all factors are almost equally important. This study finds that there is a significant change among factors for on-time software project delivery before and during the COVID-19 scenario.
Fake News Detection Using Machine Learning Techniques
(Independent University, Bangladesh, 2023-03) Sultana, Achhiya; Islam, Mahmudul; Hasan, Mahady; Ahmed, Farruk
A lot of information is spread by people in the social media to update their status and share crucial news with others. But the majority of these platforms don’t promptly validate the individuals or their posts and people aren’t able to identify the fake news manually. Therefore, there is a need for an automated system capable of detecting fake news. This research has proposed to build a model using four machine learning algorithms. The dataset employed in the experiment is a composite of two datasets containing almost equal amounts of true and fake news articles on politics. The preprocessing stages begin with cleaning the data by removing punctuation, tokenization, special characters, white spaces, redundant word elimination, numerals, and English letters followed by stemming and stop with data discretization. Then, we analyzed the collected data and 80% of the data has been used to train each model initially. After that, the four manifested classification algorithms are applied. For identifying fake news from news articles, methods like Logistic Regression, Decision Tree, Random Forest, and Gradient Boosting Classifier were used. The trained classifiers’ accuracy has been evaluated using the remaining 20% of the data. The results show that the decision tree model produces the best accuracy of 99.60% and gradient boosting of 99.55%. Besides, the random forest shows 99.10% along with the logistic regression 98.99%. Moreover, we have explored the best model to achieve the highest precision, recall, F1-score based on the confusion matrix’s outcome.
Automatic Detection of Natural Disaster Effect on Paddy Field from Satellite Images using Deep Learning Techniques
(Independent University, Bangladesh, 2023-03) Alavi Ishmam, Tahmid; Ali, Amin Ahsan; Amin, Md Ahsraful; Rahman, A K M Mahbubur
Rice is the main food in Bangladesh and for more than the world’s half population. As Bangladesh is prone to natural disasters, it’s important to detect where rice production is compromised so that government can reach with aid and can take preventive measures. This paper used high-resolution satellite imagery (Sentinal-2) to detect damaged rice penology from natural disaster. The authors developed ground truth data for rice field damage from the field level. At first, NDVI differences before and after the disaster are calculated to identify possible crop loss. The areas equal to and above the 0.33 threshold are marked as crop loss areas as significant changes are observed. The authors also verified crop loss areas by collecting data from local farmers. Later, different bands of satellite data (Red, Green, Blue) and (False Color Infrared) are useful to detect crop loss area. We used the NDVI different images as ground truth to train the DeepLabV3plus model. With RGB, we got IoU 0.41 and with FCI, we got IoU 0.51. As FCI uses NIR, Red, Blue bands and NDVI is normalized difference between NIR and Red bands, so greater FCI’s IoU score than RGB is expected. But RGB does not perform very badly here. So, where other bands are not available, RGB can use to understand crop loss areas to some extent. The ground truth developed in this paper can be used for segmentation models with very high resolution RGB only images such as Bing, Google etc.
Optimizing B2B Relationships with Post-Covid Sales and Marketing Automation through ERP Implementation: A Value-Driven Approach in Bangladesh
(Independent University, Bangladesh, 2023-05) Akthar, Sayeda Rahnuma; Khan, Mohammad Sajjad; Sadia, Farzana; Hasan, Mahady
Enterprise Resource Planning (ERP) is a powerful software program created to combine and streamline several company processes. Companies in developed countries frequently use it to increase their general effectiveness. ERP enables companies to meet the specific demands, interests, and behaviours of their customers, who are increasingly seeking individualized experiences. Even with the help of an implementation partner, deploying ERP properly necessitates a solid understanding of the involved process. During the implementation stage, effective communication and responsibility sharing can be challenging. The main goal of this paper is to create a thorough guidebook that offers insightful advice to businesses and their clients throughout the whole ERP deployment process. Six interviews were done to obtain information and address any implementation-related worries in order to accomplish this. These talks led to the creation of an implementation handbook that addressed important challenges and offered useful solutions. A methodology model diagram is also suggested to act as a visual roadmap for an organized and fruitful ERP installation. The objective is to equip businesses and their clients with the information and resources they need to successfully complete the deployment process and get the most out of their ERP system.
Morphological Classification of Radio Galaxies using Semi-Supervised Group Equivariant CNNs
(Independent University, Bangladesh, 2023-05) Hossain, Mir Sazzat; Roy, Sugandha; Asad, K. M. B.; Momen, Arshad; Ali, Amin Ahsan; Amin, M Ashraful; Rahman, A. K. M. Mahbubur
Out of the estimated few trillion galaxies, only around a million have been detected through radio frequencies, and only a tiny fraction, approximately a thousand, have been manually classified. We have addressed this disparity between labelled and unlabeled images of radio galaxies by employing a semi-supervised learning approach to classify them into the known FRI and FRII types. A Group Equivariant Convolutional Neural Network was used as an encoder that preserves the equivariance for the Euclidean Group E(2) to learn the representation of globally oriented feature maps through new SelfSupervised Learning (SSL) techniques SimCLR and BYOL. After representation learning, we trained a fully-connected classifier and fine-tuned the trained encoder with labelled data. We have found that this semi-supervised approach helps our method outperform a state-of-the-art method of classifying radio galaxies in many metrics. Our work reiterates the importance of semisupervised learning in radio galaxy classification, where labelled data are scarce, but prospects are immense.
A Hybrid Approach to Overcome Requirements Challenges in the Software Industry
(Independent University, Bangladesh, 2023-05) Hasan, Md. Tarek; Bakar, Nabil Mohammad Abu; Nahar, Nujhat; Hasan, Mahady; Rokonuzzaman, M.
This research paper presents a hybrid approach to overcome the challenges related to inadequate or insufficient client involvement and understanding during the software requirements phase. The aim of this study is to investigate the factors that contribute to this challenge and propose a solution that combines traditional and agile methodologies. To accomplish this, a survey was conducted to collect responses from industry professionals in the software development sector. The survey results showed that inadequate or insufficient client involvement and understanding is a common issue that leads to delays and misunderstandings in software development projects. To address this challenge, the proposed hybrid approach combines the traditional requirements engineering process with agile techniques such as user stories, prototypes, and continuous feedback loops. The hybrid approach aims to improve communication and collaboration between the client and the development team, ensuring that the software’s requirements are well-understood and documented. The results of this study indicate that the proposed hybrid approach is effective in overcoming the challenges related to inadequate or insufficient client involvement and understanding. The findings of this research have practical implications for software development organizations, highlighting the importance of adopting a hybrid approach to ensure successful software development projects.
Time Series Analysis and Forecasting of Air Quality Index of Dhaka City of Bangladesh
(Independent University, Bangladesh, 2023-05) Sakib, Sheikh Rahmatulla; Sara, Kamarun Nahar; Rasel, Md. Tahmid Hossain; Asif, Md. Masudul Islam; Nahid, Md. Aynul Hasan; Rahman, Md. Saifur; Mridha, M. F.; Islam, Ashraful
In Dhaka, the capital city of Bangladesh, various sources including vehicle emissions, industrial activities, brick kilns, building sites, and open rubbish burning contribute to the air pollution problem. To assess the air quality, the Air Quality Index (AQI) is utilized, which categorizes air quality based on pollutant concentration. In this study, we have built ARIMA, Auto-ARIMA, SARIMAX, and VAR models to predict the air quality of Dhaka. Unlike previous studies, we have utilized hourly air pollutants factors such as PM2.5, PM10, SO2, CO, NO2, and O3 to forecast air quality. Our novel approach enables us to predict the monthly and weekly air quality of Dhaka city. Our analysis reveals that the SARIMAX model, which takes into account seasonal patterns, trends, and external factors, is the most accurate in predicting Dhaka city’s air quality. The model’s prediction performance is assessed using statistical indicators such as mean absolute percentage error and root mean square error. The study highlights that the SARIMAX model could aid policymakers in evaluating the efficacy of air pollution control measures.
Enhancing Product Market Payoff in Small and Medium Internet-Based Firms: A Survey-Based Analysis of Innovation and Competition Factors
(Independent University, Bangladesh, 2023-05) Bakar, Nabil Mohammad Abu; Hasan, Mahady; Rokonuzzaman, M.
This paper aims to explore the relationship between innovation and competition and identify the conditions that affect the industry’s innovation and welfare. The authors analyze various factors, including the properties of product market payoffs, to determine whether competition increases or decreases industry innovation. The ultimate goal of this study is to provide policy recommendations that support innovation and competition.To achieve this objective, the authors conducted a survey to collect responses from industry professionals and identify the challenges that hinder innovation and competition. They examined several models and case studies,to understand how competition drives innovation. Despite the challenges, leading business consultants found that increased competition leads to higher productivity in various industries, including manufacturing and services.The study’s findings reveal that the recommended policy effectively overcomes the challenges related to innovation and competition sustainability. These findings have practical implications for regulatory organizations, highlighting the importance of adopting these policies to ensure innovation and competition for small and medium-sized internet-based firms.This paper presents a thorough examination of the correlation between innovation and competition, outlines the key elements that foster both innovation and competition, and suggests policy measures to bolster innovation and competition while improving product market outcomes.
SMOTE Oversampling and Near Miss Undersampling Based Diabetes Diagnosis from Imbalanced Dataset with XAI Visualization
(Independent University, Bangladesh, 2023-05) Nayan, Nasim Mahmud; Islam, Ashraful; Islam, Muhammad Usama; Ahmed, Eshtiak; Hossain, Mohammad Mobarak; Alam, Md Zahangir
This study investigated the predictive ability of ten different machine learning (ML) models for diabetes using a dataset that was not evenly distributed. Additionally, the study evaluated the effectiveness of two oversampling and undersampling methods, namely the Synthetic Minority Oversampling Technique (SMOTE) and the Near-Miss algorithm. Explainable Artificial Intelligence (XAI) techniques were employed to enhance the interpretability of the model’s predictions. The results indicate that the extreme gradient boosting (XGB) model combined with SMOTE oversampling technique exhibited the highest accuracy and an F1-score of 99% and 1.00 respectively. Furthermore, the utilization of XAI methods increased the dependability of the model’s decision-making process, rendering it more appropriate for clinical use. These results imply that integrating XAI with ML and oversampling techniques can enhance the early detection and management of diabetes, leading to better diagnosis and intervention.
An IoT Intensive AI-integrated System for Optimized Surface Water Quality Profiling
(Independent University, Bangladesh, 2023-05) Syeed, M M Mahbubul; Karim, Md. Rajaul; Hossain, Md Shakhawat; Fatema, Kaniz; Uddin, Mohammad Faisal; Khan, Razib Hayat
Surface water is heavily exposed to contamination as this is the ubiquitous source for the majority of water needs. This situation is exaggerated by excessive population, heavy industrialization, rapid urbanization, and ad-hoc monitoring. Comprehensive measurement and knowledge extraction of surface water pollution is therefore pivotal for ensuring safe and hygienic water use. However, current process of surface water quality profiling involves laboratory-based manual sample collection and testing, which is tardy, expensive, error-prone, and untraceable. This paper, therefore presents the design and development of an IoT integrated water quality profiling system that possesses a novel plug-and-play physical layer for the sensor actuation, and an AI powered fog computing based cloud application layer for remote water quality parameter measurement and data acquisition, remote data logging, monitoring and control, with data analytic for critical reasoning and decision making
A Gradient Boosted ML Approach to Feature Selection for Wireless Intrusion Detection
(Independent University, Bangladesh, 2023-05) Mondal, Birupaxha; Faisal, Fahim; Towshi, Zeba Tusnia; Monir, Md. Fahad; Ahmed, Tarem
The proliferation of Wi-Fi-enabled devices makes security a non-negotiable part of connectivity. As new attacks are discovered that compromise the security of devices in the wireless ecosystem, it is becoming increasingly crucial for intrusion detection systems to generalize to these novel attacks. Machine Learning gives us an approach to do that. In this paper, we provide a feature elimination technique to narrow down the set of features necessary to build such an ML-based solution that takes into account possible class imbalance issues in intrusion datasets. With features extracted using this technique from the AWID dataset, we use a gradient-boosted model to show that these features are necessary to generalize to new attack types in the AWID test dataset.
CaViT: Early Stage Dental Caries Detection from Smartphone-image using Vision Transformer
(Independent University, Bangladesh, 2023-05) Hossain, Md Shakhawat; Rahman, Md Mahmudur; Syeed, M M MAHBUBUL; Hannan, Ummae Hamida; Uddin, Mohammad Faisal; Mumu, Sahria Bakar
Caries detection is a routine clinical task in dental practice. If caries are detected at an early stage, non-invasive ormicro-invasive treatment such as ﬁllings and a root canal can be effective and thereby invasive treatment and therapies such as gum surgery and dental implants can be avoided. Invasive treatments are expensive and inappropriate for patients with low blood cell counts, cardiac problems and other health issues.Consequently, early caries detection is critical in dentistry. Caries are typically identiﬁed through a visual tactile examination in support of radiographic imaging. Fluorescence imaging, cone beam computed tomography or optical coherence tomography are also used. However, these procedures are time-consuming and expensive and require a physical examination of the patient.Moreover, the COVID-19 lessons taught us that such diagnoses should be avoided to prevent contagious diseases. Existing auto-mated caries detection methods fail to achieve sufﬁcient accuracy.Therefore, in this paper, we propose a highly accurate automatic system to detect early caries without any face-to-face interaction with the patient. This system is economical, rapid and easy to use. The proposed system uses a smartphone to capture teeth images and then relies on a vision transformer (ViT) to classify the images as advanced, early or no caries. Finally, the caries are segmented using a U-Net network. The proposed method outperformed the existing methods and achieved a sensitivity of95%, 91% and 100% for the no caries, early caries and advanced caries classes when tested on a dataset of 300 images, developed for this study.
Genre Classification of Bangla Poem Using Machine Learning and Deep Learning Techniques
(Independent University, Bangladesh, 2023-05) Pasha, Syed Tangim; Islam, Ashraful; Rahman, Mohammed Masudur; Ahmed, Eshtiak; Foysal, Md. Ferdouse Ahmed; Alam, Md Zahangir
The computational analysis of the Bangla poems is a challenging task due to the diverse linguistic, stylistic, and semantic features of the Bangla language. In this work, we prepared a dataset of 1311 Bangla poems of two separate categories: Love and Miscellaneous poem, which contain 500 and 811 poems respectively. We used word or semantic-based features to classify Bangla poems using the TF-IDF feature techniques. We used Logistic Regression, Naïve Bayes (NB), and Support Vector Machine (SVM) models for classification through machine learning, and we used Bayesian optimization techniques for hyperparameters tuning of these three models. We also used LSTM, CNN, and transformer models for this research. For the performance evaluation of the classification models, we used four evaluation metrics of precision, recall, F1-score, and accuracy. We also used the ROC-AUC curve to distinguish between all the machine learning and deep learning models. The experimental results expressed that, the transformer model achieved the highest accuracy compared to all the typical machine learning and deep learning models with an accuracy of 87%.
The Rise Of Remote Project Management- A New Norm? - A Survey on IT Organizations in Bangladesh
(Independent University, Bangladesh, 2023-05) Ahamed, Azaz; Khan, Touseef Aziz; Sadman, Nafiz; Hannan, Mahfuz Ibne; Nahar, Nujhat; Hasan, Mahady
The rise of remote work has brought about a significant shift in the way software development projects are managed. With teams spread out across different locations and time zones, project managers must adapt to new challenges to ensure the success of their projects. These challenges include difficulties in communication, coordination, and motivation. It is seen that project managers are using a range of tools and tactics, including agile methodologies, online communication tools, and best practices for remote work, to address these issues. Other strategies may be required to successfully handle remote software development projects as conventional ones are not always sufficient. In this paper, an in-depth and exploratory survey has been conducted on a sample size of 250 employees from various IT organizations in Bangladesh. The results are analyzed to understand the benefits and challenges that come with the Work from Anywhere (WFX) approach to software development projects. The survey data is compared and analyzed against an extensive list of research papers in a similar field and categorized in three dimensions: tools and productivity, work-life balance, and career growth. The results support a strong correlation between WFX with increased productivity and better health.
Late and Early Blight Diseases Identification of Potatoes with a Light Weight Hybrid Transfer Learning Model
(Independent University, Bangladesh, 2023-05) Siddique, Abu Zobayer Bin; Das, Shoibal; Tabassum, Poonam; Tasir, All Moon; Roy, Shovon; Rahman, Md. Saifur; Mridha, M. F.; Islam, Ashraful
Potatoes are one of the world’s most important commodities, and leaf maladies such as early and late blight can substantially reduce their yield and quality. Hence, both farmers and researchers must prioritize quick and precise illness diagnosis. In our research, we propose a strategy based on transfer learning for classifying toxic and diseased potato leaf tissue. We specifically used our dataset of potato leaf photos to fine-tune the Mobile-Net model, which was a pre-trained convolutional neural network. To enhance the model’s functionality, we also added a few more layers. Our study found that, in comparison to other state-of-the-art methods, our methodology outperformed them all by achieving a multi-class classification accuracy of 99%. Our method can be used to detect and monitor potato leaf maladies in real-world situations, which could eventually contribute to enhancing potato productivity and food security
Renewable Energy Optimization: The Role of Solar Tracking Systems
(Independent University, Bangladesh, 2023-06) Hossen, Ishtiaq; Rahman, Nazib; Islam, Rezaul; Zaman, Niaz Murshed Uz; Kabbo, Kawshik Rahman; Azim, Abdul; Uddin, Mohammad Rejwan; Hasan, Mahady
The ecosystem is being harmed using fossil fuels, so renewable energy sources are required to protect the climate for coming generations. The most popular green energy source is solar power, but its effectiveness is impacted by the sun's varying positions. This paper suggests a practical method—a single-axis solar tracking system—for increasing the electricity output of solar panels. Compared to a stationary system, the suggested system can boost a panel's power production according to tests conducted in a variety of weather conditions and positioning. The device uses very little energy and is a practical way to increase the efficiency of solar panels. Thus, putting this method into place can aid in maximizing solar energy and lowering reliance on fossil fuels.