Browsing by Author "Rahman, Md. Sadekur"

Now showing 1 - 18 of 18

A Context-Sensitive Approach to Find Optimum Language Model for Automatic Bangla Spelling Correction
(nternational Journal of Advanced Computer Science and Applications, 2018) Islam, Muhammad Ifte Khairul; Habib, Md. Tarek; Rahman, Md. Sadekur; Rahman, Md. Riazur; Ahmed, Farruk
Automated spelling correction is an important phenomenon in typing that has intense effect on aiding both literate and semi-literate people while using keyboard or other similar devices. Such automated spelling correction technique also helps students significantly in learning process through applying proper words during word processing. A lot of work has been conducted for English language, but for Bangla, it is still not adequate. All work done so far in Bangla is context-free. Bangla is one of the mostly spoken languages (3.05% of world population) and considered seventh language of all languages in the world. In this paper, we propose a context-sensitive approach for automated spelling correction in Bangla. We make combined use of edit distance and stochastic, i.e. N-gram language model. We use six N-gram models in total. A novel approach is deployed in order to find the optimum language model in terms of performance. In addition, for finding out better performance, a large Bangla corpus of different word types is used. We have achieved a satisfactory and promising accuracy of 87.58%.
A Method for Bengali Author Detection Using State of the Arts Supervised Machine Learning Classifiers
(Springer, 2023-09-15) Hamid, Md. Abdul; Marjana, Nusrat Jahan; Tumpa, Eteka Sultana; Khan, Md. Rafidul Hasan; Afroz, Umme Sanzida; Rahman, Md. Sadekur
Text classification is an important topic of study in the area of natural language Processing. To identify the authorship of the provided Bangla text, we create a model using the State of Arts Supervised method. Because our work is a multi-class categorization, we may use it to determine who wrote articles, news, emails, or messages. It can also use to find ghostwriters, identify anonymous authors, and detect plagiarism. This article focuses on the categorization of five Bengali authors. They are well-known writers in Bengali literature and poetry. Humayun Ahmed, Rabindranath Tagore, Muhammad Zafar Iqbal, Kazi Nazrul Islam, and Sarat Chandra Chattopadhyay are the five writers. Data were manually collected from various sources in the novels or books of these five writers, and we contained over 4500 paragraphs. A completely new dataset is created for the experimental evaluation. We preprocess Bengali text for training reasons. Logistic regression, naive Bayes, decision trees, support vector machines, random forests, XG-Boost, and K-nearest neighbor are among the seven classification methods employed. In our experiment, the Support Vector Machine produces the best experimental classification report. Support vector machine gives 82% model accuracy.
An Investigative Design of Optimum Stochastic Language Model for Bangla Autocomplete
(Indonesian Journal of Electrical Engineering and Computer Science, 2019) Eyamin, Md.Iftakher Alam; Habib, Md. Tarek; Muhammad Ifte Khairul Islam; Rahman, Md. Sadekur; Khan, Md. Abbas Ali
Word completion and word prediction are two important phenomena in typing that have extreme effect on aiding disable people and students while using keyboard or other similar devices. Such autocomplete technique also helps students significantly during learning process through constructing proper keywords during web searching. A lot of works are conducted for English language, but for Bangla, it is still very inadequate as well as the metrics used for performance computation is not rigorous yet. Bangla is one of the mostly spoken languages (3.05% of world population) and ranked as seventh among all the languages in the world. In this paper, word prediction on Bangla sentence by using stochastic, i.e. N-gram based language models are proposed for autocomplete a sentence by predicting a set of words rather than a single word, which was done in previous work. A novel approach is proposed in order to find the optimum language model based on performance metric. In addition, for finding out better performance, a large Bangla corpus of different word types is used.
An IoT-Based System to Measure Methane and Carbon Dioxide Emissions Along with Temperature and Humidity in Urban Areas
(Springer, 2022-10-14) Afroz, Umme Sanzida; Khan, Md. Rafidul Hasan; Rahman, Md. Sadekur; Jahan, Israt
The increment of the world population is the main reason behind the emission of greenhouse gasses like methane and carbon dioxide which affect temperature and humidity. To identify this problem, we implemented an IoT system using an ESP32 microcontroller and the necessary sensor to measure data from the outdoor environment and deploy our system to monitor the outdoor temperature and humidity quality due to increasing the level of methane and carbon dioxide. In the next step, the measured data was sent into the ThingSpeak cloud server and later was made available for visualization. This research work seeks to estimate the level of methane and carbon dioxide emissions which has a great impact on increasing temperature and humidity and also highlights some complications to integrating modern sensors based on the emergence of the Internet of things (IoT) concept. However, the data preserved in the cloud server can be used for further analysis to find other impacts of gasses in our environment.
Breaking the Fake News Barrier: Deep Learning Approaches in Bangla Language
(Scopus, 2024-06-24) Mondal;, Pronoy Kumar; Khan;, Sadman Sadik; Rana, Md. Masud; Ramit, Shahriar Sultan; Sattar;, Abdus; Rahman, Md. Sadekur
The rapid development of digital stages has greatly compounded the dispersal of untrue data, dissolving certainty and judgment in society, especially among the Bengalispeaking community. Our ponder addresses this critical issue by presenting an interesting strategy that utilizes a profound learning innovation, particularly the Gated Repetitive Unit (GRU), to recognize fake news within the Bangla dialect. The strategy of our proposed work incorporates intensive information preprocessing, which includes lemmatization, tokenization, and tending to course awkward nature by oversampling. This comes about in a dataset containing 58,478 passages. We appreciate the creation of a demonstration based on GRU (Gated Repetitive Unit) that illustrates remarkable execution with a noteworthy precision rate of 94%. This ponder gives an intensive clarification of the methods included in planning the information, selecting the show, preparing it, and assessing its execution. The performance of the model is investigated by reliable metrics like precision, recall, F1 score, and accuracy. The commitment of the work incorporates making a huge fake news dataset in Bangla and a demonstration that has outperformed other Bangla fake news location models.
Convolutional Neural Network Based Partial Face Detection
(Daffodil International University, 2022-08-01) Islam, Md. Towfiqul; Ahmed, Tanzim; Rashid, A.B.M. Raihanur; Islam, Taminul; Rahman, Md. Sadekur; Habib, Md. Tarek
Due to the massive explanation of artificial intelligence, machine learning technology is being used in various areas of our day-to-day life. In the world, there are a lot of scenarios where a simple crime can be prevented before it may even happen or find the person responsible for it. A face is one distinctive feature that we have and can differentiate easily among many other species. But not just different species, it also plays a significant role in determining someone from the same species as us, humans. Regarding this critical feature, a single problem occurs most often nowadays. When the camera is pointed, it cannot detect a person’s face, and it becomes a poor image. On the other hand, where there was a robbery and a security camera installed, the robber’s identity is almost indistinguishable due to the low-quality camera. But just making an excellent algorithm to work and detecting a face reduces the cost of hardware, and it doesn’t cost that much to focus on that area. Facial recognition, widget control, and such can be done by detecting the face correctly. This study aims to create and enhance a machine learning model that correctly recognizes faces. Total 627 Data have been collected from different Bangladeshi people's faces on four angels. In this work, CNN, Harr Cascade, Cascaded CNN, Deep CNN & MTCNN are these five machine learning approaches implemented to get the best accuracy of our dataset. After creating and running the model, Multi-Task Convolutional Neural Network (MTCNN) achieved 96.2% best model accuracy with training data rather than other machine learning models.
EduBot
(Springer, 2020-07-19) Al Muid, Md. Abdullah; Reza, Md. Masum; Kalim, Ruhullah Bin; Ahmed, Nayeem; Habib, Md. Tarek; Rahman, Md. Sadekur
In the present time Chatbot is an essential tool used by many organizations to provide services to their targeted customers round the clock. This research focuses on a domain-specific Chatbot that can be helpful for educational institutes. This Chatbot will be a virtual (representation) to the admission seekers. It will provide answers regarding the university, its departments, admission fees and other admission related FAQ. For the sake of the research, frequently asked questions of a university were collected and an unsupervised learning model along with natural language processing techniques was deployed to answer the questions of the admission candidates. Tokenization, stop words removal followed by vectorization were implemented for preprocessing the training data. User’s inputs were similarly processed and then tf-idf based cosine similarity applied to retrieve the best answer. Later, a user-centric evaluation metric was used to evaluate the model and as per the metric, our current model showed approximately 80% accuracy.
EduBot:
(Springer, 2021) Muid, Md. Abdullah Al; Reza, Md. Masum; Kalim, Ruhullah Bin; Ahmed, Nayeem; Habib, Md. Tarek; Rahman, Md. Sadekur
In the present time Chatbot is an essential tool used by many organizations to provide services to their targeted customers round the clock. This research focuses on a domain-specific Chatbot that can be helpful for educational institutes. This Chatbot will be a virtual (representation) to the admission seekers. It will provide answers regarding the university, its departments, admission fees and other admission related FAQ. For the sake of the research, frequently asked questions of a university were collected and an unsupervised learning model along with natural language processing techniques was deployed to answer the questions of the admission candidates. Tokenization, stop words removal followed by vectorization were implemented for preprocessing the training data. User’s inputs were similarly processed and then tf-idf based cosine similarity applied to retrieve the best answer. Later, a user-centric evaluation metric was used to evaluate the model and as per the metric, our current model showed approximately 80% accuracy.
Human Activity Recognition Using DTW Algorithm
(Proceedings of 2019 IEEE Region 10 Symposium, IEEE, 2019-06-09) Masnad, Mohshi; Hasan, G. M. Mukit; Iftekhar, Kazi Md.; Rahman, Md. Sadekur
Human activity recognition is now a well-known field of Human Computer Interaction (HCI) because of its capability to provide personalized support using different applications. For the purpose of recognizing human activities, we selected three activities (running, walking, and steady state, e.g., sitting and lying). We used the Dynamic Time Warping (DTW) algorithm as a classifier to learn and detect activities. Due to its inherent nature, DTW can provide satisfactory accuracy even with very few training samples. Using smartphone's gyroscope and accelerometer sensors, we recorded user data during various activities. To encounter personal traits, we made sure the users were of different age, height and gender. With the help of DTW as a real time classifier, we then identify the activities against matching templates. The obtained results showed sufficient accuracy, showing the effectiveness of the approach.
Identifying Genre of a Book from Its Summary Using Machine Learning Approach
(Springer Nature, 2022-10-14) Bhuiyan, Saidur Rahman; Khan, Md. Rafidul Hasan; Afroz, Umme Sanzida; Rahman, Md. Sadekur
Categorical search and category-wise book recommendation are two common tasks for online booksellers. But for a machine to understand this category from a given text is still challenging work, where machine learning is a widely used tool at present. Though in the English language, with the availability of rich datasets and corpus, machine learning-based categorization and recommendation have reached a standard level, in the Bengali language, to reach the standard, still needs a long way to go. One key reason is the lack of availability of a rich Bengali dataset. The aim of this research was to make a dataset first for the book’s genre identification from its given summary and to explore which supervised classifier performed best on that dataset for classifying the genres. Before that, we performed several essential preprocessing steps essential to prepare our dataset fit for the algorithms. Six machine learning classifiers were applied to the dataset, and it was observed that Naive Bayes performed best with an accuracy of 68% followed by XGB with an accuracy of 67%.
Lung Disease Classification Using Deep Learning Models from Chest X-ray Images
(IEEE, 2023-03-27) Sultana, Salma; Pramanik, Anik; Rahman, Md. Sadekur
In the very recent past, Infectious disease-related sickness has long posed a concern on a global scale. Each year, COVID-19, pneumonia, and tuberculosis cause a large number of deaths because they all affect the lungs. Early detection and diagnosis can increase the likelihood of receiving quality treatment in all circumstances. A low-cost, simple imaging approach called chest X-ray imaging enables to detection and screen lung abnormalities brought on by infectious diseases for example Covid-19, pneumonia, and tuberculosis. This paper provided a thorough analysis of current deep-learning methods for diagnosing Covid-19, pneumonia, and TB. According to the research papers reviewed, Deep Convolutional Neural Network is the most used deep learning method for identifying Covid-19, pneumonia, and TB from chest X-ray (CXR) images. We compared the proposed DNN to well-known DNNs like Efficient-NetB0, DenseNet169, and DenseNet201 in order to more accurately assess how well it performed. Our findings are equivalent to the state-of-the-art, and since the proposed CNN is lightweight, it may be employed for widespread screening in areas with limited resources. From three diverse publicly accessible datasets merged into one dataset, the suggested DNN generated the following precisions for that dataset: 99.15%, 98.89%, and 97.79% for EfficientNetB0, DenseNet169, and DenseNet201 respectively. The proposed network can help radiologists make quick and accurate diagnoses because it is effective at identifying COVID-19 and other lung contagious disorders utilizing chest X-ray images. This paper also gives young scientists a good insight into how to create CNN models that are highly efficient when used with medical images to identify diseases early.
MKRF Stacking-Voting
(Daffodil International University, 2022-07-19) Bijoy, Md. Hasan Imam; Pramanik, Anik; Rahman, Md. Sadekur; Hasan, Mehedi; Akhi, Sumiya Alam; Rahman, Md. Mahbubur
Data mining is most efficient when used deliberately to achieve a corporate goal, answer business or research questions, or contribute to a problem-solving solution. Data mining aids in the accurate prediction of outcomes, the recognition of patterns and anomalies, and frequently inform forecasts. Online education is becoming more popular all around the world because of the COVID-19 pandemic. The main goal of this research is to Predict Educational Satisfaction Level of Bangladeshis Students During the Pandemic using data mining approaches by only filling up with some basic questionnaires which are related to the satisfaction level of online education collected through a public survey. By surveying 1004 students from various academic institutions, schools, colleges, and universities on the quality of online education in COVID-19 pandemic scenarios, we were able to determine how productive it would be. Influence how online learning is measured and how satisfied people are with it. To achieve our aim of predicting satisfaction levels, we used a total of eight classifiers, six of which were based classifiers, which we combined with the best three top-scoring classifiers to build a novel ensemble approach called MKRF Stacking and MKRF Voting ensemble classifier. Among those classifiers, the Random Forest classifier outperforms the other six base classifiers with 97.21% accuracy. Our proposed data mining ensemble approaches MKRF Stacking and MKRF Voting outperform applied classifiers. Typically, voting ensemble classifiers outperform voting ensemble classifiers, but in this case, MKRF Stacking defeated MKRF Voting and all applied classifiers with a supreme accuracy of 97.68% (Average). The proposed method would be used in a framework where education counselors find the root causes and minor explanations for dissatisfaction in online education among students so that they can better understand all aspects and provide them with the best advice and solutions to their problem...
Monkeypox Detection from Various Types of Poxes: A Deep Learning Approach
(IEEE, 2023-05-23) Pramanik, Anik; Chowdhury, Fayazunnesa; Sultana, Salma; Rahman, Md. Mahbubur; Imam, Md. Hasan; Bijoy, Md. Hasan Imam; Rahman, Md. Sadekur
According to World Health Organization (WHO) statistics, monkeypox has been identified as an epidemic in 127 nations so far, and it is spreading quickly over the globe. While the rashes and skin lesions associated with monkeypox usually mimic those of other poxes, including chickenpox and measles. Due to these similarities, it could be challenging for medical professionals to identify monkeypox based just on the appearance of lesions and rashes. Because monkeypox was uncommon before in the current outbreak, healthcare professionals lack knowledge in this area. But the scientific community has demonstrated a rising interest in implementing Artificial Intelligence in Monkeypox prediction and detection from digital skin images as a result of the success of image processing approaches in COVID-19 detection. In this study, we have applied three cutting-edge deep learning models which are InceptionV3, MobileNetV3, and DenseNet201, referred to as transfer learning models, to detect monkeypox on skin images using the publicly available Monkeypox Skin Image Dataset 2022 with four classes. According to our research, transfer learning models can detect monkeypox with a top 93.59% accuracy for the InceptionNet-V3 pre-trained model from three implemented algorithms on digitized skin images. For further research, larger training images are required to train those deep learning models to achieve a higher vigorous detection rate.
Optimizing Vehicle Insurance Processing through Advanced Deep Learning Models
(IEEE, 2024-11-04) Khan, Sadman Sadik; Rupak, Afraz Ul Haque; Rahman, Md. Sadekur
This paper presents a novel method for predicting insurance claims by utilizing artificial intelligence, specifically a deep learning model within the Convolutional Neural Network (CNN) framework. The purpose is to automate and simplify the insurance claims process. The model utilizes computer vision technology to precisely detect and identify car damage, resulting in a substantial reduction in the processing time for insurance claims. The article describes the creation of a specialized dataset consisting of actual photographs of vehicles that have been wrecked. It also assesses multiple deep learning models, ultimately finding that InceptionV3 is the most successful, achieving an accuracy rate of 97%. The suggested AI system seeks to improve the efficiency of claims processing and decrease the need for human evaluation. This would provide a more dependable and efficient method for handling automobile insurance claims after accidents.
Performance Evaluation of YOLO Models for Detecting Bangladeshi License Plates
(2024-10-04) Ramit, Shahriar Sultan; Alo, Alaya Parvin; Shaqib, Sm; Khan, Sadman Sadik; Rupak, Afraz Ul Haque; Rahman, Md. Sadekur
This research paper presents a comprehensive investigation into the effectiveness of YOLO (You Only Look Once) models, namely YOLOv5, YOLOv7, and YOLOv8, in the domain of Bangladeshi license plate detection. With the escalating demand for precise license plate recognition systems to facilitate efficient traffic management and bolster law enforcement efforts, this study conducts an in-depth evaluation of these models. Central to our methodology is the development of a specialized dataset comprising Bangladeshi license plate images, reflecting the unique characteristics and challenges prevalent in this geographical context. Through meticulous dataset curation, model training, and rigorous testing procedures, we ascertain the performance metrics of each YOLO variant. Notably, our findings reveal YOLOv8 as the most proficient model, achieving a remarkable mean average precision (mAP) score of 0.934 with precision 0.93 and recall 0.906. The insights gleaned from this research contribute significantly to the advancement of intelligent transportation systems and public safety initiatives in Bangladesh, offering tailored solutions for license plate detection challenges in this specific locale.
Prediction of Addiction to Drugs and Alcohol Using Machine Learning
(International Journal of Electrical and Computer Engineering, 2021) Arif, Md. Ariful Islam; Sany, Saiful Islam; Sharmin, Farah; Rahman, Md. Sadekur; Habib, Md. Tarek
Nowadays addiction to drugs and alcohol has become a significant threat to the youth of the society as Bangladesh’s population. So, being a conscientious member of society, we must go ahead to prevent these young minds from life-threatening addiction. In this paper, we approach a machine learning-based way to forecast the risk of becoming addicted to drugs using machine-learning algorithms. First, we find some significant factors for addiction by talking to doctors, drug-addicted people, and read relevant articles and write-ups. Then we collect data from both addicted and no addicted people. After preprocessing the data set, we apply nine conspicuous machine learning algorithms, namely k-nearest neighbors, logistic regression, SVM, naïve Bayes, classification, and regression trees, random forest, multilayer perception, adaptive boosting, and gradient boosting machine on our processed data set and measure the performances of each of these classifiers in terms of some prominent performance metrics. Logistic regression is found outperforming all other classifiers in terms of all metrics used by attaining an accuracy approaching 97.91%. On the contrary, CART shows poor results of an accuracy approaching 59.37% after applying principal component analysis.
SweetSight: A Deep Convolutional Neural Network Approach for Automatic Categorization of Bengal Sweets
(Scopus, 2024-08-20) Supriya, Soummo; Rimi, Iffat Firozy; Islam, Md. Moinul; Rahman, Md. Sadekur; Nawshin, Samia; Habib, Md. Tarek
The manufacture of a wide variety of sweets is on the rise in the entire Bengal (both Bangladesh and West Bengal). As a consequence, the sweet’s name escapes the vast majority of individuals in our country. Computer vision advancements have made object recognition from photos easier in recent years. Using computer vision to automatically categorize sweets is still a challenge because of the similarity between various sorts and characteristics such as their placement or lighting conditions. Classifying sweets may be useful in a variety of domains, including autonomous economic robots and the creation of mobile apps for identifying certain sweets on the market. In this article, we employed deep convolutional neural network (DCCN) methods to evaluate five alternative models for sweet detection. The endemic Bengali delicacies we used to train my model included Inception-v3, ResNet-50, VGG15, AlexNet, and CNN. This model was efficient. Our dataset comprised images of confections from thirteen distinct sweet categories. Two portions of the dataset were separated: 80% for training and 20% for testing. The training dataset was enhanced and increased to make preparation simpler. Using the Inception-v3 model, we were able to attain a 100% accuracy rate with our dataset.
SweetSight: A Deep Convolutional Neural Network Approach for Automatic Categorization of Bengal Sweets
(Springer Nature, 2024-08-20) Supriya, Soummo; Rimi, Iffat Firozy; Moinul Islam, Md.; Rahman, Md. Sadekur; Nawshin, Samia; Habib, Md. Tarek
The manufacture of a wide variety of sweets is on the rise in the entire Bengal (both Bangladesh and West Bengal). As a consequence, the sweet’s name escapes the vast majority of individuals in our country. Computer vision advancements have made object recognition from photos easier in recent years. Using computer vision to automatically categorize sweets is still a challenge because of the similarity between various sorts and characteristics such as their placement or lighting conditions. Classifying sweets may be useful in a variety of domains, including autonomous economic robots and the creation of mobile apps for identifying certain sweets on the market. In this article, we employed deep convolutional neural network (DCCN) methods to evaluate five alternative models for sweet detection. The endemic Bengali delicacies we used to train my model included Inception-v3, ResNet-50, VGG15, AlexNet, and CNN. This model was efficient. Our dataset comprised images of confections from thirteen distinct sweet categories. Two portions of the dataset were separated: 80% for training and 20% for testing. The training dataset was enhanced and increased to make preparation simpler. Using the Inception-v3 model, we were able to attain a 100% accuracy rate with our dataset.