Browsing by Author "Mahmud, Abdullah Al"

Now showing 1 - 13 of 13

An Effective Approach To Address Processing Time and Computational Complexity Employing Modified CCT for Lung Disease Classification
(Scopus, 22-11-22) Khan, Inam Ullah; Azam, Sami; Montaha, Sidratul; Mahmud, Abdullah Al; Rafid, A.K.M. Rakibul Haque; Hasan, Md. Zahid; Jonkman, Mirjam
Early identification and adequate treatment can help prevent lung disorders from becoming chronic, severe, and life-threatening. X-ray images are commonly used and an automated and effective method involving deep learning techniques can potentially contribute to quick and accurate diagnosis of lung disorders. However, in the study of medical imaging using deep learning, two obstacles limit interpretability. One is an insufficient and imbalanced number of training samples in most medical datasets. The other is excessive training time. Although training time can be reduced by decreasing the number of pixels in the images, training with low resolution images tends to result in poor performance. This study represents a solution to overcome these impediments by balancing the number of images and reducing overall processing time while preserving accuracy. The dataset used in this research contains an unequal number of images in the different classes. The quantity of data in the classes is balanced by creating synthetic images based on the patterns and characteristics of the original images, using a Deep Convolutional Generative Adversarial Network (DCGAN). Unwanted regions are removed from the X-ray images, the brightness and contrast of the images are enhanced, and the abnormalities are highlighted by using different artifact removal, noise reduction, and enhancement techniques. We propose a Modified Compact Convolutional Transformer (MCCT) model using 32 × 32 sized images for the categorization of lung disorders into four classes. An ablation study of eleven cases is employed to adjust several hyper parameters and layer topologies. This reduces training time while preserving accuracy. Six transfer learning models, VGG19, VGG16, ResNet152, ResNet50, ResNet50V2, and MobileNet are applied with the same image size the performance is compared with the proposed MCCT model. Our MCCT model records the greatest test accuracy of 95.37%, requiring a short training time, 10-12 s/epoch, whereas the other models only reach near-moderate performance with accuracies ranging from 43% to 79% and training times of 80-90 s/epoch. The robustness of the model with regards to the number of training samples is validated by training the model multiple times reducing the number of training images gradually from 49621 images to 6204 images. Results suggest that even with a smaller dataset, the performance is sustained. Our proposed approach may contribute to an effective CAD based diagnostic system by addressing the issues of insufficient and imbalanced numbers of medical images, excessive training times and low-resolution images.
Performance study of lead-acid battery in a solar car under different traffic and weather conditions
(© 2016 Institute of Electrical and Electronics Engineers Inc., 2016-01) Mahmud, Abdullah Al; Jafar, Imran Bin; Zaman, Asif; Rahman, Mosaddequr Abdur
In this paper performance of lead-acid battery in a solar car, proposed earlier for Dhaka city dwellers, has been studied under different traffic and weather conditions while the car travels in a long route of Dhaka city. A model of the electric drive system of the proposed car, developed in Matlab Simulink, is used to investigate the battery response. It has been observed that the energy required by the car when it travels 30% of the road at the top speed of 60 km/hour is about 65% of the energy required when it travels 90% of the road at that speed, which further decreases with the increase of number of stoppages. Because of this feature, the proposed solar car appears to be an appropriate means of transport in Dhaka, the city with quotidian traffic congestion. In addition of being emission free, it also ensures less energy consumption during heavy traffic congestion compared to a fuel driven car which burns more fuel and releases more toxic gases under the same road conditions.
Performance study of lead-acid battery in a solar car under different traffic and weather conditions
(© 2016 Institute of Electrical and Electronics Engineers Inc., 2016-01) Mahmud, Abdullah Al; Jafar, Imran Bin; Zaman, Asif; Rahman, Mosaddequr Abdur
In this paper performance of lead-acid battery in a solar car, proposed earlier for Dhaka city dwellers, has been studied under different traffic and weather conditions while the car travels in a long route of Dhaka city. A model of the electric drive system of the proposed car, developed in Matlab Simulink, is used to investigate the battery response. It has been observed that the energy required by the car when it travels 30% of the road at the top speed of 60 km/hour is about 65% of the energy required when it travels 90% of the road at that speed, which further decreases with the increase of number of stoppages. Because of this feature, the proposed solar car appears to be an appropriate means of transport in Dhaka, the city with quotidian traffic congestion. In addition of being emission free, it also ensures less energy consumption during heavy traffic congestion compared to a fuel driven car which burns more fuel and releases more toxic gases under the same road conditions.
Performance study of lead-acid battery in a solar car under different traffic and weather conditions
(© 2016 Institute of Electrical and Electronics Engineers Inc., 2016-01) Mahmud, Abdullah Al; Jafar, Imran Bin; Zaman, Asif; Rahman, Mosaddequr Abdur
In this paper performance of lead-acid battery in a solar car, proposed earlier for Dhaka city dwellers, has been studied under different traffic and weather conditions while the car travels in a long route of Dhaka city. A model of the electric drive system of the proposed car, developed in Matlab Simulink, is used to investigate the battery response. It has been observed that the energy required by the car when it travels 30% of the road at the top speed of 60 km/hour is about 65% of the energy required when it travels 90% of the road at that speed, which further decreases with the increase of number of stoppages. Because of this feature, the proposed solar car appears to be an appropriate means of transport in Dhaka, the city with quotidian traffic congestion. In addition of being emission free, it also ensures less energy consumption during heavy traffic congestion compared to a fuel driven car which burns more fuel and releases more toxic gases under the same road conditions.
Power pilferage in Bangladesh: a simple and economically affordable solution
(© 2015 Institute of Electrical and Electronics Engineers Inc., 2016-03) Ahmed, Sadita; Mahmud, Abdullah Al; Rahman, Mosaddequr
In Bangladesh, illegal consumption of electricity takes a considerable part of the revenue as a non-technical loss in power distribution system. Unlike technical losses, this has no easy way to be calculated and fixed. This paper discusses the methodology for detection of power pilferage and its implementation using embedded electronics system. The idea is to detect unauthorized tapping on distribution lines by comparing the energy transferred through the energy meter of a pole and the summation of energy consumed through energy meters of all premises connected to that pole. If these two data differ by an alarming measure, the respective authority will be notified. The communication between pole and premises and between pole and authority have been done wirelessly in the lowest possible cost considering the economic condition of the developing country. The cost analysis shows that if the system is implemented, total power system loss will reduce at least 15% of the current loss.
Power pilferage in Bangladesh: a simple and economically affordable solution
(© 2015 Institute of Electrical and Electronics Engineers Inc., 2016-03) Ahmed, Sadita; Mahmud, Abdullah Al; Rahman, Mosaddequr
In Bangladesh, illegal consumption of electricity takes a considerable part of the revenue as a non-technical loss in power distribution system. Unlike technical losses, this has no easy way to be calculated and fixed. This paper discusses the methodology for detection of power pilferage and its implementation using embedded electronics system. The idea is to detect unauthorized tapping on distribution lines by comparing the energy transferred through the energy meter of a pole and the summation of energy consumed through energy meters of all premises connected to that pole. If these two data differ by an alarming measure, the respective authority will be notified. The communication between pole and premises and between pole and authority have been done wirelessly in the lowest possible cost considering the economic condition of the developing country. The cost analysis shows that if the system is implemented, total power system loss will reduce at least 15% of the current loss.
Power pilferage in Bangladesh: a simple and economically affordable solution
(© 2015 Institute of Electrical and Electronics Engineers Inc., 2016-03) Ahmed, Sadita; Mahmud, Abdullah Al; Rahman, Mosaddequr
In Bangladesh, illegal consumption of electricity takes a considerable part of the revenue as a non-technical loss in power distribution system. Unlike technical losses, this has no easy way to be calculated and fixed. This paper discusses the methodology for detection of power pilferage and its implementation using embedded electronics system. The idea is to detect unauthorized tapping on distribution lines by comparing the energy transferred through the energy meter of a pole and the summation of energy consumed through energy meters of all premises connected to that pole. If these two data differ by an alarming measure, the respective authority will be notified. The communication between pole and premises and between pole and authority have been done wirelessly in the lowest possible cost considering the economic condition of the developing country. The cost analysis shows that if the system is implemented, total power system loss will reduce at least 15% of the current loss.
Recruitment and selection process of PRAN Group
(Department of Business and Technology Management(BTM), Islamic University of Technology(IUT), Board Bazar, Gazipur-1704, Bangladesh, 2022-04-30) Mahmud, Abdullah Al
FMCG sector is also one of the career-oriented sectors which really attract me as an aspiring student stepping into the job sector. At part of the Internship program, I am placed in the Human Resource Division of Pran Group. In the report I have studied ‘Recruitment and Selection process of Pran Group and attempted to provide some ways so as to make recruitment more effective. The recruitment and selection decision are of prime importance as it is the medium for obtaining the best possible person to job fit that will contribute significantly towards the company’s effectiveness. I am privileged to be one of the students who got an opportunity to do my internship from PRAN Group. Since I have done my major in Technology Management, this internship will help me hone my practical skills and obtain knowledge in this field for my career development.
SkinNet-14: a deep learning framework for accurate skin cancer classification using low-resolution dermoscopy images with optimized training time
(Scopus, 2024-08-01) Mahmud, Abdullah Al; Azam, Sami; Khan, Inam Ullah; Montaha, Sidratul; Karim, Asif; Haque, Aminul; Hasan, Md. Zahid; Brady, Mark; Biswas, Ritu; Jonkman, Mirjam
The increasing incidence of skin cancer necessitates advancements in early detection methods, where deep learning can be beneficial. This study introduces SkinNet-14, a novel deep learning model designed to classify skin cancer types using low-resolution dermoscopy images. Unlike existing models that require high-resolution images and extensive training times, SkinNet-14 leverages a modified compact convolutional transformer (CCT) architecture to effectively process 32 × 32 pixel images, significantly reducing the computational load and training duration. The framework employs several image preprocessing and augmentation strategies to enhance input image quality and balance the dataset to address class imbalances in medical datasets. The model was tested on three distinct datasets—HAM10000, ISIC and PAD—demonstrating high performance with accuracies of 97.85%, 96.00% and 98.14%, respectively, while significantly reducing the training time to 2–8 s per epoch. Compared to traditional transfer learning models, SkinNet-14 not only improves accuracy but also ensures stability even with smaller training sets. This research addresses a critical gap in automated skin cancer detection, specifically in contexts with limited resources, and highlights the capabilities of transformer-based models that are efficient in medical image analysis.
Skinnet-14: A Deep Learning Framework for Accurate Skin Cancer Classification Using Low-resolution Dermoscopy Images with Optimized Training Time
(Springer, 2024-08-15) Mahmud, Abdullah Al; Azam, Sami; Khan, Inam Ullah; Montaha, Sidratul; Karim, Asif; Haque, Aminul; Hasan, Md. Zahid; Brady, Mark; Biswas, Ritu; Jonkman, Mirjam
The increasing incidence of skin cancer necessitates advancements in early detection methods, where deep learning can be beneficial. This study introduces SkinNet-14, a novel deep learning model designed to classify skin cancer types using low-resolution dermoscopy images. Unlike existing models that require high-resolution images and extensive training times, SkinNet-14 leverages a modified compact convolutional transformer (CCT) architecture to effectively process 32 × 32 pixel images, significantly reducing the computational load and training duration. The framework employs several image preprocessing and augmentation strategies to enhance input image quality and balance the dataset to address class imbalances in medical datasets. The model was tested on three distinct datasets—HAM10000, ISIC and PAD—demonstrating high performance with accuracies of 97.85%, 96.00% and 98.14%, respectively, while significantly reducing the training time to 2–8 s per epoch. Compared to traditional transfer learning models, SkinNet-14 not only improves accuracy but also ensures stability even with smaller training sets. This research addresses a critical gap in automated skin cancer detection, specifically in contexts with limited resources, and highlights the capabilities of transformer-based models that are efficient in medical image analysis.
Skinnet-14: A Deep Learning Framework for Accurate Skin Cancer Classification Using Low-resolution Dermoscopy Images with Optimized Training Time
(Springer Nature, 2024-08-01) Mahmud, Abdullah Al; Azam, Sami; Khan, Inam Ullah; Montaha, Sidratul; Karim, Asif; Haque, Aminul; Hasan, Md. Zahid; Brady, Mark; Biswas, Ritu; Jonkman, Mirjam
The increasing incidence of skin cancer necessitates advancements in early detection methods, where deep learning can be beneficial. This study introduces SkinNet-14, a novel deep learning model designed to classify skin cancer types using low-resolution dermoscopy images. Unlike existing models that require high-resolution images and extensive training times, SkinNet-14 leverages a modified compact convolutional transformer (CCT) architecture to effectively process 32 × 32 pixel images, significantly reducing the computational load and training duration. The framework employs several image preprocessing and augmentation strategies to enhance input image quality and balance the dataset to address class imbalances in medical datasets. The model was tested on three distinct datasets—HAM10000, ISIC and PAD—demonstrating high performance with accuracies of 97.85%, 96.00% and 98.14%, respectively, while significantly reducing the training time to 2–8 s per epoch. Compared to traditional transfer learning models, SkinNet-14 not only improves accuracy but also ensures stability even with smaller training sets. This research addresses a critical gap in automated skin cancer detection, specifically in contexts with limited resources, and highlights the capabilities of transformer-based models that are efficient in medical image analysis.
The Environment of E-Commerce in Bangladesh
(Daffodil International University, 2007-07-01) Debnath, Nitai Chandra; Mahmud, Abdullah Al
Electronic commerce is rapidly growing as an impressive manifestation of globalization. The rapid expansion of e-commerce is a major opportunity for local and international trade development of LDCs including Bangladesh. However, infrastructure, culture, and attitude are significant barriers to e-commerce. Bangladesh government has gone forward except minor lacking to create an e- commerce friendly environment. One thing to be noted is that success in e- commerce does not depend on access and connectivity alone. Information Technology also depends on the fortitude of the entrepreneur of taking advantage of the e-commerce revolution. Proper awareness building programs among the people and business community are required to establish, maintain and growth of e- commerce. This study concentrates on assessing the elements of the environment relating to e-commerce to get the best picture of the prevailing condition and then recommending some areas of improvement on the basis of the assessment for deploying e-commerce in Bangladesh.
What is relevant in a text document a machine learning based approach
(BRAC University, 2021-06) Mahmud, Abdullah Al; Noor, Jannat-E; Reshad, Sadman Alam; Fuad, Syed Nafis; Chakrabarty, Amitabha; Alam, Md. Golam Rabiul
Text Documents often contain valuable data. But not all data is relevant. That is why extracting relevant data from text documents is an essential task. Extracting relevant data from text documents refers to the study of classifying text documents into such groups that describe the contents of documents. There are many methods to find out relevant data from a cluster of text or a text document. Classifying extensive textual data helps to organize the records better, make the search easier and relevant and simplify navigation. That makes this task still an open research issue. This paper uses three techniques of classifying text documents: convolution neural networks (CNN) with deep learning, Gaussian Na¨ıve Bayes and support vector machines (SVM). With these three algorithms, the text we want to classify goes through three layers of checks. So, it gives us more reliability.