Browsing by Author "Das, Aka"

Now showing 1 - 5 of 5

KGR-Rainfall: Temperature-Based Rainfall Prediction in Bangladesh with Novel KGR Stacking Ensemble
(IEEE, 2023-07-06) Bitto, Abu Kowshir; Rubi, Maksuda Akter; Bijoy, Md. Hasan Imam; Shuvo, Subrata Das; Das, Aka; Chowdhury, Amit
Climate change factors such as wet or dry, cold or warm seasons have a significant impact on both the economy and culture. Extreme rainfall events have historically posed a major threat to many parts of the world. In Bangladesh, during monsoon seasons, wet southern airflows from the Bay of Bengal collide with dry mainland air, causing heavy rainfall that negatively affects various socio-economic sectors. These include agriculture, food production, urban planning, energy, water resource management, fisheries, forest management, healthcare, disaster management, transportation, tourism, sports, and leisure. To address this issue, the paper proposes a machine-learning approach to forecast rainfall in Bangladesh using multiple regression models and a novel Stacked Ensemble Model (KGR Stacking). The study also investigates the relationship between rainfall and temperature. The KGR Stacking model outperforms the other 12 regression models, achieving an accuracy of 86.43% and lower error.
MobileNet-Eye: An Efficient Transfer Learning for Eye Disease Classification
(2024-04-22) Niloy, Golam Mohiuddin; Bitto, Abu Kowshir; Biplob, Khalid Been Md. Badruzzaman; Sammak, Musabbir Hasan; Das, Aka; Hridoy, Golam Gouse
Precisely and promptly diagnosing eye illnesses is crucial for preventing and managing them. Transfer learning shows potential for automatically identifying different eye diseases. This is beneficial for avoiding and addressing eye issues. Enhanced computer vision has significantly benefited eye doctors by allowing computers to assist them extensively. We researched transfer learning strategies to address three eye issues: uveitis, Eyelid (Lid), and Healthy eyes. We examined the performance, precision, and efficacy of 3 popular pre-trained computer algorithms (MobileNetV2, ResNet50, EfficientNetB7) in detecting eye disorders. We utilized 3,000 images of eyes for this task: 1000 images of healthy eyes, 1,000 images of eyes affected by uveitis, and 1,000 images of eyes with Lid problems. MobileNetV2 was the most precise model, achieving a 96% accuracy in detecting eye disorders. EfficientNet-B7 achieved a 95% accuracy, whereas ResNet-50 had a 94% accuracy.
Potato-Net: Classifying Potato Leaf Diseases Using Transfer Learning Approach
(Springer Nature, 2023-06-11) Bitto, Abu Kowshir; Bijoy, Md.Hasan Imam; Das, Aka; Rahman, Md.Ashikur; Rabbani, Masud
Research on pertinent topics is more important than ever for the long-term development of agriculture, given the advancements in contemporary farming and use of artificial intelligence (AI) for identifying crop illnesses. There are numerous diseases, and they all significantly affect the amount and quality of potatoes. Early and automated detection of these illnesses during the budding phase can assist increase the output of potato crops, but it requires a high level of ability. Several models have already been created to identify various plant diseases. In this study, we use a variety of convolutional neural network designs to recognize potato leaf disease and assess their early detection accuracy against that of other researchers’ work. The learning sample for our algorithm included both the original and enhanced photos, as a learning option. The model was then evaluated to ensure that it was accurate. After being trained on the dataset for the potato leaf disease using the Inception-v3, Xception, and ResNet50 models, the model’s performance was evaluated using test images. ResNet50 has the highest accuracy and lowest error rate for detecting potato leaf disease, followed by Inception-v3 with an accuracy of nighty four point two five percent (94.25%) and Xception with an accuracy of eighty-nine point seven one percent (89.71%).
Sentiment Analysis From Bangladeshi Food Delivery Startup Based on User Reviews Using Machine Learning and Deep Learning
(Institute of Advanced Engineering and Science (IAES), 2023-08-15) Bitto, Abu Kowshir; Bijoy, Md. Hasan Imam; Arman, Md. Shohel; Mahmud, Imran; Das, Aka; Majumder, Joy
Food delivery methods are at the top of the list in today's world. People's attitudes toward food delivery systems are usually influenced by food quality and delivery time. We did a sentiment analysis of consumer comments on the Facebook pages of Food Panda, HungryNaki, Pathao Food, and Shohoz Food, and data was acquired from these four sites’ remarks. In natural language processing (NLP) task, before the model was implemented, we went through a rigorous data pre-processing process that included stages like adding contractions, removing stop words, tokenizing, and more. Four supervised classification techniques are used: extreme gradient boosting (XGB), random forest classifier (RFC), decision tree classifier (DTC), and multi nominal Naive Bayes (MNB). Three deep learning (DL) models are used: convolutional neural network (CNN), long term short memory (LSTM), and recurrent neural network (RNN). The XGB model exceeds all four machine learning (ML) algorithms with an accuracy of 89.64%. LSTM has the highest accuracy rate of the three DL algorithms, with an accuracy of 91.07%. Among ML and DL models, LSTM DL takes the lead to predict the sentiment.
Suicidal Ratio Prediction Among the Continent of World: A Machine Learning Approach
(IEEE, 2023-07-06) Biplob, Khalid Been Badruzzaman; Bijoy, Md. Hasan Imam; Bitto, Abu Kowshir; Das, Aka; Chowdhry, Amit; Hossain, Sayed Md. Minhaz
Suicide is a global health issue with significant negative effects. Individuals at risk of suicide often avoid seeking help due to stigma or fear of forced treatment, and those with mental illnesses, who make up the majority of suicide victims, may not be aware of their condition or risk. Detecting those at risk of suicide is a challenge for healthcare providers. However, advances in artificial intelligence (AI) may lead to the development of new suicide prediction technologies. This study used machine learning to predict suicide rates across different continents using six common classification algorithms: Stochastic Gradient Descent Classifier (SGDC), Random Forest Classifier (RFC), Gaussian Naive Bayes Classifier (GNBC), K-Neighbors Classifier (KNNC), Logistic Regression Classifier (LRC), and Linear Support Vector Classifier (LSVC). The KNNC algorithm had the highest training accuracy at 100%, and a 97% test accuracy. The RFC algorithm achieved the highest test accuracy at 99%, with a corresponding training accuracy of 99%.