Today, diabetes is one of the most common, chronic, and, due to some complications, deadliest diseases in the world. The early detection of diabetes is very important for its timely treatment since it can stop the progression of the disease. The proposed method can help not only to predict the occurrence of diabetes in the future but also to determine the type of the disease that a person experiences. Considering that type 1 diabetes and type 2 diabetes have many differences in their treatment methods, this method will help to provide the right treatment for the patient. By transforming the task into a classification problem, our model is mainly built using the hidden layers of a deep neural network and uses dropout regularization to prevent overfitting. We tuned a number of parameters and used the binary cross-entropy loss function, which obtained a deep neural network prediction model with high accuracy. The experimental results show the effectiveness and adequacy of the proposed DLPD (Deep Learning for Predicting Diabetes) model. The best training accuracy of the diabetes type data set is 94.02174%, and the training accuracy of the Pima Indians diabetes data set is 99.4112%. Extensive experiments have been conducted on the Pima Indians diabetes and diabetic type datasets. The experimental results show the improvements of our proposed model over the state-of-the-art methods.

/pdf/diabetes-prediction-model-based-on-an-enhanced-deep-neural-2rufzl4zv3.pdf

Diabetes prediction model based on an enhanced deep neural network

A steady rise has been observed in the percentage of elderly people who want and are still able to contribute to society. Therefore, early retirement or exit from the labour market, due to health-related issues, poses a significant problem. Nowadays, thanks to technological advances and various data from different populations, the risk factors investigation and health issues screening are moving towards automation. In the context of this work, a worker-centric, IoT enabled unobtrusive users health, well-being and functional ability monitoring framework, empowered with AI tools, is proposed. Diabetes is a high-prevalence chronic condition with harmful consequences for the quality of life and high mortality rate for people worldwide, in both developed and developing countries. Hence, its severe impact on humans’ life, e.g., personal, social, working, can be considerably reduced if early detection is possible, but most research works in this field fail to provide a more personalized approach both in the modeling and prediction process. In this direction, our designed system concerns diabetes risk prediction in which specific components of the Knowledge Discovery in Database (KDD) process are applied, evaluated and incorporated. Specifically, dataset creation, features selection and classification, using different Supervised Machine Learning (ML) models are considered. The ensemble WeightedVotingLRRFs ML model is proposed to improve the prediction of diabetes, scoring an Area Under the ROC Curve (AUC) of 0.884. Concerning the weighted voting, the optimal weights are estimated by their corresponding Sensitivity and AUC of the ML model based on a bi-objective genetic algorithm. Also, a comparative study is presented among the Finnish Diabetes Risk Score (FINDRISC) and Leicester risk score systems and several ML models, using inductive and transductive learning. The experiments were conducted using data extracted from the English Longitudinal Study of Ageing (ELSA) database.

/pdf/machine-learning-tools-for-long-term-type-2-diabetes-risk-2rusnxo810.pdf

Machine Learning Tools for Long-Term Type 2 Diabetes Risk Prediction

The phishing attack is one of the main cybersecurity threats in web phishing and spear phishing. Phishing websites continue to be a problem. One of the main contributions to our study was working and extracting the URL & Domain Identity feature, Abnormal Features, HTML and JavaScript Features, and Domain Features as semantic features to detect phishing websites, which makes the process of classification using those semantic features, more controllable and more effective. The current study used machine learning model algorithms to detect phishing websites, and comparisons were made. We have used 16 machine learning models adopted with 10 semantic features that represent the most effective features for the detection of phishing webpages extracted from two datasets. The GradientBoostingClassifier and RandomForestClassifier had the best accuracy based on the comparison results (i.e., about 97%). In contrast, GaussianNB and the stochastic gradient descent (SGD) classifier represent the lowest accuracy results; 84% and 81% respectively, in comparison with other classifiers.

/pdf/phishing-website-detection-with-semantic-features-based-on-28vjdy9n.pdf

Phishing Website Detection With Semantic Features Based on Machine Learning Classifiers: A Comparative Study

The growth of the Internet has expanded the amount of data expressed by users across multiple platforms. The availability of these different worldviews and individuals’ emotions empowers sentiment analysis. However, sentiment analysis becomes even more challenging due to a scarcity of standardized labeled data in the Bangla NLP domain. The majority of the existing Bangla research has relied on models of deep learning that significantly focus on context-independent word embeddings, such as Word2Vec, GloVe, and fastText, in which each word has a fixed representation irrespective of its context. Meanwhile, context-based pre-trained language models such as BERT have recently revolutionized the state of natural language processing. In this work, we utilized BERT’s transfer learning ability to a deep integrated model CNN-BiLSTM for enhanced performance of decision-making in sentiment analysis. In addition, we also introduced the ability of transfer learning to classical machine learning algorithms for the performance comparison of CNN-BiLSTM. Additionally, we explore various word embedding techniques, such as Word2Vec, GloVe, and fastText, and compare their performance to the BERT transfer learning strategy. As a result, we have shown a state-of-the-art binary classification performance for Bangla sentiment analysis that significantly outperforms all embedding and algorithms.

/pdf/transfer-learning-for-sentiment-analysis-using-bert-based-2kkw6etv.pdf

Transfer Learning for Sentiment Analysis Using BERT Based Supervised Fine-Tuning

Maternal mortality and childbirth complications are major delivery issues in most developing countries, especially in rural areas. The proper identification of risks associated with the delivery of an expecting woman at an earlier stage can substantially reduce the mortality rate. A few studies have been conducted on using Machine Learning (ML) techniques for predicting birth mode i.e. caesarean section or normal delivery. The most commonly used methods are Decision Tree (DT), K-Nearest Neighbour (KNN), Naive Bayes (NB) and Support Vector Machine (SVM). In this study we have implemented Bagging Ensemble Classifiers based on these traditional machine learning algorithms, which is a novel approach in the area of birth mode prediction. This paper examines the performance of four ML algorithms, with bagging ensemble classifiers (DT-Bagging, KNN-Bagging, NB-Bagging, SVM-Bagging). The result shows that bagging ensemble models outperformed the traditional models in this domain. Besides, we have identified the association between important factors and caesarean section. This study may later be used to create a decision support system by extracting knowledge from the hidden patterns in data to reduce the rate of caesarean delivery in Bangladesh.

Birth Mode Prediction Using Bagging Ensemble Classifier: A Case Study of Bangladesh

This Scientific Research paper is a procedure of an automated system "Doly: Bengali Chatbot" which gives a reply to a user query on behalf of a human for the education system in the Bengali language. This is an AI-based Chatbot, mainly based on machine learning algorithms and Bengali Natural Language Processing (BNLP). The machine gets embedded with this knowledge to identify the desired sentences and making a decision within itself, as a response to answer questions. There are many English Chatbot’s which used in education, web query, banking sector & various sectors. In this research, we have propounded a complete data-driven retrieval based closed domain Chatbot which is easily colloquy in the Bengali language with the users. We’ve created the train function adapter to train the Doly by encoding (encoding="utf8") our corpus from bot data. An input adapter has been created to take input and for output, an output adapter has been created to generate automated responses to a user’s input. We have also used a machine learning algorithm like search algorithm for finding an appropriate list of matching results from the corpus and use Naive Bayesian algorithm to generate the right answer from data. The main aim of this Chatbot based system is to bridge the gap between the knowledge sources by providing instant replies to the questions and queries that have to ask in the Bengali language.

Doly: Bengali Chatbot for Bengali Education

For the time being, Cricket is an indisputably one of the most interesting game in the world, especially in the territory of South Asian. As human beings are prone to error, sometimes errors have happened to an umpire and about a constant time is taken by the third umpire for an exact decision of a review. The two different domains artificial intelligence and computer vision have become pop in cricket analysis and decision making. Using different types of computer vision’s procedures in exploring several Cricket cases and automatically turning into decisions have become exoteric in recent days. In this research paper, we have propounded a classification method by the aids of Convolutional Neural Network (CNN) with Inception V3 in order to automatically unroll the decisions of third umpire and scoring system such as umpire signal detection. We have also proposed the deep CNN technique that aids to increase the performance of CNN.

Detecting Third Umpire Decisions & Automated Scoring System of Cricket

The advent of pre-trained language models has directed a new era of Natural Language Processing (NLP), enabling us to create powerful language models. Among these models, Transformer-based models like BERT have grown in popularity due to their cutting-edge effectiveness. However, these models heavily rely on resource-intensive languages, forcing other languages into multilingual models(mBERT). The two fundamental challenges with mBERT become significantly more challenging in a resource-constrained language like Bangla. It was trained on a limited and organized dataset and contained weights for all other languages. Besides, current research on other languages suggests that a language-specific BERT model will exceed multilingual ones. This paper introduces Bangla-BERT,a a monolingual BERT model for the Bangla language. Despite the limited data available for NLP tasks in Bangla, we perform pre-training on the largest Bangla language model dataset, BanglaLM, which we constructed using 40 GB of text data. Bangla-BERT achieves the highest results in all datasets and vastly improves the state-of-the-art performance in binary linguistic classification, multilabel extraction, and named entity recognition, outperforming multilingual BERT and other previous research. The pre-trained model is assessed against several non-contextual models such as Bangla fasttext and word2vec the downstream tasks. Finally, this model is evaluated by transfer learning based on hybrid deep learning models such as LSTM, CNN, and CRF in NER, and it is observed that Bangla-BERT outperforms state-of-the-art methods. The proposed Bangla-BERT model is assessed by using benchmark datasets, including Banfakenews, Sentiment Analysis on Bengali News Comments, and Cross-lingual Sentiment Analysis in Bengali. Finally, it is concluded that Bangla-BERT surpasses all prior state-of-the-art results by 3.52%, 2.2%, and 5.3%.

https://ieeexplore.ieee.org/ielx7/6287639/6514899/09852438.pdf

Bangla-BERT: Transformer-Based Efficient Model for Transfer Learning and Language Understanding

The ultimate goal of this research paper is to introduce a robust machine learning algorithm called Impact-Learning, which is being used widely to achieve more advanced results on many machine-learning related challenges. Impact learning is a supervised machine learning algorithm for resolving classification and linear or polynomial regression knowledge from examples. It also contributes to analyzing systems for competitive data. This algorithm is unique for being capable of learning from a competition, which is the impact of independent features. In other words, it is trained by the impacts of the features from the intrinsic rate of natural increase (RNI). The input to the Impact Learning is a training set of numerical data. In this work, we used six datasets related to regressions and classifications as the experiment of the Impact Learning, and the comparison indicates that at outperforms other standard machine learning regressions and classifications algorithms such as Random forest tree, SVM, Naive Bayes, Logistic regression and so forth.

Md. Kowsher

Papers

Transfer Learning for Sentiment Analysis Using BERT Based Supervised Fine-Tuning

Doly: Bengali Chatbot for Bengali Education

Detecting Third Umpire Decisions & Automated Scoring System of Cricket

Bangla-BERT: Transformer-Based Efficient Model for Transfer Learning and Language Understanding

Impact-Learning: A Robust Machine Learning Algorithm