Pub Date : 2023-05-11DOI: 10.1109/ICDT57929.2023.10151334
Ashlesha Kolarkar, Surinder Kumar
A natural language interface to query databases (NLIDB) permits users to access data stored in databases by typing queries stated in some natural language. It is typical for the elements in a SQL statement to refer to specific information in the database table rather than being expressly mentioned in the natural language query. As a result, it is vital to gauge the similarity of two texts from the standpoint of semantics rather than strings because it can be challenging to deduce the proper element values purely based on the original query in this circumstance. It can be challenging for a text-to-SQL model to decide which column to utilize if the query includes numerous columns with identical semantics. First, the column representations are improved using table contents, and then, the query representation is improved using the revised column representations. As a result of the efficiency of the column contents being less than ideal, a small-scale introduction is planned. A generic module for use in nlidb systems that enables such systems to conduct queries using aggregations is presented in great detail in this survey. These methods include query-based, pattern-based, general, keyword-based, and grammar-based systems.
{"title":"A Peer Review on Natural Language Interface: Various Challenges and Scope","authors":"Ashlesha Kolarkar, Surinder Kumar","doi":"10.1109/ICDT57929.2023.10151334","DOIUrl":"https://doi.org/10.1109/ICDT57929.2023.10151334","url":null,"abstract":"A natural language interface to query databases (NLIDB) permits users to access data stored in databases by typing queries stated in some natural language. It is typical for the elements in a SQL statement to refer to specific information in the database table rather than being expressly mentioned in the natural language query. As a result, it is vital to gauge the similarity of two texts from the standpoint of semantics rather than strings because it can be challenging to deduce the proper element values purely based on the original query in this circumstance. It can be challenging for a text-to-SQL model to decide which column to utilize if the query includes numerous columns with identical semantics. First, the column representations are improved using table contents, and then, the query representation is improved using the revised column representations. As a result of the efficiency of the column contents being less than ideal, a small-scale introduction is planned. A generic module for use in nlidb systems that enables such systems to conduct queries using aggregations is presented in great detail in this survey. These methods include query-based, pattern-based, general, keyword-based, and grammar-based systems.","PeriodicalId":266681,"journal":{"name":"2023 International Conference on Disruptive Technologies (ICDT)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-05-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131331160","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-11DOI: 10.1109/ICDT57929.2023.10150921
N. Bindu Madhavi, G. Kannan, K. Vinayagam, M. Jayaprakash
The traditional supply chain management and logistics aims at delivering the shipments to user end without delay. The research provides directions for the supply chain management to improve the chain of logistics for the ecommerce sites using nature inspired bee swarm optimization. The utilization of the nature inspired model improves the decision-making ability of the supply chain logistics using various input parameters. The simulation is tested with the development of supply and track model in python tool that uses bee swarm intelligence to track the logistics chain and thereby mitigating the delay in delivering the shipments to the user end. The results show that the proposed intelligence model achieves a higher tracking efficiency than the existing SOTA.
{"title":"Supply Chain Management Using Bee Swarm Optimisation to Improve the Logistics in E- Commerce Era","authors":"N. Bindu Madhavi, G. Kannan, K. Vinayagam, M. Jayaprakash","doi":"10.1109/ICDT57929.2023.10150921","DOIUrl":"https://doi.org/10.1109/ICDT57929.2023.10150921","url":null,"abstract":"The traditional supply chain management and logistics aims at delivering the shipments to user end without delay. The research provides directions for the supply chain management to improve the chain of logistics for the ecommerce sites using nature inspired bee swarm optimization. The utilization of the nature inspired model improves the decision-making ability of the supply chain logistics using various input parameters. The simulation is tested with the development of supply and track model in python tool that uses bee swarm intelligence to track the logistics chain and thereby mitigating the delay in delivering the shipments to the user end. The results show that the proposed intelligence model achieves a higher tracking efficiency than the existing SOTA.","PeriodicalId":266681,"journal":{"name":"2023 International Conference on Disruptive Technologies (ICDT)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-05-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134516788","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-11DOI: 10.1109/ICDT57929.2023.10151341
Shivam Pandey, Aditya, Seema Jain, Usha Dhankar
Images shared online have a high likelihood of being altered, and further global alterations like compression, resizing, or filtering mask the potential change. Many restrictions are placed on forgery detection systems by such manipulations. Image forgery detection is the fundamental solution to many issues, particularly social issues like those on Facebook and legal issues. The most frequent form of image fraud is called a copy-move forgery, where a portion of the original image is copied and pasted in a different spot within the same image. Because the duplicated portions' attributes are similar to those of the original image's components, this type of picture counterfeiting is simpler to carry out but more challenging to detect. The method for spotting copy-move forgeries described in this study is based on processing blocks into features and then extracting those features from the blocks' transforms. A Convolutional Neural Network (CNN) is another tool for detecting forgeries Serial pairings of convolution and pooling layers are employed to conduct feature extraction. Original and changed images are then categorised using transforms and without transformations. We use the CASIA2 dataset, which has 4795 images, of which 1701 are authentic and 3274 are forged. The accuracy of our proposed model is 97.7%. This improved the detection process's overall processing effectiveness and allowed it to fulfill real-time processing demands..
{"title":"Image Forgery Detection","authors":"Shivam Pandey, Aditya, Seema Jain, Usha Dhankar","doi":"10.1109/ICDT57929.2023.10151341","DOIUrl":"https://doi.org/10.1109/ICDT57929.2023.10151341","url":null,"abstract":"Images shared online have a high likelihood of being altered, and further global alterations like compression, resizing, or filtering mask the potential change. Many restrictions are placed on forgery detection systems by such manipulations. Image forgery detection is the fundamental solution to many issues, particularly social issues like those on Facebook and legal issues. The most frequent form of image fraud is called a copy-move forgery, where a portion of the original image is copied and pasted in a different spot within the same image. Because the duplicated portions' attributes are similar to those of the original image's components, this type of picture counterfeiting is simpler to carry out but more challenging to detect. The method for spotting copy-move forgeries described in this study is based on processing blocks into features and then extracting those features from the blocks' transforms. A Convolutional Neural Network (CNN) is another tool for detecting forgeries Serial pairings of convolution and pooling layers are employed to conduct feature extraction. Original and changed images are then categorised using transforms and without transformations. We use the CASIA2 dataset, which has 4795 images, of which 1701 are authentic and 3274 are forged. The accuracy of our proposed model is 97.7%. This improved the detection process's overall processing effectiveness and allowed it to fulfill real-time processing demands..","PeriodicalId":266681,"journal":{"name":"2023 International Conference on Disruptive Technologies (ICDT)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-05-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132041328","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-11DOI: 10.1109/ICDT57929.2023.10151211
D. Praveenadevi, S. Kowsalyadevi, B. Girimurugan, Penugonda. Sreemai, Kolli. Nandini, Sumit Pareek
It is not an easy decision to make when deciding whether or not to let a student continue their studies in a graduate program. There are several factors to take into consideration. An application is analyzed based on a variety of different criteria, and the results of this examination are utilized to provide a prediction of the applicant's likelihood of being successful. Through the course of human history, regression analysis has been used as a methodology for the development of many kinds of prediction systems. On the other hand, it has been demonstrated that the models that were presented in this research had a very limited capacity for predictive ability. An empirical examination of these relationships was carried out by these authors using survey data acquired from MBA students attending a private university. The structural equation models that were generated using this information were used in the investigation. It was found that the content of the courses themselves was the single most critical factor in correctly predicting all learning, satisfaction, and quality.
{"title":"Best Ways Using AI in Impacting Success on MBA Graduates","authors":"D. Praveenadevi, S. Kowsalyadevi, B. Girimurugan, Penugonda. Sreemai, Kolli. Nandini, Sumit Pareek","doi":"10.1109/ICDT57929.2023.10151211","DOIUrl":"https://doi.org/10.1109/ICDT57929.2023.10151211","url":null,"abstract":"It is not an easy decision to make when deciding whether or not to let a student continue their studies in a graduate program. There are several factors to take into consideration. An application is analyzed based on a variety of different criteria, and the results of this examination are utilized to provide a prediction of the applicant's likelihood of being successful. Through the course of human history, regression analysis has been used as a methodology for the development of many kinds of prediction systems. On the other hand, it has been demonstrated that the models that were presented in this research had a very limited capacity for predictive ability. An empirical examination of these relationships was carried out by these authors using survey data acquired from MBA students attending a private university. The structural equation models that were generated using this information were used in the investigation. It was found that the content of the courses themselves was the single most critical factor in correctly predicting all learning, satisfaction, and quality.","PeriodicalId":266681,"journal":{"name":"2023 International Conference on Disruptive Technologies (ICDT)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-05-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"113951783","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-11DOI: 10.1109/ICDT57929.2023.10150611
Naman Goel, Nikhil Prabhat Yadav, Prakarti Prakarti, Anukul Pandey
The current focus of research is on using machine learning (ML) algorithms to predict heart disease. Using the UC Irvine (UCI) Cleveland Heart Disease dataset, this study investigates the effectiveness of various types of classifiers, including K-Nearest Neighbours (KNN), AdaBoost, Gaussian Naïve Bayes (GNB), support vector machines (SVM), multilayer perceptron (MLP) and random forests. The objective of this study is to assess the precision and speed of each classifier and gauge their effectiveness by utilizing measures like accuracy and F1 score for comparison. The study also looks into the potential benefits of fusion methods for improving the accuracy of heart disease prediction. The study concludes that combining various models could lead to improving the metrics. Our study contributes to the ongoing research on heart disease prediction using ML algorithms. The findings of our study can be used to develop more precise models for predicting heart disease, which can aid in improving clinical decision-making for heart disease prevention and treatment.
{"title":"Comparative Analysis of Single Classifier Models against Aggregated Fusion Models for Heart Disease Prediction","authors":"Naman Goel, Nikhil Prabhat Yadav, Prakarti Prakarti, Anukul Pandey","doi":"10.1109/ICDT57929.2023.10150611","DOIUrl":"https://doi.org/10.1109/ICDT57929.2023.10150611","url":null,"abstract":"The current focus of research is on using machine learning (ML) algorithms to predict heart disease. Using the UC Irvine (UCI) Cleveland Heart Disease dataset, this study investigates the effectiveness of various types of classifiers, including K-Nearest Neighbours (KNN), AdaBoost, Gaussian Naïve Bayes (GNB), support vector machines (SVM), multilayer perceptron (MLP) and random forests. The objective of this study is to assess the precision and speed of each classifier and gauge their effectiveness by utilizing measures like accuracy and F1 score for comparison. The study also looks into the potential benefits of fusion methods for improving the accuracy of heart disease prediction. The study concludes that combining various models could lead to improving the metrics. Our study contributes to the ongoing research on heart disease prediction using ML algorithms. The findings of our study can be used to develop more precise models for predicting heart disease, which can aid in improving clinical decision-making for heart disease prevention and treatment.","PeriodicalId":266681,"journal":{"name":"2023 International Conference on Disruptive Technologies (ICDT)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-05-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115502973","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-11DOI: 10.1109/ICDT57929.2023.10150836
Sahote Tejinder Singh, Madhuri Dinesh Gabhane, C. Mahamuni
Spam is the act of sending unsolicited emails to a large number of users for phishing, spreading malware, etc. Internet Service Providers (ISPs) and email inbox providers (like Gmail, Yahoo Mail, AOL, etc.) rely on SPAM filters, firewalls, and blacklist directories to prevent "unsolicited" SPAM emails from entering your inbox. Spam mails are overrunning email inboxes, which significantly slows down internet performance. It is crucial to properly analyze the connections between these spammers and spam because the majority of us tend to provide them with crucial information, such as our contact information. Since the benefactor covers a large percentage of the costs related to spamming, it effectively serves as advertising for the cost of mailing. The study of existing work shows that machine learning and deep learning are frequently employed to effectively identify email spam. This research paper is secondary work in which we have studied, and implemented the various machine learning and deep learning approaches to identify email spam in Python. The four machine learning algorithms—KNN, Navies Bayes, BiLSTM, and Deep CNN—show that they can be utilized effectively to detect spam. Yet the Deep CNN outperforms the other three based on accuracy and the F1 score.
{"title":"Study of Machine Learning and Deep Learning Algorithms for the Detection of Email Spam based on Python Implementation","authors":"Sahote Tejinder Singh, Madhuri Dinesh Gabhane, C. Mahamuni","doi":"10.1109/ICDT57929.2023.10150836","DOIUrl":"https://doi.org/10.1109/ICDT57929.2023.10150836","url":null,"abstract":"Spam is the act of sending unsolicited emails to a large number of users for phishing, spreading malware, etc. Internet Service Providers (ISPs) and email inbox providers (like Gmail, Yahoo Mail, AOL, etc.) rely on SPAM filters, firewalls, and blacklist directories to prevent \"unsolicited\" SPAM emails from entering your inbox. Spam mails are overrunning email inboxes, which significantly slows down internet performance. It is crucial to properly analyze the connections between these spammers and spam because the majority of us tend to provide them with crucial information, such as our contact information. Since the benefactor covers a large percentage of the costs related to spamming, it effectively serves as advertising for the cost of mailing. The study of existing work shows that machine learning and deep learning are frequently employed to effectively identify email spam. This research paper is secondary work in which we have studied, and implemented the various machine learning and deep learning approaches to identify email spam in Python. The four machine learning algorithms—KNN, Navies Bayes, BiLSTM, and Deep CNN—show that they can be utilized effectively to detect spam. Yet the Deep CNN outperforms the other three based on accuracy and the F1 score.","PeriodicalId":266681,"journal":{"name":"2023 International Conference on Disruptive Technologies (ICDT)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-05-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124436118","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-11DOI: 10.1109/ICDT57929.2023.10150712
I. Jeya Daisy, G. Diyaneshwaran, K. Ravivarmaa, S. Shobana, M. Sneha, N. S. Monessha
Modern obstetrics places a high priority on foetal health monitoring. Although foetal movement is frequently used as a proxy for foetal health, it is difficult to accurately monitor foetal movement over an extended period of time without causing any harm. In high-risk pregnancies and in high-risk moms who have previously experienced miscarriages, it is highly helpful to determine the foetus position because, in the majority of cases, an incorrect foetal position results in both foetal and maternal mortality. Pregnant women may benefit from the design and construction of a device that can accurately identify the location of the foetus. Recent years have seen the development of a few accelerometer-based systems to address frequent problems with ultrasound measurement and allow for remote, self-managed monitoring of foetal movement throughout pregnancy. The optimum design for body-worn accelerometers, data processing, and deep learning methods used to identify foetal movement. This study will explore four alternative techniques for determining the location of the foetus. Ultrasonograms are the most popular methods for foetal position detection. The wearable ambulatory device known as Femom, which has been made available to women on home prescription, can also be used to determine the location of the foetus. Deep learning techniques and thermal imaging cameras are also utilised to determine the position of the foetus.
{"title":"Review On Foetal Position Detection Using Different Techniques","authors":"I. Jeya Daisy, G. Diyaneshwaran, K. Ravivarmaa, S. Shobana, M. Sneha, N. S. Monessha","doi":"10.1109/ICDT57929.2023.10150712","DOIUrl":"https://doi.org/10.1109/ICDT57929.2023.10150712","url":null,"abstract":"Modern obstetrics places a high priority on foetal health monitoring. Although foetal movement is frequently used as a proxy for foetal health, it is difficult to accurately monitor foetal movement over an extended period of time without causing any harm. In high-risk pregnancies and in high-risk moms who have previously experienced miscarriages, it is highly helpful to determine the foetus position because, in the majority of cases, an incorrect foetal position results in both foetal and maternal mortality. Pregnant women may benefit from the design and construction of a device that can accurately identify the location of the foetus. Recent years have seen the development of a few accelerometer-based systems to address frequent problems with ultrasound measurement and allow for remote, self-managed monitoring of foetal movement throughout pregnancy. The optimum design for body-worn accelerometers, data processing, and deep learning methods used to identify foetal movement. This study will explore four alternative techniques for determining the location of the foetus. Ultrasonograms are the most popular methods for foetal position detection. The wearable ambulatory device known as Femom, which has been made available to women on home prescription, can also be used to determine the location of the foetus. Deep learning techniques and thermal imaging cameras are also utilised to determine the position of the foetus.","PeriodicalId":266681,"journal":{"name":"2023 International Conference on Disruptive Technologies (ICDT)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-05-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115958939","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-11DOI: 10.1109/ICDT57929.2023.10150962
Asif Khan, Dev Paliwal, Ritank Jaikar, S. Attri
An animal's normal state is altered by sickness which can stop or change critical processes. Concerns over animal diseases have existed as animal lovers interacted with animals and this concern is reflected in the first ideas about religion and magic. Animal illnesses still pose a threat, primarily due to the potential financial costs and risk of human transmission. The study, prevention, and treatment of diseases in animals including wild animals and those utilized in scientific research are the focus of the medical specialty known as veterinary medicine. This research examines recent developments in image-based animal illness detection and predicting the best deep learning model to detect the animal diseases. People now have a better grasp of machine learning and its potential uses in treating animal diseases as a result of the discussion of this paper. Regarding accuracy, DenseNet169 has performed remarkably better than other models whereas ResNet50V2 has least accuracy. These models are trained on the dataset which is built using images collected by the Authors.
{"title":"Demystifying the Transfer Learning based Detection of Animal Diseases from Images","authors":"Asif Khan, Dev Paliwal, Ritank Jaikar, S. Attri","doi":"10.1109/ICDT57929.2023.10150962","DOIUrl":"https://doi.org/10.1109/ICDT57929.2023.10150962","url":null,"abstract":"An animal's normal state is altered by sickness which can stop or change critical processes. Concerns over animal diseases have existed as animal lovers interacted with animals and this concern is reflected in the first ideas about religion and magic. Animal illnesses still pose a threat, primarily due to the potential financial costs and risk of human transmission. The study, prevention, and treatment of diseases in animals including wild animals and those utilized in scientific research are the focus of the medical specialty known as veterinary medicine. This research examines recent developments in image-based animal illness detection and predicting the best deep learning model to detect the animal diseases. People now have a better grasp of machine learning and its potential uses in treating animal diseases as a result of the discussion of this paper. Regarding accuracy, DenseNet169 has performed remarkably better than other models whereas ResNet50V2 has least accuracy. These models are trained on the dataset which is built using images collected by the Authors.","PeriodicalId":266681,"journal":{"name":"2023 International Conference on Disruptive Technologies (ICDT)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-05-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116191703","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-11DOI: 10.1109/ICDT57929.2023.10151148
S. Mushtaq, K. S. Saini, Saimul Bashir
Brain stroke is a serious medical condition that needs timely diagnosis and action to avoid irretrievable harm to the brain. Machine learmusht (ML) techniques have been extensively used in the healthcare industry to build predictive models for various medical conditions, including brain stroke, heart stroke and diabetes disease. In this paper, we present an advanced stroke detection algorithm for predicting the occurrence of stroke. We used a dataset contaimusht detail of important parameters which are responsible for the brain stroke like: Age: Body Mass Index (BMI): Gender: Heart Disease: Smoking Status etc, to develop a predictive model. The dataset was preprocessed to handle missing values, handle categorical features and to balance the dataset. We used different classification algorithms such as Naïve Bayes, logistic regression, XgBoost, decision trees, AdaBoost, K-Nearest Neighbor, random forests, Voting classifier and support vector machines to develop our predictive model. The evaluation of the models was conducted using several metrics such as accuracy, F1-score, recall, precision. Moreover an additional metrics parameter is calculated in this paper known as Specificity which was not calculated in earlier studies. Our results showed that the Support Vector Machine algorithm outperformed other models, achieving an accuracy of 99.5%, precision of 99.9% , recall of 99.1%, F1-score of 99.5% and specificity of 99%.
脑中风是一种严重的疾病，需要及时诊断和采取行动，以避免对大脑造成不可挽回的伤害。机器学习(ML)技术已广泛应用于医疗保健行业，用于建立各种医疗状况的预测模型，包括脑中风、心脏病和糖尿病疾病。本文提出了一种先进的脑卒中检测算法，用于预测脑卒中的发生。我们使用了一个数据集，其中包含了与脑中风有关的重要参数的详细信息，如:年龄、体重指数(BMI)、性别、心脏病、吸烟状况等，以建立一个预测模型。对数据集进行预处理，处理缺失值，处理分类特征，平衡数据集。我们使用了不同的分类算法，如Naïve贝叶斯，逻辑回归，XgBoost，决策树，AdaBoost, k -近邻，随机森林，投票分类器和支持向量机来开发我们的预测模型。采用准确性、f1评分、召回率、精度等指标对模型进行评价。此外，本文还计算了一个额外的度量参数，称为特异性，这在早期的研究中没有计算。结果表明，支持向量机算法的准确率为99.5%，精密度为99.9%，召回率为99.1%，f1评分为99.5%，特异性为99%，优于其他模型。
{"title":"Machine Learmusht for Brain Stroke Prediction","authors":"S. Mushtaq, K. S. Saini, Saimul Bashir","doi":"10.1109/ICDT57929.2023.10151148","DOIUrl":"https://doi.org/10.1109/ICDT57929.2023.10151148","url":null,"abstract":"Brain stroke is a serious medical condition that needs timely diagnosis and action to avoid irretrievable harm to the brain. Machine learmusht (ML) techniques have been extensively used in the healthcare industry to build predictive models for various medical conditions, including brain stroke, heart stroke and diabetes disease. In this paper, we present an advanced stroke detection algorithm for predicting the occurrence of stroke. We used a dataset contaimusht detail of important parameters which are responsible for the brain stroke like: Age: Body Mass Index (BMI): Gender: Heart Disease: Smoking Status etc, to develop a predictive model. The dataset was preprocessed to handle missing values, handle categorical features and to balance the dataset. We used different classification algorithms such as Naïve Bayes, logistic regression, XgBoost, decision trees, AdaBoost, K-Nearest Neighbor, random forests, Voting classifier and support vector machines to develop our predictive model. The evaluation of the models was conducted using several metrics such as accuracy, F1-score, recall, precision. Moreover an additional metrics parameter is calculated in this paper known as Specificity which was not calculated in earlier studies. Our results showed that the Support Vector Machine algorithm outperformed other models, achieving an accuracy of 99.5%, precision of 99.9% , recall of 99.1%, F1-score of 99.5% and specificity of 99%.","PeriodicalId":266681,"journal":{"name":"2023 International Conference on Disruptive Technologies (ICDT)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-05-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128757677","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2023-05-11DOI: 10.1109/icdt57929.2023.10150838
Ajay Kumar, J. Jaiswal, Naman Tiwari
Blind spot is the region that is not visible to the driver while driving car via side or rear mirrors. The blind spot is usually located at the rear of the vehicle, but may also be found on both sides. It is caused due to obstruction from other vehicles, objects or pedestrians. Other names for blind areas include "blind zones," "fatal zones," and "dead spots.". This blind spot can be dangerous for drivers, especially when they are driving at night or in bad weather conditions. When drivers neglect to examine their blind areas before changing lanes or making a right turn, this can result in accidents and injuries. Our proposed model will be able to identify the objects that lies in the vehicle's blind spot area using an Arduino and an ultrasonic sensor. The use of a BSMS while driving can help you stay safe. It can be installed on the car’s rear fender and if there are any objects in the vicinity of the model then an alarm will be generated and the driver will have enough time to react before he gets into an accident We have suggested the idea of implementing machine learning algorithms for better accuracy and reliability.
{"title":"Blind Spot Monitoring System Using Ultrasonic Sensor","authors":"Ajay Kumar, J. Jaiswal, Naman Tiwari","doi":"10.1109/icdt57929.2023.10150838","DOIUrl":"https://doi.org/10.1109/icdt57929.2023.10150838","url":null,"abstract":"Blind spot is the region that is not visible to the driver while driving car via side or rear mirrors. The blind spot is usually located at the rear of the vehicle, but may also be found on both sides. It is caused due to obstruction from other vehicles, objects or pedestrians. Other names for blind areas include \"blind zones,\" \"fatal zones,\" and \"dead spots.\". This blind spot can be dangerous for drivers, especially when they are driving at night or in bad weather conditions. When drivers neglect to examine their blind areas before changing lanes or making a right turn, this can result in accidents and injuries. Our proposed model will be able to identify the objects that lies in the vehicle's blind spot area using an Arduino and an ultrasonic sensor. The use of a BSMS while driving can help you stay safe. It can be installed on the car’s rear fender and if there are any objects in the vicinity of the model then an alarm will be generated and the driver will have enough time to react before he gets into an accident We have suggested the idea of implementing machine learning algorithms for better accuracy and reliability.","PeriodicalId":266681,"journal":{"name":"2023 International Conference on Disruptive Technologies (ICDT)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-05-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121770701","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：