Pub Date : 2022-06-06DOI: 10.1109/aiiot54504.2022.9817287
Hebah Alquran, Shadi Banitaan
Fake news is propagated by intentionally spreading false information on social media platforms. Fake news intends to mislead the public and damage the reputation of a person or entity. Detecting misinformation over digital platforms is essential to minimizing its adverse effects. While false comments and news can be easily posted on social media without any oversight, identifying real information from false information is often the most challenging part. This work examined the most relevant features that can be used for fake news detection. After selecting the significant features, prediction models are built and compared in terms of precision, recall, and F-score evaluation metrics using Naive Bayes, Bayesian Network, and J48 classification methods. Based on our experiments on a benchmark dataset, we obtained an overall F-score of 69.7% by employing the J48 classifier on the politician's brief statement, and the counts of the speaker's statement history feature set.
{"title":"Fake News Detection in Social Networks Using Data Mining Techniques","authors":"Hebah Alquran, Shadi Banitaan","doi":"10.1109/aiiot54504.2022.9817287","DOIUrl":"https://doi.org/10.1109/aiiot54504.2022.9817287","url":null,"abstract":"Fake news is propagated by intentionally spreading false information on social media platforms. Fake news intends to mislead the public and damage the reputation of a person or entity. Detecting misinformation over digital platforms is essential to minimizing its adverse effects. While false comments and news can be easily posted on social media without any oversight, identifying real information from false information is often the most challenging part. This work examined the most relevant features that can be used for fake news detection. After selecting the significant features, prediction models are built and compared in terms of precision, recall, and F-score evaluation metrics using Naive Bayes, Bayesian Network, and J48 classification methods. Based on our experiments on a benchmark dataset, we obtained an overall F-score of 69.7% by employing the J48 classifier on the politician's brief statement, and the counts of the speaker's statement history feature set.","PeriodicalId":409264,"journal":{"name":"2022 IEEE World AI IoT Congress (AIIoT)","volume":"88 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124558846","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 : 2022-06-06DOI: 10.1109/aiiot54504.2022.9817168
A. Ornatelli, A. Giuseppi, A. Tortorelli
This paper deals with the transmission power control problem in wireless networks. Such a problem represents a well known and relevant issue as it allows to efficiently manage the network's required energy and the interference experienced by end-users. With the widespread diffusion of smart devices, the relevance of this aspect further increased and has been identified as such also in 5G standards. The problem has been formalized as a Multi-Agent Reinforcement Learning approach (MARL) to guarantee scalability and robustness. These two aspects also drove the development of an original Distributed Average-Cost Temporal-Difference (TD) Learning algorithm. To adopt such an algorithm, a Markov Game formulation of the power control problem has also been derived. The effectiveness of the proposed distributed framework in reducing the total network's transmission power has been proved by means of simulations in a specific case study.
{"title":"A Distributed Average Cost Reinforcement Learning approach for Power Control in Wireless 5G Networks","authors":"A. Ornatelli, A. Giuseppi, A. Tortorelli","doi":"10.1109/aiiot54504.2022.9817168","DOIUrl":"https://doi.org/10.1109/aiiot54504.2022.9817168","url":null,"abstract":"This paper deals with the transmission power control problem in wireless networks. Such a problem represents a well known and relevant issue as it allows to efficiently manage the network's required energy and the interference experienced by end-users. With the widespread diffusion of smart devices, the relevance of this aspect further increased and has been identified as such also in 5G standards. The problem has been formalized as a Multi-Agent Reinforcement Learning approach (MARL) to guarantee scalability and robustness. These two aspects also drove the development of an original Distributed Average-Cost Temporal-Difference (TD) Learning algorithm. To adopt such an algorithm, a Markov Game formulation of the power control problem has also been derived. The effectiveness of the proposed distributed framework in reducing the total network's transmission power has been proved by means of simulations in a specific case study.","PeriodicalId":409264,"journal":{"name":"2022 IEEE World AI IoT Congress (AIIoT)","volume":"60 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121172374","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 : 2022-06-06DOI: 10.1109/aiiot54504.2022.9817160
Khan Md Hasib, S. Sakib, J. Mahmud, Kamruzzaman Mithu, Md. Saifur Rahman, Mohammad Shafiul Alam
The severely infectious virus known as “COVID- 19” has wreaked havoc on the planet, trapping to keep the disease from spreading, while billions of people are staying inside. Every experts and professionals in many disciplines are working tirelessly to create a vaccine and preventative techniques to help the globe overcome this difficult crisis. In Bangladesh, the number of persons infected with Coronavirus is particularly alarming. A accurate prognosis of the epidemic, on the other hand, may aid in the management of this contagious illness until a remedy is discovered. This study aims to forecast impending COVID-19 exposed instances and fatalities using a time series dataset utilizing proposed deep transfer learning model where encoder-decoder CNN-LSTM along with deep CNN pretrained models such as: ResNet-50, DenseNet-201, MobileNet-V2, and Inception-ResNet-V2 performed. We also predict the regular exposed instances and fatalities throughout the following 180 days in data visualization segment using AIC and BIC selection criteria. The suggested paradigms are also used to anticipate Bangladesh's daily confirmed cases and daily which is evaluated by error based on three performance criteria. We discovered that ResNet-50 performs better among others for predicting infected case and deaths owing to COVID-19 in Bangladesh in terms of MAPE, MAE and RMSE evaluations.
{"title":"COVID-19 Prediction based on Infected Cases and Deaths of Bangladesh using Deep Transfer Learning","authors":"Khan Md Hasib, S. Sakib, J. Mahmud, Kamruzzaman Mithu, Md. Saifur Rahman, Mohammad Shafiul Alam","doi":"10.1109/aiiot54504.2022.9817160","DOIUrl":"https://doi.org/10.1109/aiiot54504.2022.9817160","url":null,"abstract":"The severely infectious virus known as “COVID- 19” has wreaked havoc on the planet, trapping to keep the disease from spreading, while billions of people are staying inside. Every experts and professionals in many disciplines are working tirelessly to create a vaccine and preventative techniques to help the globe overcome this difficult crisis. In Bangladesh, the number of persons infected with Coronavirus is particularly alarming. A accurate prognosis of the epidemic, on the other hand, may aid in the management of this contagious illness until a remedy is discovered. This study aims to forecast impending COVID-19 exposed instances and fatalities using a time series dataset utilizing proposed deep transfer learning model where encoder-decoder CNN-LSTM along with deep CNN pretrained models such as: ResNet-50, DenseNet-201, MobileNet-V2, and Inception-ResNet-V2 performed. We also predict the regular exposed instances and fatalities throughout the following 180 days in data visualization segment using AIC and BIC selection criteria. The suggested paradigms are also used to anticipate Bangladesh's daily confirmed cases and daily which is evaluated by error based on three performance criteria. We discovered that ResNet-50 performs better among others for predicting infected case and deaths owing to COVID-19 in Bangladesh in terms of MAPE, MAE and RMSE evaluations.","PeriodicalId":409264,"journal":{"name":"2022 IEEE World AI IoT Congress (AIIoT)","volume":"33 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129292040","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 : 2022-06-06DOI: 10.1109/aiiot54504.2022.9817228
Cesar Martinez Melgoza, Kayla Lee, Tyler Groom, Nate Ruppert, K. George, Henry Lin
Over the years, the increase in electronic devices and innovation towards technological capabilities have resulted in an increase in traffic in the electromagnetic spectrum, thus making it harder for radar systems to distinguish multiple emitters with added interference. Traditional methods for classification, such as machine learning, prove to be a suitable solution for this problem, however these models require an enormous amount of data to train and evaluate. This experiment implements a Few-Shot learning framework and evaluates the performance of different Neural Network Architectures such as a standard Convolutional Neural Network, and a Siamese Network from a previous experiment. The experiment will utilize different kinds of hardware equipment. This includes the ZCU104 FPGA board, AD-FMCOMMS2-EBZ RF module, the Jetson TX2, and NVIDIA Titan RTX. The hardware equipment will be evaluated using performance metrics such as hardware acceleration, to find the best medium between computational power, acceleration speed, and evaluation accuracy.
{"title":"Comparing Pretrained Image-Net CNN with a Siamese Architecture for Few-Shot Learning Applications in Radar Systems","authors":"Cesar Martinez Melgoza, Kayla Lee, Tyler Groom, Nate Ruppert, K. George, Henry Lin","doi":"10.1109/aiiot54504.2022.9817228","DOIUrl":"https://doi.org/10.1109/aiiot54504.2022.9817228","url":null,"abstract":"Over the years, the increase in electronic devices and innovation towards technological capabilities have resulted in an increase in traffic in the electromagnetic spectrum, thus making it harder for radar systems to distinguish multiple emitters with added interference. Traditional methods for classification, such as machine learning, prove to be a suitable solution for this problem, however these models require an enormous amount of data to train and evaluate. This experiment implements a Few-Shot learning framework and evaluates the performance of different Neural Network Architectures such as a standard Convolutional Neural Network, and a Siamese Network from a previous experiment. The experiment will utilize different kinds of hardware equipment. This includes the ZCU104 FPGA board, AD-FMCOMMS2-EBZ RF module, the Jetson TX2, and NVIDIA Titan RTX. The hardware equipment will be evaluated using performance metrics such as hardware acceleration, to find the best medium between computational power, acceleration speed, and evaluation accuracy.","PeriodicalId":409264,"journal":{"name":"2022 IEEE World AI IoT Congress (AIIoT)","volume":"103 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116975154","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 : 2022-06-06DOI: 10.1109/aiiot54504.2022.9817368
Jack Li, Yim-Fun Hu
Although the success in application of the Internet pushes development and applications of the Internet-of-Things (IoT) on market quickly, IoT devices are also exposed to attacks from the network, which raises the security issues of IoT devices. Most IoT devices are embedded systems, and there was little work on device security as a part of the device design because most applications force engineers to mainly focus on how to implement the systems' functions with less hardware and software design as well as less power consumption. There are many new technologies, such as AI, machine learning that could provide good solutions to device security. However, all these new technologies rely on complex calculation and large amount of memory etc., which is not part of most IoT devices, such as a smart sensor. Using edge computing to provide some security solutions for IoT devices is one approach to solve the IoT security problems. Detecting some malfunctions in the system at an IoT device by edge computing is proposed in this work to make an IoT device more secure.
{"title":"Employing Edge Computing to Enhance Self-Defense Capabilities of IoT Devices","authors":"Jack Li, Yim-Fun Hu","doi":"10.1109/aiiot54504.2022.9817368","DOIUrl":"https://doi.org/10.1109/aiiot54504.2022.9817368","url":null,"abstract":"Although the success in application of the Internet pushes development and applications of the Internet-of-Things (IoT) on market quickly, IoT devices are also exposed to attacks from the network, which raises the security issues of IoT devices. Most IoT devices are embedded systems, and there was little work on device security as a part of the device design because most applications force engineers to mainly focus on how to implement the systems' functions with less hardware and software design as well as less power consumption. There are many new technologies, such as AI, machine learning that could provide good solutions to device security. However, all these new technologies rely on complex calculation and large amount of memory etc., which is not part of most IoT devices, such as a smart sensor. Using edge computing to provide some security solutions for IoT devices is one approach to solve the IoT security problems. Detecting some malfunctions in the system at an IoT device by edge computing is proposed in this work to make an IoT device more secure.","PeriodicalId":409264,"journal":{"name":"2022 IEEE World AI IoT Congress (AIIoT)","volume":"48 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116750036","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 : 2022-06-06DOI: 10.1109/aiiot54504.2022.9817298
Shereen S. Ismail, H. Reza
Wireless Sensor Network (WSN) is one of the Internet of Things (IoT) operating platforms, which has proliferated into a wide range of applications. These networks comprise many resource-restricted sensors in terms of sensing, communication, storage, and power. Security becomes a critical concern to protect the network of scarce resources from any malicious activities that target the network. Several solutions have been presented in the literature; however, machine learning has proven its appropriateness in designing energy-efficient detection measures for cyber-attacks targeting WSNs. This paper presents a WSN security performance evaluation of three Naïve Bayesian machine learning classification technique variants: Gaussian Naïve Bayes, Multinomial Naïve Bayes, and Bernoulli Naïve Bayes, compared to three well-known base algorithms: K-Nearest Neighbors, Support Vector Machine, and Multilayer Perceptron. We applied Spearman correlation as a univariate feature selection. The specialized dataset, WSN-DS, was used for training and testing purposes. The performance of the six classifiers was evaluated in terms of accuracy, probability of detection, positive prediction value, probability of false alarm, probability of misdetection, memory usage, processing time, prediction time, and complexity.
{"title":"Evaluation of Naïve Bayesian Algorithms for Cyber-Attacks Detection in Wireless Sensor Networks","authors":"Shereen S. Ismail, H. Reza","doi":"10.1109/aiiot54504.2022.9817298","DOIUrl":"https://doi.org/10.1109/aiiot54504.2022.9817298","url":null,"abstract":"Wireless Sensor Network (WSN) is one of the Internet of Things (IoT) operating platforms, which has proliferated into a wide range of applications. These networks comprise many resource-restricted sensors in terms of sensing, communication, storage, and power. Security becomes a critical concern to protect the network of scarce resources from any malicious activities that target the network. Several solutions have been presented in the literature; however, machine learning has proven its appropriateness in designing energy-efficient detection measures for cyber-attacks targeting WSNs. This paper presents a WSN security performance evaluation of three Naïve Bayesian machine learning classification technique variants: Gaussian Naïve Bayes, Multinomial Naïve Bayes, and Bernoulli Naïve Bayes, compared to three well-known base algorithms: K-Nearest Neighbors, Support Vector Machine, and Multilayer Perceptron. We applied Spearman correlation as a univariate feature selection. The specialized dataset, WSN-DS, was used for training and testing purposes. The performance of the six classifiers was evaluated in terms of accuracy, probability of detection, positive prediction value, probability of false alarm, probability of misdetection, memory usage, processing time, prediction time, and complexity.","PeriodicalId":409264,"journal":{"name":"2022 IEEE World AI IoT Congress (AIIoT)","volume":"35 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115463968","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 : 2022-06-06DOI: 10.1109/aiiot54504.2022.9817161
Nicholas Foley, Chen-Hsiang Yu
- Muscles are an essential part of everyday life and any damage or illness that affects them can cause massive problems. Patients who are diagnosed with muscle injuries and illnesses largely remain unmonitored, even though a few appointments they have with doctors annually. Moving from unmonitored to constant monitoring can not only paint a better picture of how a muscle condition is progressing, but it also can inform medical professionals if their treatment regimen is actually working. In this paper, we propose a new system that can monitor muscle health of a patient and predict the muscle conditions. This system mainly focuses on the shoulder but could be expanded to other areas of the body. By utilizing the strength of machine learning and the Android platform, we created a platform that can monitor muscle health quickly and easily. The current prototype system is not only able to display live data gathered from an EMG sensor, but it can also predict whether the muscle is currently flexed or relaxed. Although there is a limitation in current prototype system, a more robust machine learning algorithm could be trained to give a wide array of muscle health predictions.
{"title":"MusCare+: Muscle Monitoring for Anomalies","authors":"Nicholas Foley, Chen-Hsiang Yu","doi":"10.1109/aiiot54504.2022.9817161","DOIUrl":"https://doi.org/10.1109/aiiot54504.2022.9817161","url":null,"abstract":"- Muscles are an essential part of everyday life and any damage or illness that affects them can cause massive problems. Patients who are diagnosed with muscle injuries and illnesses largely remain unmonitored, even though a few appointments they have with doctors annually. Moving from unmonitored to constant monitoring can not only paint a better picture of how a muscle condition is progressing, but it also can inform medical professionals if their treatment regimen is actually working. In this paper, we propose a new system that can monitor muscle health of a patient and predict the muscle conditions. This system mainly focuses on the shoulder but could be expanded to other areas of the body. By utilizing the strength of machine learning and the Android platform, we created a platform that can monitor muscle health quickly and easily. The current prototype system is not only able to display live data gathered from an EMG sensor, but it can also predict whether the muscle is currently flexed or relaxed. Although there is a limitation in current prototype system, a more robust machine learning algorithm could be trained to give a wide array of muscle health predictions.","PeriodicalId":409264,"journal":{"name":"2022 IEEE World AI IoT Congress (AIIoT)","volume":"56 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122002038","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 : 2022-06-06DOI: 10.1109/aiiot54504.2022.9817157
Bhargavi Krishnamurthy, S. Shiva
-Nowadays the data is growing at a faster pace and the big data applications are required to be more agile and flexible. There is a need for a decentralized model to carry out the required substantial amount of computation across edge devices as they has led to the innovation of fog computing. Energy consumption among the edge devices is one of the potential threatening issues in fog computing. Their high energy demand also contributes to higher computation cost. In this paper Double Critic (DC) approach is enforced over the Deep Deterministic Policy Gradient (DDPG) technique to design the DC-DDPG framework which formulates high quality energy efficiency policies for fog computing. The performance of the proposed framework is outstanding compared to existing works based on the metrics like energy consumption, response time, total cost, and throughput. They are measured under two different fog computing scenarios i.e., fog layer with multiple entities in a region and fog layer with multiple entities in multiple regions. Mathematical modeling reveals that the energy efficiency policies formulated are of high quality as they satisfy the quality assurance metrics, such as empirical correctness, robustness, model relevance, and data privacy.
{"title":"Energy Efficient Double Critic Deep Deterministic Policy Gradient Framework for Fog Computing","authors":"Bhargavi Krishnamurthy, S. Shiva","doi":"10.1109/aiiot54504.2022.9817157","DOIUrl":"https://doi.org/10.1109/aiiot54504.2022.9817157","url":null,"abstract":"-Nowadays the data is growing at a faster pace and the big data applications are required to be more agile and flexible. There is a need for a decentralized model to carry out the required substantial amount of computation across edge devices as they has led to the innovation of fog computing. Energy consumption among the edge devices is one of the potential threatening issues in fog computing. Their high energy demand also contributes to higher computation cost. In this paper Double Critic (DC) approach is enforced over the Deep Deterministic Policy Gradient (DDPG) technique to design the DC-DDPG framework which formulates high quality energy efficiency policies for fog computing. The performance of the proposed framework is outstanding compared to existing works based on the metrics like energy consumption, response time, total cost, and throughput. They are measured under two different fog computing scenarios i.e., fog layer with multiple entities in a region and fog layer with multiple entities in multiple regions. Mathematical modeling reveals that the energy efficiency policies formulated are of high quality as they satisfy the quality assurance metrics, such as empirical correctness, robustness, model relevance, and data privacy.","PeriodicalId":409264,"journal":{"name":"2022 IEEE World AI IoT Congress (AIIoT)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126705267","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 : 2022-06-06DOI: 10.1109/aiiot54504.2022.9817303
M. Mamun, Afia Farjana, Miraz Al Mamun, Md Salim Ahammed, Md Minhazur Rahman
Heart Failure (HF) is a prevalent ailment worldwide, and despite significant medical advancements in the past few decades, cardiovascular disease is still the leading cause of death. Although HF itself is a critical risk for patient survival, other co-existing pathophysiological conditions can present a significant threat to patient survival. Because so many elements contribute to a patient's survival in heart failure, predicting the chances of survival without using a computational technique can be difficult for cardiac doctors, eventually preventing the patient from receiving correct care. Fortunately, categorization and prediction models exist, which can assist cardiologists in designing proper treatment schemes using relevant medical data. This study aims to develop prediction models for patient survival in HF conditions. In this paper, we analyzed the UCI heart failure dataset containing relevant medical information of 299 HF patients. We applied several machine learning classifiers to predict the patient survival from HF-related pathophysiological parameters and analyzed the features corresponding to the most crucial risk factors using the correlation matrix. Our prediction models used the following machine learning techniques- Logistic Regression, Decision Tree, Support Vector Machine, XGBoost, LightGBM, Random Forest, KNN, and Bagging and were able to find a better result. Also, this paper presents a comparative study by analyzing the performance of different machine learning algorithms. Our analysis indicates that LightGBM achieved the highest Accuracy of 85% and AUC of 93% in predicting patient survival of HF patients compared to other machine learning algorithms.
{"title":"Heart failure survival prediction using machine learning algorithm: am I safe from heart failure?","authors":"M. Mamun, Afia Farjana, Miraz Al Mamun, Md Salim Ahammed, Md Minhazur Rahman","doi":"10.1109/aiiot54504.2022.9817303","DOIUrl":"https://doi.org/10.1109/aiiot54504.2022.9817303","url":null,"abstract":"Heart Failure (HF) is a prevalent ailment worldwide, and despite significant medical advancements in the past few decades, cardiovascular disease is still the leading cause of death. Although HF itself is a critical risk for patient survival, other co-existing pathophysiological conditions can present a significant threat to patient survival. Because so many elements contribute to a patient's survival in heart failure, predicting the chances of survival without using a computational technique can be difficult for cardiac doctors, eventually preventing the patient from receiving correct care. Fortunately, categorization and prediction models exist, which can assist cardiologists in designing proper treatment schemes using relevant medical data. This study aims to develop prediction models for patient survival in HF conditions. In this paper, we analyzed the UCI heart failure dataset containing relevant medical information of 299 HF patients. We applied several machine learning classifiers to predict the patient survival from HF-related pathophysiological parameters and analyzed the features corresponding to the most crucial risk factors using the correlation matrix. Our prediction models used the following machine learning techniques- Logistic Regression, Decision Tree, Support Vector Machine, XGBoost, LightGBM, Random Forest, KNN, and Bagging and were able to find a better result. Also, this paper presents a comparative study by analyzing the performance of different machine learning algorithms. Our analysis indicates that LightGBM achieved the highest Accuracy of 85% and AUC of 93% in predicting patient survival of HF patients compared to other machine learning algorithms.","PeriodicalId":409264,"journal":{"name":"2022 IEEE World AI IoT Congress (AIIoT)","volume":"29 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125228930","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 : 2022-06-06DOI: 10.1109/aiiot54504.2022.9817331
Bhargavi Krishnamurthy, S. Shiva, Saikat Das, Ph.D.
Fog computing can process big data generated by the IoT (IoT) architectures. The hierarchical, heterogeneous and distributed form of fog computing makes the application placement a challenging task. IoT applications are time-sensitive, and their placement decision is dependent on the user's Quality of Experience (QoE). This paper proposes an explainable Model Value Evaluation based Reinforcement Learning (MVERL) framework for placing applications among appropriate fog nodes. The quality of the application placement policies is good in terms of metrics related to quality like correctness, model relevance, $in$-differential privacy, and robustness. The performance results of the proposed MVERL are evaluated considering fog nodes with both limited and unlimited processors. The simulation found that the proposed MVERL outperforms existing works concerning a few performance metrics.
{"title":"MVE-based Reinforcement Learning Framework with Explainability for improving Quality of Experience of Application Placement in Fog Computing","authors":"Bhargavi Krishnamurthy, S. Shiva, Saikat Das, Ph.D.","doi":"10.1109/aiiot54504.2022.9817331","DOIUrl":"https://doi.org/10.1109/aiiot54504.2022.9817331","url":null,"abstract":"Fog computing can process big data generated by the IoT (IoT) architectures. The hierarchical, heterogeneous and distributed form of fog computing makes the application placement a challenging task. IoT applications are time-sensitive, and their placement decision is dependent on the user's Quality of Experience (QoE). This paper proposes an explainable Model Value Evaluation based Reinforcement Learning (MVERL) framework for placing applications among appropriate fog nodes. The quality of the application placement policies is good in terms of metrics related to quality like correctness, model relevance, $in$-differential privacy, and robustness. The performance results of the proposed MVERL are evaluated considering fog nodes with both limited and unlimited processors. The simulation found that the proposed MVERL outperforms existing works concerning a few performance metrics.","PeriodicalId":409264,"journal":{"name":"2022 IEEE World AI IoT Congress (AIIoT)","volume":"45 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134147491","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：