Pub Date : 2022-06-06DOI: 10.1109/aiiot54504.2022.9817159
S. K. Mahmud, Yue Chen, K. K. Chai
In this work we present an ensemble reinforcement learning (ERL) framework comprising of deep-Q networks (DQNs). The aim is to optimize sum rate for non orthogonal multiple access unmanned aerial network (NOMA-UAV) network. Power in downlink (DL) and bandwidth allotment for a NOMA cluster is managed over fixed UAV trajectory. The environment is dynamic and quality of service (QoS) requirements are varying for each node on ground. A comparative analysis between conventional reinforcement learning (CRL) framework and proposed ensemble of ERL yields a performance gain in undermentioned metrics. The ERL achieves 20 percent performance gain in average sum rate and the gain in spectral efficiency is 2 percent, over conventional reinforcement learning framework with single DQN. It also achieves high performance over different UAV speeds in cumulative sum rate and device coverage.
{"title":"Ensemble Reinforcement Learning Framework for Sum Rate Optimization in NOMA-UAV Network","authors":"S. K. Mahmud, Yue Chen, K. K. Chai","doi":"10.1109/aiiot54504.2022.9817159","DOIUrl":"https://doi.org/10.1109/aiiot54504.2022.9817159","url":null,"abstract":"In this work we present an ensemble reinforcement learning (ERL) framework comprising of deep-Q networks (DQNs). The aim is to optimize sum rate for non orthogonal multiple access unmanned aerial network (NOMA-UAV) network. Power in downlink (DL) and bandwidth allotment for a NOMA cluster is managed over fixed UAV trajectory. The environment is dynamic and quality of service (QoS) requirements are varying for each node on ground. A comparative analysis between conventional reinforcement learning (CRL) framework and proposed ensemble of ERL yields a performance gain in undermentioned metrics. The ERL achieves 20 percent performance gain in average sum rate and the gain in spectral efficiency is 2 percent, over conventional reinforcement learning framework with single DQN. It also achieves high performance over different UAV speeds in cumulative sum rate and device coverage.","PeriodicalId":409264,"journal":{"name":"2022 IEEE World AI IoT Congress (AIIoT)","volume":"50 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":"132397382","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/aiiot54504.2022.9817252
Carlos Jose Martinez, S. Galmés
Technological evolution and the current situation in the world have given rise to new emerging realities in medicine, such as point-of-care diagnosis and individualized health service without limitations such as time and place. Thanks to IoMT, new forms of contact between healthcare providers and patients have been developed using the Internet. This research aims to analyze the security of IoMT communication protocols and their limitations. For this purpose, an exhaustive review of the literature on intelligent healthcare is performed because there are many articles published in indexed journals that have addressed these issues in recent years. This review involves a bibliometric analysis of the variables Protocols, Communication, and IoMT, which bases its search on Scopus to fulfill the objective. A total of 27 documents published during the period 2015–2020 were identified as 2019, the year in which the most significant record was achieved with 11 publications; it is also established that 51.9% of the identified documents are of type Journal Article 33.3 % Conference Articles. Next, this literature is analyzed, particularly those publications emphasizing the security measures of the protocols, their vulnerabilities, and the attacks to which IoMT implementations are subjected, to make known the position of different authors regarding the subject proposed in this document. Finally, future research directions are provided to further progress in constructing theories related to the topic studied.
{"title":"Analysis of the primary attacks on IoMT Internet of Medical Things communications protocols","authors":"Carlos Jose Martinez, S. Galmés","doi":"10.1109/aiiot54504.2022.9817252","DOIUrl":"https://doi.org/10.1109/aiiot54504.2022.9817252","url":null,"abstract":"Technological evolution and the current situation in the world have given rise to new emerging realities in medicine, such as point-of-care diagnosis and individualized health service without limitations such as time and place. Thanks to IoMT, new forms of contact between healthcare providers and patients have been developed using the Internet. This research aims to analyze the security of IoMT communication protocols and their limitations. For this purpose, an exhaustive review of the literature on intelligent healthcare is performed because there are many articles published in indexed journals that have addressed these issues in recent years. This review involves a bibliometric analysis of the variables Protocols, Communication, and IoMT, which bases its search on Scopus to fulfill the objective. A total of 27 documents published during the period 2015–2020 were identified as 2019, the year in which the most significant record was achieved with 11 publications; it is also established that 51.9% of the identified documents are of type Journal Article 33.3 % Conference Articles. Next, this literature is analyzed, particularly those publications emphasizing the security measures of the protocols, their vulnerabilities, and the attacks to which IoMT implementations are subjected, to make known the position of different authors regarding the subject proposed in this document. Finally, future research directions are provided to further progress in constructing theories related to the topic studied.","PeriodicalId":409264,"journal":{"name":"2022 IEEE World AI IoT Congress (AIIoT)","volume":"328 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":"133569842","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/aiiot54504.2022.9817372
Christopher Le, Tauheed Khan Mohd
Detecting faces in low-light environments is an important new technology and have been under development for years. In surveillance, some security cameras with thermal technology can recognize humans based on the heat that the object radiates. However, with only thermal techniques, it is still challenging to recognize specific people. Recognizing human with heat vision makes it hard to tell the identity of the person with existing Computer Vision techniques such as CNN. In this research, we present a system to recognize human faces in a low-light environment by enhancing low-light images and applying facial detection to them. Another technique of image super-resolution will also be applied to enhance the quality of the images for better detection.
{"title":"Facial Detection in Low Light Environments Using OpenCV","authors":"Christopher Le, Tauheed Khan Mohd","doi":"10.1109/aiiot54504.2022.9817372","DOIUrl":"https://doi.org/10.1109/aiiot54504.2022.9817372","url":null,"abstract":"Detecting faces in low-light environments is an important new technology and have been under development for years. In surveillance, some security cameras with thermal technology can recognize humans based on the heat that the object radiates. However, with only thermal techniques, it is still challenging to recognize specific people. Recognizing human with heat vision makes it hard to tell the identity of the person with existing Computer Vision techniques such as CNN. In this research, we present a system to recognize human faces in a low-light environment by enhancing low-light images and applying facial detection to them. Another technique of image super-resolution will also be applied to enhance the quality of the images for better detection.","PeriodicalId":409264,"journal":{"name":"2022 IEEE World AI IoT Congress (AIIoT)","volume":"48 22","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-06-06","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114044062","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/aiiot54504.2022.9817311
Elsa Jelista Sari, Y. Priyadi, Rosa Reska Riskiana
The SRS used in this study is an application called “Sipjabs”. This application processes data regarding the position of human resources to meet the needs of a company. This research aims to implement semantic textual similarity in software requirements specification through functional requirements with use case diagrams using the Wu Palmer (WUP) method in finding semantics. This research method is presented in a flow chart consisting of three main activities: research object analysis, semantic textual similarity, and validity and reliability testing. In this research, an extraction process for the Requirement Specification has been produced, divided into five documents: FR01, FR02, FR03, FR04, FR05. Then the steps performed in the use case description are divided into UD01, UD02, UD03, UD04, UD05. The highest similarity value is found in documents UD03 and FR03, where the number of similarities is 0.626640. In addition, the highest score of the sentence that has been calculated using the Wu Palmer concept is 0.8000, which is found in the words “page” and “user”. The highest kappa value with Gwet's AC1 formula is 0.02547770700636931, which means “Fair Agreement”. For the results of the calculation of the questionnaire filled in by the expert, namely 0.82022, which means “Almost Perfect”.
{"title":"Implementation of Semantic Textual Similarity between Requirement Specification and Use Case Description Using WUP Method (Case Study: Sipjabs Application)","authors":"Elsa Jelista Sari, Y. Priyadi, Rosa Reska Riskiana","doi":"10.1109/aiiot54504.2022.9817311","DOIUrl":"https://doi.org/10.1109/aiiot54504.2022.9817311","url":null,"abstract":"The SRS used in this study is an application called “Sipjabs”. This application processes data regarding the position of human resources to meet the needs of a company. This research aims to implement semantic textual similarity in software requirements specification through functional requirements with use case diagrams using the Wu Palmer (WUP) method in finding semantics. This research method is presented in a flow chart consisting of three main activities: research object analysis, semantic textual similarity, and validity and reliability testing. In this research, an extraction process for the Requirement Specification has been produced, divided into five documents: FR01, FR02, FR03, FR04, FR05. Then the steps performed in the use case description are divided into UD01, UD02, UD03, UD04, UD05. The highest similarity value is found in documents UD03 and FR03, where the number of similarities is 0.626640. In addition, the highest score of the sentence that has been calculated using the Wu Palmer concept is 0.8000, which is found in the words “page” and “user”. The highest kappa value with Gwet's AC1 formula is 0.02547770700636931, which means “Fair Agreement”. For the results of the calculation of the questionnaire filled in by the expert, namely 0.82022, which means “Almost Perfect”.","PeriodicalId":409264,"journal":{"name":"2022 IEEE World AI IoT Congress (AIIoT)","volume":"28 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":"128684175","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/aiiot54504.2022.9817151
Bhargavi Krishnamurthy, S. Shiva
Because of exponential growth in the availability of large number of Internet of Things (IoT) devices there is an increase in the latency of IoT applications that is managed by performing computation on edge devices/fog nodes. Man-in-the-Middle (MitM) attack is very much common in fog computing as the Fog computing architecture is vulnerable to MitM attack because of the positioning of fog nodes between cloud and end devices. Several machine learning approaches are designed and developed in literature for detection of MitM attacks in fog computing but they lack interpretability/explainability feature. In this paper a novel interpretable Q-learning algorithm with soft actor critic approach is designed for detecting MitM attacks in Fog computing with proper reasoning. Entropy regularized reinforcement learning is performed at each time step which prevents the loss during of every Q-function during approximation of the target. The attack detection policies formulated are of high quality as it satisfies the quality assurance metrics of robustness and correctness the conduct of the proposed interpretable Q-learning framework is encouraging towards the metrics like latency, attack detection time, energy consumption, and accuracy.
{"title":"Man-in-the-Middle attack Explainer for Fog computing using Soft Actor Critic Q-Learning Approach","authors":"Bhargavi Krishnamurthy, S. Shiva","doi":"10.1109/aiiot54504.2022.9817151","DOIUrl":"https://doi.org/10.1109/aiiot54504.2022.9817151","url":null,"abstract":"Because of exponential growth in the availability of large number of Internet of Things (IoT) devices there is an increase in the latency of IoT applications that is managed by performing computation on edge devices/fog nodes. Man-in-the-Middle (MitM) attack is very much common in fog computing as the Fog computing architecture is vulnerable to MitM attack because of the positioning of fog nodes between cloud and end devices. Several machine learning approaches are designed and developed in literature for detection of MitM attacks in fog computing but they lack interpretability/explainability feature. In this paper a novel interpretable Q-learning algorithm with soft actor critic approach is designed for detecting MitM attacks in Fog computing with proper reasoning. Entropy regularized reinforcement learning is performed at each time step which prevents the loss during of every Q-function during approximation of the target. The attack detection policies formulated are of high quality as it satisfies the quality assurance metrics of robustness and correctness the conduct of the proposed interpretable Q-learning framework is encouraging towards the metrics like latency, attack detection time, energy consumption, and accuracy.","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":"121773415","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/aiiot54504.2022.9817225
Joseph Squillace, May Bantan
Advancements in technology have brought with it widespread use and utilization of the lnternet-of-Things (IoT); a catch-all moniker identifying the collection of eclectic hardware and software mediums connecting people with small electronic computing and monitoring devices, such as SMART products: television, thermostat, speaker, refrigerator, etc. As personal IoT adaptation has grown in parallel with IoT home integration, so too have IoT concerns related to Privacy, Trust, and the Security risks associated with individual end-user protection, especially from IoT devices linked to Facebook (Meta) (F), Apple (A), Amazon (A), Netflix (N), and Google (G); conversationally identifiable as F(M).A.A.N.G. corporations. As a result, the list of IoT threats, security weaknesses, and vulnerabilities has grown exponentially. Nefarious actors have taken advantage of this growing technology, reaching into homes and offices far beyond established limits set by end-users. This research will use a multiple case study approach to analyze IoT security breach events to identify current Privacy, Trust, and Security risks to end-users associated with IoT devices. Furthermore, this research will also utilize security behavioral research by Anderson and Agarwal to introduce a Safe Computing Practices Model (SCPM) that can be implemented by end-users when integrating IoT devices.
{"title":"A Taxonomy of Privacy, Trust, and Security Breach Incidents of Internet-of-Things Linked to F(M).A.A.N.G. Corporations","authors":"Joseph Squillace, May Bantan","doi":"10.1109/aiiot54504.2022.9817225","DOIUrl":"https://doi.org/10.1109/aiiot54504.2022.9817225","url":null,"abstract":"Advancements in technology have brought with it widespread use and utilization of the lnternet-of-Things (IoT); a catch-all moniker identifying the collection of eclectic hardware and software mediums connecting people with small electronic computing and monitoring devices, such as SMART products: television, thermostat, speaker, refrigerator, etc. As personal IoT adaptation has grown in parallel with IoT home integration, so too have IoT concerns related to Privacy, Trust, and the Security risks associated with individual end-user protection, especially from IoT devices linked to Facebook (Meta) (F), Apple (A), Amazon (A), Netflix (N), and Google (G); conversationally identifiable as F(M).A.A.N.G. corporations. As a result, the list of IoT threats, security weaknesses, and vulnerabilities has grown exponentially. Nefarious actors have taken advantage of this growing technology, reaching into homes and offices far beyond established limits set by end-users. This research will use a multiple case study approach to analyze IoT security breach events to identify current Privacy, Trust, and Security risks to end-users associated with IoT devices. Furthermore, this research will also utilize security behavioral research by Anderson and Agarwal to introduce a Safe Computing Practices Model (SCPM) that can be implemented by end-users when integrating IoT devices.","PeriodicalId":409264,"journal":{"name":"2022 IEEE World AI IoT Congress (AIIoT)","volume":"28 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":"130334992","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/aiiot54504.2022.9817175
Christopher Faircloth, Gavin Hartzell, Nathan Callahan, S. Bhunia
The 2021 T-Mobile breach conducted by John Erin Binns resulted in the theft of 54 million customers' personal data. The attacker gained entry into T-Mobile's systems through an unprotected router and used brute force techniques to access the sensitive information stored on the internal servers. The data stolen included names, addresses, Social Security Numbers, birthdays, driver's license numbers, ID information, IMEIs, and IMSIs. We analyze the data breach and how it opens the door to identity theft and many other forms of hacking such as SIM Hijacking. SIM Hijacking is a form of hacking in which bad actors can take control of a victim's phone number allowing them means to bypass additional safety measures currently in place to prevent fraud. This paper thoroughly reviews the attack methodology, impact, and attempts to provide an understanding of important measures and possible defense solutions against future attacks. We also detail other social engineering attacks that can be incurred from releasing the leaked data.
{"title":"A Study on Brute Force Attack on T-Mobile Leading to SIM-Hijacking and Identity-Theft","authors":"Christopher Faircloth, Gavin Hartzell, Nathan Callahan, S. Bhunia","doi":"10.1109/aiiot54504.2022.9817175","DOIUrl":"https://doi.org/10.1109/aiiot54504.2022.9817175","url":null,"abstract":"The 2021 T-Mobile breach conducted by John Erin Binns resulted in the theft of 54 million customers' personal data. The attacker gained entry into T-Mobile's systems through an unprotected router and used brute force techniques to access the sensitive information stored on the internal servers. The data stolen included names, addresses, Social Security Numbers, birthdays, driver's license numbers, ID information, IMEIs, and IMSIs. We analyze the data breach and how it opens the door to identity theft and many other forms of hacking such as SIM Hijacking. SIM Hijacking is a form of hacking in which bad actors can take control of a victim's phone number allowing them means to bypass additional safety measures currently in place to prevent fraud. This paper thoroughly reviews the attack methodology, impact, and attempts to provide an understanding of important measures and possible defense solutions against future attacks. We also detail other social engineering attacks that can be incurred from releasing the leaked data.","PeriodicalId":409264,"journal":{"name":"2022 IEEE World AI IoT Congress (AIIoT)","volume":"40 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":"129954984","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/aiiot54504.2022.9817288
Talha Hassan, Tauheed Khan Mohd
Advancements in Internet of things (IoT) and sensor technologies along with deep learning techniques have enabled the design of several types of virtual keyboards including holographic keyboards. However, there is a need to better understand how its performance compares against the standard QWERTY keyboard in order to identify specific aspects of performance that are better as well as others than need to be further optimized. We conducted an initial two-part study. The first part was a controlled experimental study with 12 participants to see how a holographic keyboard compares with a QWERTY keyboard as a text entry tool on various metrics including task success, speed, and user satisfaction. The second part consisted of a short semi-structured interview with the participants, specifically based on the observations made by the researchers during the first part. Our initial findings indicate that although there is a speed advantage when users initially use standard QWERTY keyboard, there is a learning effect on speed with holographic keyboard and participants improve over time. Participants also indicated they would learn the holographic Keyboard quickly on a system usability sub-scale. This is still a work in progress and we aim to improve the design of the holographic keyboard based on the feedback from participants, and conduct a final summative evaluation.
{"title":"Comparison of Task Performance and User Satisfaction Between Holographic and Standard QWERTY Keyboard","authors":"Talha Hassan, Tauheed Khan Mohd","doi":"10.1109/aiiot54504.2022.9817288","DOIUrl":"https://doi.org/10.1109/aiiot54504.2022.9817288","url":null,"abstract":"Advancements in Internet of things (IoT) and sensor technologies along with deep learning techniques have enabled the design of several types of virtual keyboards including holographic keyboards. However, there is a need to better understand how its performance compares against the standard QWERTY keyboard in order to identify specific aspects of performance that are better as well as others than need to be further optimized. We conducted an initial two-part study. The first part was a controlled experimental study with 12 participants to see how a holographic keyboard compares with a QWERTY keyboard as a text entry tool on various metrics including task success, speed, and user satisfaction. The second part consisted of a short semi-structured interview with the participants, specifically based on the observations made by the researchers during the first part. Our initial findings indicate that although there is a speed advantage when users initially use standard QWERTY keyboard, there is a learning effect on speed with holographic keyboard and participants improve over time. Participants also indicated they would learn the holographic Keyboard quickly on a system usability sub-scale. This is still a work in progress and we aim to improve the design of the holographic keyboard based on the feedback from participants, and conduct a final summative evaluation.","PeriodicalId":409264,"journal":{"name":"2022 IEEE World AI IoT Congress (AIIoT)","volume":"19 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":"127900765","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/aiiot54504.2022.9817283
A. Mashhadi, Ali Tabaraei, Yuting Zhan, R. Parizi
Federated learning enables remote devices such as smartphones to train statistical models while ensuring that data remains private and secure. Performing privacy-preserving data analysis becomes increasingly crucial as our model is potentially being trained within heterogeneous and massive networks. While federated learning offers the potential to boost diversity in many existing models through on-device learning and enabling a wider range of users to participate, developing fair federated learning models is a challenging task. Throughout this paper, we propose a fairness auditing system for FL models that rely on spatial-temporal data. Borrowing tenets from mobility literature, we propose a set of metrics to define individual fairness using spatial-temporal data. We also introduce a set of approaches for measuring these metrics in distributed settings, as well as building a framework that can monitor the fairness of FL models dynamically.
{"title":"An Auditing Framework for Analyzing Fairness of Spatial-Temporal Federated Learning Applications","authors":"A. Mashhadi, Ali Tabaraei, Yuting Zhan, R. Parizi","doi":"10.1109/aiiot54504.2022.9817283","DOIUrl":"https://doi.org/10.1109/aiiot54504.2022.9817283","url":null,"abstract":"Federated learning enables remote devices such as smartphones to train statistical models while ensuring that data remains private and secure. Performing privacy-preserving data analysis becomes increasingly crucial as our model is potentially being trained within heterogeneous and massive networks. While federated learning offers the potential to boost diversity in many existing models through on-device learning and enabling a wider range of users to participate, developing fair federated learning models is a challenging task. Throughout this paper, we propose a fairness auditing system for FL models that rely on spatial-temporal data. Borrowing tenets from mobility literature, we propose a set of metrics to define individual fairness using spatial-temporal data. We also introduce a set of approaches for measuring these metrics in distributed settings, as well as building a framework that can monitor the fairness of FL models dynamically.","PeriodicalId":409264,"journal":{"name":"2022 IEEE World AI IoT Congress (AIIoT)","volume":"5 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":"117038647","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}
In recent years, the number of deaths due to suicide has increased. Suicide is becoming one of the major causes of death across the whole world. This has led to an alarming situation as it is endangering the human life. A lot of studies have been done to find the reason behind such suicides and its prevention. The literature has suggested that the detection of suicide thoughts at an early stage can help to rescue the life of people. The idea of early detection has led various researchers to carry out research in this direction. Many such studies have used machine learning and deep learning models to predict the idea of suicide. So, this paper reviews the existing study that has been performed towards detection of suicidal thoughts using machine learning and deep learning techniques.
{"title":"A Review on Suicidal Ideation Detection Based on Machine Learning and Deep Learning Techniques","authors":"Tanya Bhardwaj, Paridhi Gupta, Akshita Goyal, Akanksha Nagpal, Vivekanand Jha","doi":"10.1109/aiiot54504.2022.9817373","DOIUrl":"https://doi.org/10.1109/aiiot54504.2022.9817373","url":null,"abstract":"In recent years, the number of deaths due to suicide has increased. Suicide is becoming one of the major causes of death across the whole world. This has led to an alarming situation as it is endangering the human life. A lot of studies have been done to find the reason behind such suicides and its prevention. The literature has suggested that the detection of suicide thoughts at an early stage can help to rescue the life of people. The idea of early detection has led various researchers to carry out research in this direction. Many such studies have used machine learning and deep learning models to predict the idea of suicide. So, this paper reviews the existing study that has been performed towards detection of suicidal thoughts using machine learning and deep learning techniques.","PeriodicalId":409264,"journal":{"name":"2022 IEEE World AI IoT Congress (AIIoT)","volume":"104 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":"122391165","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：