Pub Date : 2023-04-05DOI: 10.1109/ICNWC57852.2023.10127355
P. Muthuvel, T. Jaswanth, S. Firoz, S. Sri, N. Mukhesh
In the domain of artificial intelligence, it’s becoming more crucial than ever to classify emotions from both text and speech (AI). In order to promote and enhance human-ma-chine interaction, it is essential to establish a broader frame-work for speech emotion recognition. Machines are currently unable to reliably classify human emotions, hence machine learning development models were created for this purpose. Many academics worldwide are attempting to improve the ac-curacy of emotion categorization systems. The two steps of this study’s creation of a speech emotion detection model are (I) tasked with managing and (ii) classification. The most pertinent feature subset was discovered using feature selection (FS). A wide variety of different vision -based paradigms were employed to address the growing demand for accurate emotion categorization all across the domain of ai technology, taking into account how crucial feature selection is. This study strategy for both the emotion categorization problem and the establishment of ml algorithms and deep learning methods. This same aforementioned work focuses on speech expression analysis & proposes a paradigm for bettering human-computer interaction through into the construction on prototype cognitive computing that categorizes feelings. The investigation aims to boost this same precision for eg in voice by applying methods for selecting features and now a spectrum different deep learning methodology, notably TensorFlow. A research also high-lights the contribution on component choice mostly in creation of powerful machine-learning algorithms towards feelings categorization.
{"title":"Emotion Recognition in Speech Signals using MFCC and Mel-Spectrogram Analysis","authors":"P. Muthuvel, T. Jaswanth, S. Firoz, S. Sri, N. Mukhesh","doi":"10.1109/ICNWC57852.2023.10127355","DOIUrl":"https://doi.org/10.1109/ICNWC57852.2023.10127355","url":null,"abstract":"In the domain of artificial intelligence, it’s becoming more crucial than ever to classify emotions from both text and speech (AI). In order to promote and enhance human-ma-chine interaction, it is essential to establish a broader frame-work for speech emotion recognition. Machines are currently unable to reliably classify human emotions, hence machine learning development models were created for this purpose. Many academics worldwide are attempting to improve the ac-curacy of emotion categorization systems. The two steps of this study’s creation of a speech emotion detection model are (I) tasked with managing and (ii) classification. The most pertinent feature subset was discovered using feature selection (FS). A wide variety of different vision -based paradigms were employed to address the growing demand for accurate emotion categorization all across the domain of ai technology, taking into account how crucial feature selection is. This study strategy for both the emotion categorization problem and the establishment of ml algorithms and deep learning methods. This same aforementioned work focuses on speech expression analysis & proposes a paradigm for bettering human-computer interaction through into the construction on prototype cognitive computing that categorizes feelings. The investigation aims to boost this same precision for eg in voice by applying methods for selecting features and now a spectrum different deep learning methodology, notably TensorFlow. A research also high-lights the contribution on component choice mostly in creation of powerful machine-learning algorithms towards feelings categorization.","PeriodicalId":197525,"journal":{"name":"2023 International Conference on Networking and Communications (ICNWC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-04-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127785640","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}
Massive MIMO systems are being incorporated in 5G wireless networks owing to its high spectral efficiency. In order to achieve this efficiency, we require accurate Channel State Information (CSI) which is acquired by a training performing pilot transmission, CSI estimation and feedback. In this work, a novel technique for performing this task is proposed where channel estimation is performed at the base station. The work also proposes pilot compression for this system model. The base station sends compressed pilots to the user equipment in the downlink channel which amplifies and forwards the received signal and relays it back to the base station in the uplink channel. The performance analysis for this system model has been simulated using MATLAB and is expressed in terms of the NMSE values for different levels of compression.
{"title":"Pilot Compression Analysis for Feedback Based Channel Estimation Model in FDD Massive MIMO","authors":"Madhumitha Jayaram, Bhagyaveni Marcharla Anjaneyulu","doi":"10.1109/ICNWC57852.2023.10127323","DOIUrl":"https://doi.org/10.1109/ICNWC57852.2023.10127323","url":null,"abstract":"Massive MIMO systems are being incorporated in 5G wireless networks owing to its high spectral efficiency. In order to achieve this efficiency, we require accurate Channel State Information (CSI) which is acquired by a training performing pilot transmission, CSI estimation and feedback. In this work, a novel technique for performing this task is proposed where channel estimation is performed at the base station. The work also proposes pilot compression for this system model. The base station sends compressed pilots to the user equipment in the downlink channel which amplifies and forwards the received signal and relays it back to the base station in the uplink channel. The performance analysis for this system model has been simulated using MATLAB and is expressed in terms of the NMSE values for different levels of compression.","PeriodicalId":197525,"journal":{"name":"2023 International Conference on Networking and Communications (ICNWC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-04-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127923717","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 : 2023-04-05DOI: 10.1109/ICNWC57852.2023.10127558
D. Malathi, S. Gopika, Devina Awasthi, Dorathi Jayaseeli
The internet has emerged as one of the most crucial elements of modern life. Every single person uses the internet to access knowledge, information, and all communication tools available to them. However, many who are visually impaired find it difficult to use those features and need outside aid to complete their tasks. People with visual impairments all around the world now have a wide range of new opportunities because of the invention of computers. Screen readers, audio-based environments, and other assistive features have made it easier for blind persons to utilize the workspace. Today, email is required to send confidential information. Email is a type of technology that facilitates business correspondence and lets users transmit messages to other people. The main objective of this work is to develop a voice-based email system that will enable people who are blind or visually impaired to send and receive emails using computers. It will make advantage of modern features to create a working environment that enables persons with visual impairments to do their jobs independently.
{"title":"Voice Automation Mail System for Visually Impaired","authors":"D. Malathi, S. Gopika, Devina Awasthi, Dorathi Jayaseeli","doi":"10.1109/ICNWC57852.2023.10127558","DOIUrl":"https://doi.org/10.1109/ICNWC57852.2023.10127558","url":null,"abstract":"The internet has emerged as one of the most crucial elements of modern life. Every single person uses the internet to access knowledge, information, and all communication tools available to them. However, many who are visually impaired find it difficult to use those features and need outside aid to complete their tasks. People with visual impairments all around the world now have a wide range of new opportunities because of the invention of computers. Screen readers, audio-based environments, and other assistive features have made it easier for blind persons to utilize the workspace. Today, email is required to send confidential information. Email is a type of technology that facilitates business correspondence and lets users transmit messages to other people. The main objective of this work is to develop a voice-based email system that will enable people who are blind or visually impaired to send and receive emails using computers. It will make advantage of modern features to create a working environment that enables persons with visual impairments to do their jobs independently.","PeriodicalId":197525,"journal":{"name":"2023 International Conference on Networking and Communications (ICNWC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-04-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133607349","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 : 2023-04-05DOI: 10.1109/ICNWC57852.2023.10127500
R. Deepa, R. Priscilla, A. Pandi, B. Renukadevi
Chronic kidney disease (CKD) or chronic renal disease-has become a major issue with a steady growth rate. A person can survive for a maximum of 18 days, which makes a huge demand for a kidney transplant and dialysis. It is necessary to have a good model to predict this disease at an earlier stage. It can be identified using ML models. This proposal proposes a workflow to predict CKD status based on the pre-processing steps of clinical data collection, incorporating data, handling missing values with collaborative filters, and attribute selection. This proposal used seven machine models and will compare all the models and the extra tree classifier and decision tree to ensure high accuracy and minimal bias for the attribute. This research also focuses on the real-time aspects of data collection and highlights the importance of domain knowledge when using machine learning for CKD status prediction. The evolution of the proposed model shows that the model can predict CKD with an accuracy of 98.65%.
{"title":"An Early Prediction Model for Chronic Kidney Disease Using Machine Learning","authors":"R. Deepa, R. Priscilla, A. Pandi, B. Renukadevi","doi":"10.1109/ICNWC57852.2023.10127500","DOIUrl":"https://doi.org/10.1109/ICNWC57852.2023.10127500","url":null,"abstract":"Chronic kidney disease (CKD) or chronic renal disease-has become a major issue with a steady growth rate. A person can survive for a maximum of 18 days, which makes a huge demand for a kidney transplant and dialysis. It is necessary to have a good model to predict this disease at an earlier stage. It can be identified using ML models. This proposal proposes a workflow to predict CKD status based on the pre-processing steps of clinical data collection, incorporating data, handling missing values with collaborative filters, and attribute selection. This proposal used seven machine models and will compare all the models and the extra tree classifier and decision tree to ensure high accuracy and minimal bias for the attribute. This research also focuses on the real-time aspects of data collection and highlights the importance of domain knowledge when using machine learning for CKD status prediction. The evolution of the proposed model shows that the model can predict CKD with an accuracy of 98.65%.","PeriodicalId":197525,"journal":{"name":"2023 International Conference on Networking and Communications (ICNWC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-04-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134485379","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 : 2023-04-05DOI: 10.1109/ICNWC57852.2023.10127236
K. Anand, M. Priyadharshini, K. Priyadharshini
This paper proposes the Gzip algorithm for image and document compression and decompression. Gzip is a hybrid algorithm that combines Lz77 and Huffman. In document management and communication systems, picture and document compression and decompression are crucial. Image and document compression technologies are used to lower the amount of data required to represent the file. Image compression has proven to be the most advantageous and practical method in the field of digital image processing. The goal is to reduce the images’ and documents’ redundancy so that data may be stored or sent efficiently. In order to reduce data redundancy and conserve more hardware space and transmission bandwidth, the theory of data compression and decompression is therefore becoming more and more important. Compression is beneficial because it makes use of less expensive resources like hard disc space and transmission bandwidth. When we evaluate the image quality, decompression is beneficial. In the proposed system, there is no reduction in data, but there is a decrease in data size without loss of quality.
{"title":"Compression And Decompression Of Files Without Loss Of Quality","authors":"K. Anand, M. Priyadharshini, K. Priyadharshini","doi":"10.1109/ICNWC57852.2023.10127236","DOIUrl":"https://doi.org/10.1109/ICNWC57852.2023.10127236","url":null,"abstract":"This paper proposes the Gzip algorithm for image and document compression and decompression. Gzip is a hybrid algorithm that combines Lz77 and Huffman. In document management and communication systems, picture and document compression and decompression are crucial. Image and document compression technologies are used to lower the amount of data required to represent the file. Image compression has proven to be the most advantageous and practical method in the field of digital image processing. The goal is to reduce the images’ and documents’ redundancy so that data may be stored or sent efficiently. In order to reduce data redundancy and conserve more hardware space and transmission bandwidth, the theory of data compression and decompression is therefore becoming more and more important. Compression is beneficial because it makes use of less expensive resources like hard disc space and transmission bandwidth. When we evaluate the image quality, decompression is beneficial. In the proposed system, there is no reduction in data, but there is a decrease in data size without loss of quality.","PeriodicalId":197525,"journal":{"name":"2023 International Conference on Networking and Communications (ICNWC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-04-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115045290","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 : 2023-04-05DOI: 10.1109/ICNWC57852.2023.10127363
Munnaza Ramzan, G. M. Rather
The study of a new unknown phenomenon/system begins with an experimental/ observational study. Statistical and regression analysis of the recorded random data is carried out to examine the characteristic features and behavior of the new phenomenon/system. The recorded data and observed statistical features are used to develop a mathematical model which closely represents the system. This helps in duplicating the new systems through simulation studies. To observe the behavior of dependent output response vis-à-vis independent input to the system under observation, curve fitting techniques are used. Most commonly used being least square based linear regression and non-linear regression techniques. These techniques have their own merits and demerits. In this paper a new polynomial based regression technique is presented. The technique performs exceptionally well within the given range of the independent variable and perfectly maps the observed points to the curve. It helps in predicting the values of the dependent variable with good accuracy in close proximity of the considered independent variable range.
{"title":"A Polynomial Curve Mapping Technique for Random Data","authors":"Munnaza Ramzan, G. M. Rather","doi":"10.1109/ICNWC57852.2023.10127363","DOIUrl":"https://doi.org/10.1109/ICNWC57852.2023.10127363","url":null,"abstract":"The study of a new unknown phenomenon/system begins with an experimental/ observational study. Statistical and regression analysis of the recorded random data is carried out to examine the characteristic features and behavior of the new phenomenon/system. The recorded data and observed statistical features are used to develop a mathematical model which closely represents the system. This helps in duplicating the new systems through simulation studies. To observe the behavior of dependent output response vis-à-vis independent input to the system under observation, curve fitting techniques are used. Most commonly used being least square based linear regression and non-linear regression techniques. These techniques have their own merits and demerits. In this paper a new polynomial based regression technique is presented. The technique performs exceptionally well within the given range of the independent variable and perfectly maps the observed points to the curve. It helps in predicting the values of the dependent variable with good accuracy in close proximity of the considered independent variable range.","PeriodicalId":197525,"journal":{"name":"2023 International Conference on Networking and Communications (ICNWC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-04-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116052686","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 : 2023-04-05DOI: 10.1109/ICNWC57852.2023.10127259
Ms.S.Anitha Jebamani, M. Ragavi, K. Nivetha.
The direct link between computer systems and the human brain is known as a “brain computer interface,” or BCI. The BCI reads the waves composed of the brain at exclusive places inside the human head, translates those indicators into movements and instructions,which can control the computer systems. We endorse combining this generation with home automation. This interface device is especially helpful for people who are severely disabled or confined and lack reliable muscular control over the parts of their bodies that are engaged with surrounding peripherals. The machine entails two components: an EEG sensor circuit and Arduino microcontroller board. The mind waves are captured using electrodes. These indicators are filtered and amplified to take away noise. These analog signs are converted to digital. The digital alerts are decoded and are used to exchange on a device
{"title":"Biometric Recognition Using EEG Signals And Controlling The Electrical Devices","authors":"Ms.S.Anitha Jebamani, M. Ragavi, K. Nivetha.","doi":"10.1109/ICNWC57852.2023.10127259","DOIUrl":"https://doi.org/10.1109/ICNWC57852.2023.10127259","url":null,"abstract":"The direct link between computer systems and the human brain is known as a “brain computer interface,” or BCI. The BCI reads the waves composed of the brain at exclusive places inside the human head, translates those indicators into movements and instructions,which can control the computer systems. We endorse combining this generation with home automation. This interface device is especially helpful for people who are severely disabled or confined and lack reliable muscular control over the parts of their bodies that are engaged with surrounding peripherals. The machine entails two components: an EEG sensor circuit and Arduino microcontroller board. The mind waves are captured using electrodes. These indicators are filtered and amplified to take away noise. These analog signs are converted to digital. The digital alerts are decoded and are used to exchange on a device","PeriodicalId":197525,"journal":{"name":"2023 International Conference on Networking and Communications (ICNWC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-04-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123452587","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 : 2023-04-05DOI: 10.1109/ICNWC57852.2023.10127520
N. Neeraj, Aditya Nellikeri, P. Varun, Santosh Reddy, Mangesh Shanbhag, Dg Narayan, Altaf Husain
With the advancement of cloud computing technologies, it is necessary to maintain quality of service, for which service level agreements (SLAs) are one of the ways to improve this. A significant part of revenue will be contributed by the use of on demand resources (compute, network or storage) provided by this cloud service providers (CSP). Service level agreement (SLA) plays a key role in maintaining quality of services (QoS) of these on demand resources. The CSP enforces the SLA, by utilizing machine-generated logs to monitor machine performance. Additionally, some consumers may monitor resource performance themselves to ensure it meets their standards. This approach not only leads to traction between both CSP and cloud service consumers (CSC), but also leads to a duplication of the effort. The aim of this work is to create a system that is distributed in nature and could be trusted by both CSP and the consumer. The paper provides solution using a public blockchain and log based algorithm that evaluates resource performance based on several SLA criteria accepted by the CSP and consumers. If the system’s performance falls below a certain level, the CSC will be intimated through logs. This motivates the consumer to spend more money on the resources and compels the CSP to keep the number of SLA violations to a minimal. A web application with a blockchain as the back-end is used to monitor SLA violations and the compensation procedure. Finally, a multi-node ethereum blockchain network is used to conduct a performance analysis of the suggested solution.
{"title":"Service Level Agreement Violation Detection in Multi-cloud Environment using Ethereum Blockchain","authors":"N. Neeraj, Aditya Nellikeri, P. Varun, Santosh Reddy, Mangesh Shanbhag, Dg Narayan, Altaf Husain","doi":"10.1109/ICNWC57852.2023.10127520","DOIUrl":"https://doi.org/10.1109/ICNWC57852.2023.10127520","url":null,"abstract":"With the advancement of cloud computing technologies, it is necessary to maintain quality of service, for which service level agreements (SLAs) are one of the ways to improve this. A significant part of revenue will be contributed by the use of on demand resources (compute, network or storage) provided by this cloud service providers (CSP). Service level agreement (SLA) plays a key role in maintaining quality of services (QoS) of these on demand resources. The CSP enforces the SLA, by utilizing machine-generated logs to monitor machine performance. Additionally, some consumers may monitor resource performance themselves to ensure it meets their standards. This approach not only leads to traction between both CSP and cloud service consumers (CSC), but also leads to a duplication of the effort. The aim of this work is to create a system that is distributed in nature and could be trusted by both CSP and the consumer. The paper provides solution using a public blockchain and log based algorithm that evaluates resource performance based on several SLA criteria accepted by the CSP and consumers. If the system’s performance falls below a certain level, the CSC will be intimated through logs. This motivates the consumer to spend more money on the resources and compels the CSP to keep the number of SLA violations to a minimal. A web application with a blockchain as the back-end is used to monitor SLA violations and the compensation procedure. Finally, a multi-node ethereum blockchain network is used to conduct a performance analysis of the suggested solution.","PeriodicalId":197525,"journal":{"name":"2023 International Conference on Networking and Communications (ICNWC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-04-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122906766","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 : 2023-04-05DOI: 10.1109/ICNWC57852.2023.10127244
Ambika S Rao, R. Anilkumar, K. Padmapriya, K. Sudeendra Kumar
Commercial-grade electronic components are finding their way into spacecrafts intended for low-orbit applications due to their low cost and ease of availability. However, being susceptible to a variety of disturbances like radiation upsets, they are not as reliable as radiation-hardened space-grade components. Reliability of these components can be improved using design techniques for fault detection that allows for subsequent correction through hardware or software. In this paper, a fault monitoring mechanism is introduced to improve the reliability of an Embedded RISC-V System-on-Chip (SoC) intended for low-cost space applications. This mechanism alerts the system of potential faults induced by upsets in space environment and allows for recovery. Implementation of the mechanism is done in an ASIC with 180nm commercial foundry libraries, but the design can also be ported to SRAM-based FPGAs.
{"title":"Improving Reliability of Embedded RISC-V SoC for Low-cost Space Applications","authors":"Ambika S Rao, R. Anilkumar, K. Padmapriya, K. Sudeendra Kumar","doi":"10.1109/ICNWC57852.2023.10127244","DOIUrl":"https://doi.org/10.1109/ICNWC57852.2023.10127244","url":null,"abstract":"Commercial-grade electronic components are finding their way into spacecrafts intended for low-orbit applications due to their low cost and ease of availability. However, being susceptible to a variety of disturbances like radiation upsets, they are not as reliable as radiation-hardened space-grade components. Reliability of these components can be improved using design techniques for fault detection that allows for subsequent correction through hardware or software. In this paper, a fault monitoring mechanism is introduced to improve the reliability of an Embedded RISC-V System-on-Chip (SoC) intended for low-cost space applications. This mechanism alerts the system of potential faults induced by upsets in space environment and allows for recovery. Implementation of the mechanism is done in an ASIC with 180nm commercial foundry libraries, but the design can also be ported to SRAM-based FPGAs.","PeriodicalId":197525,"journal":{"name":"2023 International Conference on Networking and Communications (ICNWC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-04-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129824024","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 : 2023-04-05DOI: 10.1109/ICNWC57852.2023.10127450
Arshdeep Janjua, S. Dhalla, Savita Gupta, Sukhwinder Singh
Federated learning (FL) has undergone substantial research and has been used in numerous real-world solutions over the past few years. It shows promising results in addressing the data security and privacy issues present in the traditional centralized machine learning approach. Even though federated learning makes certain of data privacy for each contributing user, the global model and the data are still vulnerable to attacks by compromised clients and servers. Additionally, in the settings for non-independent identical data (Non-IID), federated learning performs significantly less compared to the standard centralized learning mode. To address both the security and performance issues, this paper proposes a blockchain-enabled gossip federated learning framework (BGFL). BGFL replaces the central server with a blockchain-enabled system for global model storage and exchange. Also, to achieve faster training convergence, clients communicate with each other based on a gossip training approach. Then, to evaluate the performance we perform experiments using MNIST and CIFAR datasets in Non-IID settings. The performance and effectiveness of the BGFL framework is demonstrated by the experimental results.
{"title":"A Blockchain-Enabled Decentralized Gossip Federated Learning Framework","authors":"Arshdeep Janjua, S. Dhalla, Savita Gupta, Sukhwinder Singh","doi":"10.1109/ICNWC57852.2023.10127450","DOIUrl":"https://doi.org/10.1109/ICNWC57852.2023.10127450","url":null,"abstract":"Federated learning (FL) has undergone substantial research and has been used in numerous real-world solutions over the past few years. It shows promising results in addressing the data security and privacy issues present in the traditional centralized machine learning approach. Even though federated learning makes certain of data privacy for each contributing user, the global model and the data are still vulnerable to attacks by compromised clients and servers. Additionally, in the settings for non-independent identical data (Non-IID), federated learning performs significantly less compared to the standard centralized learning mode. To address both the security and performance issues, this paper proposes a blockchain-enabled gossip federated learning framework (BGFL). BGFL replaces the central server with a blockchain-enabled system for global model storage and exchange. Also, to achieve faster training convergence, clients communicate with each other based on a gossip training approach. Then, to evaluate the performance we perform experiments using MNIST and CIFAR datasets in Non-IID settings. The performance and effectiveness of the BGFL framework is demonstrated by the experimental results.","PeriodicalId":197525,"journal":{"name":"2023 International Conference on Networking and Communications (ICNWC)","volume":null,"pages":null},"PeriodicalIF":0.0,"publicationDate":"2023-04-05","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130563111","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：