Pub Date : 2021-09-02DOI: 10.1109/ACCESS51619.2021.9563342
V.A Sumayyabeevi, Jaimy James Poovely, N. Aswathy, S. Chinnu
Artificial neural networks are very popular and fast-growing machine learning algorithms today. There exist a large number of ways for implementing ANN into reality. Generally, the main two techniques are neuromorphic programming and neural networks. This paper presents an overview of such methods. Nowadays machine learning chips are available with a high level of parallel designs, but deep neural network requires flexible and efficient hardware structure that can be perfect for any type of neural networks. Also, varieties of hardware topologies are available for FPGA implementation. This paper explains those architectural variations and suggests a new topology. The proposed architecture adopts the systolic structure and applies to any feed forward neural networks such as Multi-Layer Perceptron (MLP), Auto Encoder (AE) and, Logic Regression (LR). Unlike other hardware neural network structures, this architecture implements a single activation function block and the largest layer only. This paper also includes the implementation of a feed-forward neural network for digit recognition (0 to 9) in the Zynq-7000 board with MNIST as the dataset. Different activation functions and different parameters of each activation function are used for the network. Changes and improvements are mentioned in this paper based on Accuracy, Operating frequency and, Resource usage. Logistic Sigmoidal functions can achieve more accuracy and performance as compared with others.
{"title":"A New Hardware Architecture for FPGA Implementation of Feed Forward Neural Networks","authors":"V.A Sumayyabeevi, Jaimy James Poovely, N. Aswathy, S. Chinnu","doi":"10.1109/ACCESS51619.2021.9563342","DOIUrl":"https://doi.org/10.1109/ACCESS51619.2021.9563342","url":null,"abstract":"Artificial neural networks are very popular and fast-growing machine learning algorithms today. There exist a large number of ways for implementing ANN into reality. Generally, the main two techniques are neuromorphic programming and neural networks. This paper presents an overview of such methods. Nowadays machine learning chips are available with a high level of parallel designs, but deep neural network requires flexible and efficient hardware structure that can be perfect for any type of neural networks. Also, varieties of hardware topologies are available for FPGA implementation. This paper explains those architectural variations and suggests a new topology. The proposed architecture adopts the systolic structure and applies to any feed forward neural networks such as Multi-Layer Perceptron (MLP), Auto Encoder (AE) and, Logic Regression (LR). Unlike other hardware neural network structures, this architecture implements a single activation function block and the largest layer only. This paper also includes the implementation of a feed-forward neural network for digit recognition (0 to 9) in the Zynq-7000 board with MNIST as the dataset. Different activation functions and different parameters of each activation function are used for the network. Changes and improvements are mentioned in this paper based on Accuracy, Operating frequency and, Resource usage. Logistic Sigmoidal functions can achieve more accuracy and performance as compared with others.","PeriodicalId":409648,"journal":{"name":"2021 2nd International Conference on Advances in Computing, Communication, Embedded and Secure Systems (ACCESS)","volume":"10 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"121015942","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 : 2021-09-02DOI: 10.1109/ACCESS51619.2021.9563304
Christdas, S. Mythry, R. Yanamshetti
Biomedical IoT applications require the amplification of biomedical signals with low frequencies and small amplitudes, while suppressing Direct Current Voltage offsets. CMOS OTA-based amplifier design that meets this requirement is a research topic of interest to neuroscience scientists and clinicians. This article describes a literature review on different kinds of OTA used in neural signal capturing applications. A 120dB high gain, 4pV/√(Hz) noise and 0.5µW ECG amplifier for heart rate analyzing sensor node used in biomedical IOT application is designed using 90nm CMOS process. Wilson current-mirror method is used in designing 1V-powered CMOS ECG OTA.
{"title":"CMOS ECG amplifier for heart rate analyzer sensor node used in Biomedical IOT applications","authors":"Christdas, S. Mythry, R. Yanamshetti","doi":"10.1109/ACCESS51619.2021.9563304","DOIUrl":"https://doi.org/10.1109/ACCESS51619.2021.9563304","url":null,"abstract":"Biomedical IoT applications require the amplification of biomedical signals with low frequencies and small amplitudes, while suppressing Direct Current Voltage offsets. CMOS OTA-based amplifier design that meets this requirement is a research topic of interest to neuroscience scientists and clinicians. This article describes a literature review on different kinds of OTA used in neural signal capturing applications. A 120dB high gain, 4pV/√(Hz) noise and 0.5µW ECG amplifier for heart rate analyzing sensor node used in biomedical IOT application is designed using 90nm CMOS process. Wilson current-mirror method is used in designing 1V-powered CMOS ECG OTA.","PeriodicalId":409648,"journal":{"name":"2021 2nd International Conference on Advances in Computing, Communication, Embedded and Secure Systems (ACCESS)","volume":"3 7","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"120861474","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 : 2021-09-02DOI: 10.1109/ACCESS51619.2021.9563299
K. Manoj, A. S. Dhas
A major increase in brain tumor has been seen in recent years, and it is in the tenth position. It is sever type of cancer and influences in people of all ages. Hence, if diagnosed well at the initial stage, it will turn out to be one of the most curable types of tumors. The computer aided analysis of MRI is performed to diagnosis the tumor through the process of classifying and segmenting. From the previous years of study, the research areas are mainly concentrated on machine and deep learning for brain malignancy prediction and treatment. The two dimensional MRI images helps to detect and classify the brain cancer precisely and efficiently. Usually the MRI images are two dimensional and not give sufficient knowledge regarding the structure and exact size of the tumor can be removed, and the detection procedure has become more complex. Since two-dimensional images never offer the actual feeling of exactly how a tumor looks, diagnosis includes 3D tumor reconstruction, planning for surgery and biological studies. The survival rate shows gives us an exact picture of the number of patients who have survived after the tumor is identified. The 5-year and 10 year survival rate is approximately 36 percent and 31 percent respectively for persons with a cancerous brain or CNS tumor. For increasing the survival rate of brain tumor, 3D image reconstruction can be used and it is one of the best attractive features in virtual reality, especially because of its application in medical image processing.
{"title":"Review On Brain Tumor Malignancy Prediction By 3D Reconstruction","authors":"K. Manoj, A. S. Dhas","doi":"10.1109/ACCESS51619.2021.9563299","DOIUrl":"https://doi.org/10.1109/ACCESS51619.2021.9563299","url":null,"abstract":"A major increase in brain tumor has been seen in recent years, and it is in the tenth position. It is sever type of cancer and influences in people of all ages. Hence, if diagnosed well at the initial stage, it will turn out to be one of the most curable types of tumors. The computer aided analysis of MRI is performed to diagnosis the tumor through the process of classifying and segmenting. From the previous years of study, the research areas are mainly concentrated on machine and deep learning for brain malignancy prediction and treatment. The two dimensional MRI images helps to detect and classify the brain cancer precisely and efficiently. Usually the MRI images are two dimensional and not give sufficient knowledge regarding the structure and exact size of the tumor can be removed, and the detection procedure has become more complex. Since two-dimensional images never offer the actual feeling of exactly how a tumor looks, diagnosis includes 3D tumor reconstruction, planning for surgery and biological studies. The survival rate shows gives us an exact picture of the number of patients who have survived after the tumor is identified. The 5-year and 10 year survival rate is approximately 36 percent and 31 percent respectively for persons with a cancerous brain or CNS tumor. For increasing the survival rate of brain tumor, 3D image reconstruction can be used and it is one of the best attractive features in virtual reality, especially because of its application in medical image processing.","PeriodicalId":409648,"journal":{"name":"2021 2nd International Conference on Advances in Computing, Communication, Embedded and Secure Systems (ACCESS)","volume":"37 40 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125708201","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 : 2021-09-02DOI: 10.1109/ACCESS51619.2021.9563279
Akshay Azhikoden, Anuroop P. Das, K. Chandran, Mohan V S Syam, D. S. Divya, A. A. Kadar
Human population is increasingly getting contained in closed office or residential spaces with minimal natural ventilation. Air pollutants that get generated internally or coming in from outside tends to get trapped inside these closed spaces long enough to pose health hazards. Also with these spaces being artificially controlled using power hungry HVAC equipment leading to heavy load on the public electric infrastructure along with inflated electric bills eating into operating budgets. This paper presents a LoRaWAN based smart room monitoring solution to monitor vital room environmental parameters along with a automatic load control scheme that could potentially save up to 30 percent in energy cost along with proportional reduction in greenhouse gas emissions.
{"title":"LoRaWAN Based Smart Room Monitor","authors":"Akshay Azhikoden, Anuroop P. Das, K. Chandran, Mohan V S Syam, D. S. Divya, A. A. Kadar","doi":"10.1109/ACCESS51619.2021.9563279","DOIUrl":"https://doi.org/10.1109/ACCESS51619.2021.9563279","url":null,"abstract":"Human population is increasingly getting contained in closed office or residential spaces with minimal natural ventilation. Air pollutants that get generated internally or coming in from outside tends to get trapped inside these closed spaces long enough to pose health hazards. Also with these spaces being artificially controlled using power hungry HVAC equipment leading to heavy load on the public electric infrastructure along with inflated electric bills eating into operating budgets. This paper presents a LoRaWAN based smart room monitoring solution to monitor vital room environmental parameters along with a automatic load control scheme that could potentially save up to 30 percent in energy cost along with proportional reduction in greenhouse gas emissions.","PeriodicalId":409648,"journal":{"name":"2021 2nd International Conference on Advances in Computing, Communication, Embedded and Secure Systems (ACCESS)","volume":"68 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130696138","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 : 2021-09-02DOI: 10.1109/ACCESS51619.2021.9563319
Benjamin Koshy Alex, Chandran G Jyothish, Julian Thomas Prasad, Samanta Cottackal
A real time 3D capture system for an object is designed and developed in this paper. In a fast-moving world where whatever that is seen around is three dimensional, most of the techniques which are devised to study an object gives a two-dimensional output. In other words, a clear idea about such spatial objects is obtained rarely. The existing techniques which are available, moreover gives an output which is not real-time, that is when an object is subjected for study, the intended results cannot be obtained in a parallel manner. In real time 3D capture, the focus is on studying three- dimensional objects in real-time. This proposal solely accentuates human perception to a different perspective. The primary goal of the technique that is inculcated in this project is to recreate the purpose of the human eye through a capturing system. This is served by the technique of stereovision. This involves distance calculation to an object from point of observation. As a result, the depth is calculated in 3D space. The system consists of two camera modules synchronized together which are further connected to a processor(computer) and output is fed to a display unit. This system brings off the task of quality analysis. The locus point in quality analysis is to find the physical deformities. Predominantly, this system is devised to equip and encourage the small scale production units or factories to aid them in an inclusive way.
{"title":"Real Time 3D Capturing System for Quality Analysis","authors":"Benjamin Koshy Alex, Chandran G Jyothish, Julian Thomas Prasad, Samanta Cottackal","doi":"10.1109/ACCESS51619.2021.9563319","DOIUrl":"https://doi.org/10.1109/ACCESS51619.2021.9563319","url":null,"abstract":"A real time 3D capture system for an object is designed and developed in this paper. In a fast-moving world where whatever that is seen around is three dimensional, most of the techniques which are devised to study an object gives a two-dimensional output. In other words, a clear idea about such spatial objects is obtained rarely. The existing techniques which are available, moreover gives an output which is not real-time, that is when an object is subjected for study, the intended results cannot be obtained in a parallel manner. In real time 3D capture, the focus is on studying three- dimensional objects in real-time. This proposal solely accentuates human perception to a different perspective. The primary goal of the technique that is inculcated in this project is to recreate the purpose of the human eye through a capturing system. This is served by the technique of stereovision. This involves distance calculation to an object from point of observation. As a result, the depth is calculated in 3D space. The system consists of two camera modules synchronized together which are further connected to a processor(computer) and output is fed to a display unit. This system brings off the task of quality analysis. The locus point in quality analysis is to find the physical deformities. Predominantly, this system is devised to equip and encourage the small scale production units or factories to aid them in an inclusive way.","PeriodicalId":409648,"journal":{"name":"2021 2nd International Conference on Advances in Computing, Communication, Embedded and Secure Systems (ACCESS)","volume":"28 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"134284772","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 : 2021-09-02DOI: 10.1109/ACCESS51619.2021.9563340
S. Saurav, S. Benedict
Energy-aware scheduling algorithms are emerging as important components in economic-conscious heterogeneous computing systems such as IoT-enabled edge, fog, or cloud environments. Most of the IoT applications utilize cloud infrastructure to process information or perform analytics. The design of energy-aware scheduling algorithms for cloud infrastructures is especially challenging given the highly variable state changes of processors and virtual machines, and the changing available compute nodes typically encountered in such infrastructures. In this paper, we have reviewed the merits and demerits of the available energy-efficient scheduling techniques for cloud environments which could be applied for IoT applications. The paper, in addition, discussed a few design challenges for creating an energy-efficient scheduler.
{"title":"Energy Aware Scheduling Algorithms for Cloud Environments -A Survey","authors":"S. Saurav, S. Benedict","doi":"10.1109/ACCESS51619.2021.9563340","DOIUrl":"https://doi.org/10.1109/ACCESS51619.2021.9563340","url":null,"abstract":"Energy-aware scheduling algorithms are emerging as important components in economic-conscious heterogeneous computing systems such as IoT-enabled edge, fog, or cloud environments. Most of the IoT applications utilize cloud infrastructure to process information or perform analytics. The design of energy-aware scheduling algorithms for cloud infrastructures is especially challenging given the highly variable state changes of processors and virtual machines, and the changing available compute nodes typically encountered in such infrastructures. In this paper, we have reviewed the merits and demerits of the available energy-efficient scheduling techniques for cloud environments which could be applied for IoT applications. The paper, in addition, discussed a few design challenges for creating an energy-efficient scheduler.","PeriodicalId":409648,"journal":{"name":"2021 2nd International Conference on Advances in Computing, Communication, Embedded and Secure Systems (ACCESS)","volume":"64 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129444055","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 : 2021-09-02DOI: 10.1109/ACCESS51619.2021.9563326
Laxmi Yadav, R. K. Yadav, Vinay Kumar
Authentication of a user's identity is becoming a tough task for a system in today's era in which digital authentication becoming mandatory to satisfy the security of a system. Recognition failure of user's identity is one of the big concerns. This paper introduces an efficient mechanism to carry out the recognition of facial features in order to satisfy the authentication of a system. Earlier researches in this field have common constraints such as false acceptance and false rejection rate. The proposed method implements over video data on which deep reinforcement learning and K-nearest neighbors (KNN) have been applied to perform detection and recognize facial data accurately. The challenging task of this work is to correctly recognize the facial data under various disturbance and unprecedented noisy circumstances including bad illumination, blurring, inappropriate poses, angle, etc. The main objective of the model is to achieve a high recognition rate of facial data under different unwanted noise and attacks. Reinforcement learning is used to count the number of people in the proposed system. This concept of the KNN algorithm is used for classification based on Euclidean distance to achieve better recognition results. The average rate of accuracy for recognition is found to be 96.40%. The proposed model can be applied to an investigation into digital forensics.
{"title":"An Efficient Approach towards Face Recognition using Deep Reinforcement Learning, Viola Jones and K-nearest neighbor","authors":"Laxmi Yadav, R. K. Yadav, Vinay Kumar","doi":"10.1109/ACCESS51619.2021.9563326","DOIUrl":"https://doi.org/10.1109/ACCESS51619.2021.9563326","url":null,"abstract":"Authentication of a user's identity is becoming a tough task for a system in today's era in which digital authentication becoming mandatory to satisfy the security of a system. Recognition failure of user's identity is one of the big concerns. This paper introduces an efficient mechanism to carry out the recognition of facial features in order to satisfy the authentication of a system. Earlier researches in this field have common constraints such as false acceptance and false rejection rate. The proposed method implements over video data on which deep reinforcement learning and K-nearest neighbors (KNN) have been applied to perform detection and recognize facial data accurately. The challenging task of this work is to correctly recognize the facial data under various disturbance and unprecedented noisy circumstances including bad illumination, blurring, inappropriate poses, angle, etc. The main objective of the model is to achieve a high recognition rate of facial data under different unwanted noise and attacks. Reinforcement learning is used to count the number of people in the proposed system. This concept of the KNN algorithm is used for classification based on Euclidean distance to achieve better recognition results. The average rate of accuracy for recognition is found to be 96.40%. The proposed model can be applied to an investigation into digital forensics.","PeriodicalId":409648,"journal":{"name":"2021 2nd International Conference on Advances in Computing, Communication, Embedded and Secure Systems (ACCESS)","volume":"116 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"113999833","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 : 2021-09-02DOI: 10.1109/ACCESS51619.2021.9563334
Ujjwal Singh, D. S. Divya, A. Kadar
Virtual Reality (VR) is an innovative technology that provides a unique platform to create an imaginary world in the real world that can be addressed by physical presence. This enveloping technology develops a computer-based imaginary environment platform and the advancements give unique solutions. The project aims to create a Virtual Reality training module for Maithari, the first stage of kalarippayattu (Kalari) which includes postures, exercises and sequences practised during the Kalari with the help of Unity Engine. There are four stages of kalarippayattu i.e. First stage- Meithari (Physical exercises), Second stage- Kolthari (Wooden Weapon Training), Third stage-Angathari (Metal Weapon Training), Fourth stage- Verumkai (Bare-Hand Technique). Here we are focusing on the first stage, Meithari. It builds the basics steps for the entire training and exercises that include jumps, leaps, kicks, circular sequences, leaps, postures, leg swings, and various movement technique. The physical training performed at this stage prepares the practitioner by developing several attributes such as strength, stamina, flexibility, agility, and speed. The aim is also to develop patience, self-control, focus, and awareness. In this project, we are using the Unity Engine as a tool to develop the training module as it offers support for multiple platforms, including PC, mobile and major consoles. With Unity, we just need to have a one-click process to have multiple platforms with ease.
{"title":"Development of Virtual Reality Training module for Maithari of Martial Art Kalari","authors":"Ujjwal Singh, D. S. Divya, A. Kadar","doi":"10.1109/ACCESS51619.2021.9563334","DOIUrl":"https://doi.org/10.1109/ACCESS51619.2021.9563334","url":null,"abstract":"Virtual Reality (VR) is an innovative technology that provides a unique platform to create an imaginary world in the real world that can be addressed by physical presence. This enveloping technology develops a computer-based imaginary environment platform and the advancements give unique solutions. The project aims to create a Virtual Reality training module for Maithari, the first stage of kalarippayattu (Kalari) which includes postures, exercises and sequences practised during the Kalari with the help of Unity Engine. There are four stages of kalarippayattu i.e. First stage- Meithari (Physical exercises), Second stage- Kolthari (Wooden Weapon Training), Third stage-Angathari (Metal Weapon Training), Fourth stage- Verumkai (Bare-Hand Technique). Here we are focusing on the first stage, Meithari. It builds the basics steps for the entire training and exercises that include jumps, leaps, kicks, circular sequences, leaps, postures, leg swings, and various movement technique. The physical training performed at this stage prepares the practitioner by developing several attributes such as strength, stamina, flexibility, agility, and speed. The aim is also to develop patience, self-control, focus, and awareness. In this project, we are using the Unity Engine as a tool to develop the training module as it offers support for multiple platforms, including PC, mobile and major consoles. With Unity, we just need to have a one-click process to have multiple platforms with ease.","PeriodicalId":409648,"journal":{"name":"2021 2nd International Conference on Advances in Computing, Communication, Embedded and Secure Systems (ACCESS)","volume":"61 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"114843545","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 : 2021-09-02DOI: 10.1109/ACCESS51619.2021.9563337
V. A. Binson, M. Subramoniam, G. Ragesh, Ajay Kumar
This pilot study presents the application of the ensemble learning method, AdaBoost in the detection of early-stage lung cancers. To detect the presence and variations of volatile organic compound biomarkers in the expelled breath, an electronic nose system with metal oxide gas sensors is developed. The system is tested in ten lung cancer patients and fifteen healthy controls to differentiate the breath samples. The system attained an acceptable accuracy, sensitivity, and specificity of 76 %, 70 %, and 80 % respectively with an independent component analysis (ICA) dimensionality reduction technique. The system should be further studied with adequate number of early stage cancers to get a concluding remark about the performance of the system in the detection of early-stage lung cancers.
{"title":"Early Detection of Lung Cancer Through Breath Analysis Using AdaBoost Ensemble Learning Method","authors":"V. A. Binson, M. Subramoniam, G. Ragesh, Ajay Kumar","doi":"10.1109/ACCESS51619.2021.9563337","DOIUrl":"https://doi.org/10.1109/ACCESS51619.2021.9563337","url":null,"abstract":"This pilot study presents the application of the ensemble learning method, AdaBoost in the detection of early-stage lung cancers. To detect the presence and variations of volatile organic compound biomarkers in the expelled breath, an electronic nose system with metal oxide gas sensors is developed. The system is tested in ten lung cancer patients and fifteen healthy controls to differentiate the breath samples. The system attained an acceptable accuracy, sensitivity, and specificity of 76 %, 70 %, and 80 % respectively with an independent component analysis (ICA) dimensionality reduction technique. The system should be further studied with adequate number of early stage cancers to get a concluding remark about the performance of the system in the detection of early-stage lung cancers.","PeriodicalId":409648,"journal":{"name":"2021 2nd International Conference on Advances in Computing, Communication, Embedded and Secure Systems (ACCESS)","volume":"39 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133340592","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 : 2021-09-02DOI: 10.1109/ACCESS51619.2021.9563297
Ishita Karna, Aniket Madam, Chinmay Deokule, Rahul B. Adhao, V. Pachghare
Intrusion Detection systems play a crucial role in maintaining network security. It keeps track of network traffic for anomalous activities and detects any vulnerabilities in the network. It is not a trivial task to build one due to the high number of features in the dataset, which increases the computational overhead on the system. It is necessary that we select only the relevant features from the dataset to ensure that the model thus built provides high accuracy in low computational time. This paper works on different filter-based feature selection techniques to lower the complexity of intrusion detection systems while preserving the performance of the system. The use of feature selection techniques followed by ensemble learning provides an optimal subset of features. The proposed method attempts to handle the imbalance of classes in CIC-IDS2017 and NSL-KDD datasets by separately classifying the minority and majority classes. The model's performance is explored in terms of precision, accuracy, and F1 score, that has been observed to be superior to existing works in the field of intrusion detection.
{"title":"Ensemble-Based Filter Feature Selection Technique for Building Flow-Based IDS","authors":"Ishita Karna, Aniket Madam, Chinmay Deokule, Rahul B. Adhao, V. Pachghare","doi":"10.1109/ACCESS51619.2021.9563297","DOIUrl":"https://doi.org/10.1109/ACCESS51619.2021.9563297","url":null,"abstract":"Intrusion Detection systems play a crucial role in maintaining network security. It keeps track of network traffic for anomalous activities and detects any vulnerabilities in the network. It is not a trivial task to build one due to the high number of features in the dataset, which increases the computational overhead on the system. It is necessary that we select only the relevant features from the dataset to ensure that the model thus built provides high accuracy in low computational time. This paper works on different filter-based feature selection techniques to lower the complexity of intrusion detection systems while preserving the performance of the system. The use of feature selection techniques followed by ensemble learning provides an optimal subset of features. The proposed method attempts to handle the imbalance of classes in CIC-IDS2017 and NSL-KDD datasets by separately classifying the minority and majority classes. The model's performance is explored in terms of precision, accuracy, and F1 score, that has been observed to be superior to existing works in the field of intrusion detection.","PeriodicalId":409648,"journal":{"name":"2021 2nd International Conference on Advances in Computing, Communication, Embedded and Secure Systems (ACCESS)","volume":"29 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2021-09-02","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129313450","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：