Pub Date : 2022-11-19DOI: 10.1109/ASSIC55218.2022.10088389
Aritra Nandi, Shivam Yadav, Yashasvi Jaiswal
Crop disease is a serious problem in the agricultural sector. To prevent crop disease we have to detect the disease at an early stage. Various technologies are emerging these days to determine specific diseases in crops. Deep Learning is one of the best approaches to detecting crop disease. This research paper includes a deep learning framework to classify healthy and diseased crops. For image recognition, ResNet was built using Keras applications. It is a deep residual learning approach that was used, as its framework is easy for training networks. Our used dataset consists of 87,354 images of 14 different sets of crops including both healthy and diseased images. The dataset was collected using a cloud-based architecture with AR. The model architecture that was trained gives us an accuracy of 99.53% in finding the diseased crop images successfully. The high success rate of this model makes it very useful and most effective in real-life applications. The further expansion of this idea “Crop disease diagnosis using deep learning” will help contribute to the operation in real cultivation conditions.
{"title":"Crop disease recognition and diagnosis using Residual Neural Network","authors":"Aritra Nandi, Shivam Yadav, Yashasvi Jaiswal","doi":"10.1109/ASSIC55218.2022.10088389","DOIUrl":"https://doi.org/10.1109/ASSIC55218.2022.10088389","url":null,"abstract":"Crop disease is a serious problem in the agricultural sector. To prevent crop disease we have to detect the disease at an early stage. Various technologies are emerging these days to determine specific diseases in crops. Deep Learning is one of the best approaches to detecting crop disease. This research paper includes a deep learning framework to classify healthy and diseased crops. For image recognition, ResNet was built using Keras applications. It is a deep residual learning approach that was used, as its framework is easy for training networks. Our used dataset consists of 87,354 images of 14 different sets of crops including both healthy and diseased images. The dataset was collected using a cloud-based architecture with AR. The model architecture that was trained gives us an accuracy of 99.53% in finding the diseased crop images successfully. The high success rate of this model makes it very useful and most effective in real-life applications. The further expansion of this idea “Crop disease diagnosis using deep learning” will help contribute to the operation in real cultivation conditions.","PeriodicalId":441406,"journal":{"name":"2022 International Conference on Advancements in Smart, Secure and Intelligent Computing (ASSIC)","volume":"133 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115254736","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 : 2022-11-19DOI: 10.1109/ASSIC55218.2022.10088382
P. Ghadekar, Chirag Vaswani, Dhruva Khanwelkar, Harsh More, Nirvisha Soni, Juhi Rajani
Analysing equity instruments has become more and more important with the stock markets being more accessible. The 2 popular ways include technical analysis and fundamental analysis. While technical analysis involves studying patterns or trends over a period of time, fundamental analysis takes a more logical approach by valuing the instrument according to its underlying fundamentals such as the reported profits, current debt, etc., and is closer to the balance sheet. Fundamental Analysis puts great emphasis on quantifying the strength of the instrument using the measures that directly represent how the organisation that issues these instruments is performing. This paper aims to investigate how a high-capacity model such as a Deep Neural Network, specifically the Entity Embedding Neural Network maps fundamental and price data to predict a future price that best explains a security. Results show that the proposed approach has an R2 score of 0.9019, accuracy of 93.42%, and MSE loss of 0.047 which outperforms the results obtained by some of the other ways of modeling this data.
{"title":"Fundamental Analysis of Equity Instruments Using an Entity Embedding Neural Network","authors":"P. Ghadekar, Chirag Vaswani, Dhruva Khanwelkar, Harsh More, Nirvisha Soni, Juhi Rajani","doi":"10.1109/ASSIC55218.2022.10088382","DOIUrl":"https://doi.org/10.1109/ASSIC55218.2022.10088382","url":null,"abstract":"Analysing equity instruments has become more and more important with the stock markets being more accessible. The 2 popular ways include technical analysis and fundamental analysis. While technical analysis involves studying patterns or trends over a period of time, fundamental analysis takes a more logical approach by valuing the instrument according to its underlying fundamentals such as the reported profits, current debt, etc., and is closer to the balance sheet. Fundamental Analysis puts great emphasis on quantifying the strength of the instrument using the measures that directly represent how the organisation that issues these instruments is performing. This paper aims to investigate how a high-capacity model such as a Deep Neural Network, specifically the Entity Embedding Neural Network maps fundamental and price data to predict a future price that best explains a security. Results show that the proposed approach has an R2 score of 0.9019, accuracy of 93.42%, and MSE loss of 0.047 which outperforms the results obtained by some of the other ways of modeling this data.","PeriodicalId":441406,"journal":{"name":"2022 International Conference on Advancements in Smart, Secure and Intelligent Computing (ASSIC)","volume":"50 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125069883","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 : 2022-11-19DOI: 10.1109/ASSIC55218.2022.10088399
P. Chinnasamy, N. Kumaresan, R. Selvaraj, S. Dhanasekaran, K. Ramprathap, Sruthi Boddu
Phishing is an illegal method which involves user's personal information at high risk. Phishing websites prey individuals, the cloud storage hosting companies and government agencies. Though there are various anti-phishing approaches like hardware as they are not cost effective and they don't choose these approaches. To overcome this, many software-based techniques are used. Zero-day phishing problem cannot be omitted with the existing models. To prevail over these issues and detect phishing attack an approach using heuristic methodology has been proposed. We classify whether a link is phishing or non-phishing based on the input features we take like Web Traffic and Uniform Resource Locator (URL). The proposed methodology is executed by retrieving datasets from phishing cases and Machine Learning model using algorithms like Random Forest, SVM, Genetic.
{"title":"An Efficient Phishing Attack Detection using Machine Learning Algorithms","authors":"P. Chinnasamy, N. Kumaresan, R. Selvaraj, S. Dhanasekaran, K. Ramprathap, Sruthi Boddu","doi":"10.1109/ASSIC55218.2022.10088399","DOIUrl":"https://doi.org/10.1109/ASSIC55218.2022.10088399","url":null,"abstract":"Phishing is an illegal method which involves user's personal information at high risk. Phishing websites prey individuals, the cloud storage hosting companies and government agencies. Though there are various anti-phishing approaches like hardware as they are not cost effective and they don't choose these approaches. To overcome this, many software-based techniques are used. Zero-day phishing problem cannot be omitted with the existing models. To prevail over these issues and detect phishing attack an approach using heuristic methodology has been proposed. We classify whether a link is phishing or non-phishing based on the input features we take like Web Traffic and Uniform Resource Locator (URL). The proposed methodology is executed by retrieving datasets from phishing cases and Machine Learning model using algorithms like Random Forest, SVM, Genetic.","PeriodicalId":441406,"journal":{"name":"2022 International Conference on Advancements in Smart, Secure and Intelligent Computing (ASSIC)","volume":"2 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116476878","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}
More than half of the population of the world owns a smartphone, and many individuals are beginning to utilize smartwatches. Many real-world smartphones or smartwatch-based sensing applications are becoming available. To gain a better understanding of human behaviour, these applications recognize human activities using accelerometers and gyroscope sensors built into smartphones. In this research, we looked at the accelerometer and gyroscopes on both the smartphone and the smartwatch, as well as their combinations, to see which combination performs best for the underlying algorithms. This work demonstrates how to automatically extract discriminative features for activity recognition using Long Short Term Memory (LSTM) method, a deep learning approach. The results reported in this article show that using a smartwatch accelerometer and/or a combination of any two or four sensors can produce good results. However, we will endeavour to improve the accuracy of activity detection using raw sensor data.
{"title":"Applications of Deep Learning for Improved Recognition from Some High-Level Human Activities Using Sensors Data","authors":"Bhavantik Gondaliya, Anil Kumar Agrawal, Ankit Chouksey","doi":"10.1109/ASSIC55218.2022.10088361","DOIUrl":"https://doi.org/10.1109/ASSIC55218.2022.10088361","url":null,"abstract":"More than half of the population of the world owns a smartphone, and many individuals are beginning to utilize smartwatches. Many real-world smartphones or smartwatch-based sensing applications are becoming available. To gain a better understanding of human behaviour, these applications recognize human activities using accelerometers and gyroscope sensors built into smartphones. In this research, we looked at the accelerometer and gyroscopes on both the smartphone and the smartwatch, as well as their combinations, to see which combination performs best for the underlying algorithms. This work demonstrates how to automatically extract discriminative features for activity recognition using Long Short Term Memory (LSTM) method, a deep learning approach. The results reported in this article show that using a smartwatch accelerometer and/or a combination of any two or four sensors can produce good results. However, we will endeavour to improve the accuracy of activity detection using raw sensor data.","PeriodicalId":441406,"journal":{"name":"2022 International Conference on Advancements in Smart, Secure and Intelligent Computing (ASSIC)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131144908","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 : 2022-11-19DOI: 10.1109/ASSIC55218.2022.10088320
A. Sangeetha, V. Hema, S. Navya
Information overflow is a concern today due to the increased number of information sources available on the internet. One of the most obvious types of cybercrime has emerged: fake news, commonly known as a hoax. Hoax news spreads harm to the social community by instilling hatred in both individuals and groups. The purpose of this paper is to distinguish between hoaxes and true news using the Extreme Gradient Boosting (XGBoost) method. The dataset used is kaggle train datasets The study included 20799 news records, including individual false and actual news stories, which were split into 80 percent training data and 20 percent test data. According to the findings of this study, the machine learning model constructed using XGBoost has an accuracy of 91 percent, a precision of 90 percent, and a recall of 80 percent. As a result, we created a webapp that uses the Flask API to determine whether or not news is bogus.
{"title":"Fake News Detection Using Machine Learning Algorithm","authors":"A. Sangeetha, V. Hema, S. Navya","doi":"10.1109/ASSIC55218.2022.10088320","DOIUrl":"https://doi.org/10.1109/ASSIC55218.2022.10088320","url":null,"abstract":"Information overflow is a concern today due to the increased number of information sources available on the internet. One of the most obvious types of cybercrime has emerged: fake news, commonly known as a hoax. Hoax news spreads harm to the social community by instilling hatred in both individuals and groups. The purpose of this paper is to distinguish between hoaxes and true news using the Extreme Gradient Boosting (XGBoost) method. The dataset used is kaggle train datasets The study included 20799 news records, including individual false and actual news stories, which were split into 80 percent training data and 20 percent test data. According to the findings of this study, the machine learning model constructed using XGBoost has an accuracy of 91 percent, a precision of 90 percent, and a recall of 80 percent. As a result, we created a webapp that uses the Flask API to determine whether or not news is bogus.","PeriodicalId":441406,"journal":{"name":"2022 International Conference on Advancements in Smart, Secure and Intelligent Computing (ASSIC)","volume":"73 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"116480193","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 : 2022-11-19DOI: 10.1109/ASSIC55218.2022.10088371
Syed Naushad Ali Hashmi, Vinod Kumar Singh, Rajesh Kumar Dwivedi, R. S. Pathak, R. Tiwari
Recently the thinner and lighter devices are preferred by the consumers because of having multifunctional capabilities. That's why dual-band antennas for WLAN and Hiper LAN applications are developed to fulfill the demand of consumers. A new compact antenna operating in 2 to 11 GHz frequency range has been presented in this paper. The design is planned for the Hiper LAN and WLAN applications. The simulations of the antenna in free space produced desired results in terms of gain, radiation efficiency, and bandwidth; a maximum directivity of 4.372 dB was achieved. A slotted circular patch with line feed is made on substrate having partial ground plane of copper. The presented antenna gives dual band width of 31.84% and 87.85%.
{"title":"Tri Pentagon Slotted Antenna Using Jeans Material WLAN/Hiper LAN Application","authors":"Syed Naushad Ali Hashmi, Vinod Kumar Singh, Rajesh Kumar Dwivedi, R. S. Pathak, R. Tiwari","doi":"10.1109/ASSIC55218.2022.10088371","DOIUrl":"https://doi.org/10.1109/ASSIC55218.2022.10088371","url":null,"abstract":"Recently the thinner and lighter devices are preferred by the consumers because of having multifunctional capabilities. That's why dual-band antennas for WLAN and Hiper LAN applications are developed to fulfill the demand of consumers. A new compact antenna operating in 2 to 11 GHz frequency range has been presented in this paper. The design is planned for the Hiper LAN and WLAN applications. The simulations of the antenna in free space produced desired results in terms of gain, radiation efficiency, and bandwidth; a maximum directivity of 4.372 dB was achieved. A slotted circular patch with line feed is made on substrate having partial ground plane of copper. The presented antenna gives dual band width of 31.84% and 87.85%.","PeriodicalId":441406,"journal":{"name":"2022 International Conference on Advancements in Smart, Secure and Intelligent Computing (ASSIC)","volume":"15 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133181153","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 : 2022-11-19DOI: 10.1109/ASSIC55218.2022.10088368
Dipti Dash, Aleena Mishra
Marketing is crucial for the company's growth and long-term viability. Marketers can aid in the development of a company's brand, customer engagement, revenue growth and sales. Knowing and identifying clients' needs is one of the most difficult tasks for the marketers. Marketers can begin a focused marketing strategy that is suited to specific demands by understanding the customer. Data science can be used to do market segmentation if the customer data is accessible. We are doing a credit card segmentation using New York City bank data set. By doing analysis on this dataset, we can know about the behavior of credit card holders and can lunch an effective market campaign which would be more focused on the targeted customer. This customer-centric campaign will help to reduce the overall marketing cost, and this will boost in the number of credit card holders. In the evolving world of technology, Fintech industry is growing very fast. We are doing behavioral segmentation process to make clusters of customers. We will be using Autoencoders and then perform k-means clustering, PCA for visualization.
{"title":"Credit Card Holders Segmentation Using K-mean Clustering with Autoencoder","authors":"Dipti Dash, Aleena Mishra","doi":"10.1109/ASSIC55218.2022.10088368","DOIUrl":"https://doi.org/10.1109/ASSIC55218.2022.10088368","url":null,"abstract":"Marketing is crucial for the company's growth and long-term viability. Marketers can aid in the development of a company's brand, customer engagement, revenue growth and sales. Knowing and identifying clients' needs is one of the most difficult tasks for the marketers. Marketers can begin a focused marketing strategy that is suited to specific demands by understanding the customer. Data science can be used to do market segmentation if the customer data is accessible. We are doing a credit card segmentation using New York City bank data set. By doing analysis on this dataset, we can know about the behavior of credit card holders and can lunch an effective market campaign which would be more focused on the targeted customer. This customer-centric campaign will help to reduce the overall marketing cost, and this will boost in the number of credit card holders. In the evolving world of technology, Fintech industry is growing very fast. We are doing behavioral segmentation process to make clusters of customers. We will be using Autoencoders and then perform k-means clustering, PCA for visualization.","PeriodicalId":441406,"journal":{"name":"2022 International Conference on Advancements in Smart, Secure and Intelligent Computing (ASSIC)","volume":"43 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"130564426","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 : 2022-11-19DOI: 10.1109/ASSIC55218.2022.10088352
K. Swetha, E. N. V. Kumari, A. Kiran, Keerthana Sree Arrola
AD is a neurological illness. It ranks as the sixth most common reason for both morbidity and mortality. Alzheimer's disease can progress through three stages: mild, moderate, and severe. A timely diagnosis can assist in the provision of necessary therapy, so preventing additional harm to brain tissue. Recent research has utilised technology in an attempt to diagnose Alzheimer's disease; nevertheless, the majority of machine detection technologies are inborn. The early stages of Alzheimer's disease can be diagnosed, but it is not possible to anticipate the progression of the disease. Prediction is only possible before dementia sets in. Deep Learning (DL) has the potential to detect Alzheimer's disease in its early stages. In this article, we use two different kinds of data to predict disease categories: csv data that includes cognitive task parameters like SES, MMSE, CDR, eTIV, nWBV, ASF, delay, heredity, MOCA, SAGE, CDT; and basic patient information like gender, age, dominant hand, Education, drowsiness, and visits. The csv data includes cognitive task parameters like SES, MMSE, CDR, eTIV Calculations are done to determine the F1 score, precision, recall, and accuracy of each technique.
{"title":"Alzheimer's disease Diagnosis from MRI using Siamese Convolutional Neural Network","authors":"K. Swetha, E. N. V. Kumari, A. Kiran, Keerthana Sree Arrola","doi":"10.1109/ASSIC55218.2022.10088352","DOIUrl":"https://doi.org/10.1109/ASSIC55218.2022.10088352","url":null,"abstract":"AD is a neurological illness. It ranks as the sixth most common reason for both morbidity and mortality. Alzheimer's disease can progress through three stages: mild, moderate, and severe. A timely diagnosis can assist in the provision of necessary therapy, so preventing additional harm to brain tissue. Recent research has utilised technology in an attempt to diagnose Alzheimer's disease; nevertheless, the majority of machine detection technologies are inborn. The early stages of Alzheimer's disease can be diagnosed, but it is not possible to anticipate the progression of the disease. Prediction is only possible before dementia sets in. Deep Learning (DL) has the potential to detect Alzheimer's disease in its early stages. In this article, we use two different kinds of data to predict disease categories: csv data that includes cognitive task parameters like SES, MMSE, CDR, eTIV, nWBV, ASF, delay, heredity, MOCA, SAGE, CDT; and basic patient information like gender, age, dominant hand, Education, drowsiness, and visits. The csv data includes cognitive task parameters like SES, MMSE, CDR, eTIV Calculations are done to determine the F1 score, precision, recall, and accuracy of each technique.","PeriodicalId":441406,"journal":{"name":"2022 International Conference on Advancements in Smart, Secure and Intelligent Computing (ASSIC)","volume":"145 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124649074","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}
With the rise of the digital era, the massive shift towards everything being digital is evident. People have a plethora of digital data points to denote them in digital world. Digital personal data points, also known as Digital Identity, contains Personally Identifiable Information (PII) which is private to every individual. Managing the cycle of private, high security data from its creation till verification is a massive undertaking. As there are instances where personal data are breached, centralized servers fail to respond, attacks compromise the entire system, it becomes difficult to rely on such systems. Blockchain is a propitious technology that is immutable, decentralized, tamper-proof, highly secure and easy to use. This paper proposes to use Blockchain and leverage its advantages to establish a self-sovereign digital identity management system, collision resistant encrypted document, which ensures the integrity of document issued. It presents DigiBlock as a powerful solution to the current scenario with a robust role based permission system leveraging Distributed Ledger Technology (DLT), which provides advantage over centralized system with its exceptional sovereignty, storage-control, cost-free, security, privacy, transparency and portability features.
{"title":"DigiBlock: Digital Self-sovereign Identity on Distributed Ledger based on Blockchain","authors":"Pinaki Bhattacharjee, Chandra Prakash, Sakshi Gairola, Sai Shradha Lala, Pratyusa Mukherjee","doi":"10.1109/ASSIC55218.2022.10088367","DOIUrl":"https://doi.org/10.1109/ASSIC55218.2022.10088367","url":null,"abstract":"With the rise of the digital era, the massive shift towards everything being digital is evident. People have a plethora of digital data points to denote them in digital world. Digital personal data points, also known as Digital Identity, contains Personally Identifiable Information (PII) which is private to every individual. Managing the cycle of private, high security data from its creation till verification is a massive undertaking. As there are instances where personal data are breached, centralized servers fail to respond, attacks compromise the entire system, it becomes difficult to rely on such systems. Blockchain is a propitious technology that is immutable, decentralized, tamper-proof, highly secure and easy to use. This paper proposes to use Blockchain and leverage its advantages to establish a self-sovereign digital identity management system, collision resistant encrypted document, which ensures the integrity of document issued. It presents DigiBlock as a powerful solution to the current scenario with a robust role based permission system leveraging Distributed Ledger Technology (DLT), which provides advantage over centralized system with its exceptional sovereignty, storage-control, cost-free, security, privacy, transparency and portability features.","PeriodicalId":441406,"journal":{"name":"2022 International Conference on Advancements in Smart, Secure and Intelligent Computing (ASSIC)","volume":"19 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"127911431","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 : 2022-11-19DOI: 10.1109/ASSIC55218.2022.10088388
Sangsaptak Pal, S. Mishra, B. P. Mishra, Santwana Sagnika, Saurabh Bilgaiyan
In current scenario Convolutional Neural Network (CNN) has gained the attention of the researchers. It is a special type of feed forward neural network used to handle large images. It has the capability of adjusting the parameters. However, it is computationally expensive as it takes more training time. So in this paper we are interested to propose a new technique which will reduce the number of training parameters of CNN as well as providing a promising accuracy. The proposed technique is validated for the detection of tuberculosis.
{"title":"Quick & Lightweight Tuberculosis Detection:A CNN Based Approach","authors":"Sangsaptak Pal, S. Mishra, B. P. Mishra, Santwana Sagnika, Saurabh Bilgaiyan","doi":"10.1109/ASSIC55218.2022.10088388","DOIUrl":"https://doi.org/10.1109/ASSIC55218.2022.10088388","url":null,"abstract":"In current scenario Convolutional Neural Network (CNN) has gained the attention of the researchers. It is a special type of feed forward neural network used to handle large images. It has the capability of adjusting the parameters. However, it is computationally expensive as it takes more training time. So in this paper we are interested to propose a new technique which will reduce the number of training parameters of CNN as well as providing a promising accuracy. The proposed technique is validated for the detection of tuberculosis.","PeriodicalId":441406,"journal":{"name":"2022 International Conference on Advancements in Smart, Secure and Intelligent Computing (ASSIC)","volume":"74 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-11-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129620363","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：