Pub Date : 2022-02-12DOI: 10.1109/ICITIIT54346.2022.9744195
R. Remya, H. Prasad, S. Hariharan, C. Gopakumar
Chromosome images are susceptible to sensor and staining noises, inhomogeneity, and blurring which prevent efficient karyotyping. In this research work, image processing methods are systematically extended for the preprocessing of chromosome images, and a novel approach for denoising and enhancing the chromosome images is proposed. The proposed approach is mathematically modeled and evaluated with subjective and objective measures. Promising results are obtained which are further substantiated with the post-classification of the segmented chromosomes from the preprocessed input image. Performance of the proposed method is quantified in terms of MSE (Mean Squared Error), PSNR (Peak Signal to Noise Ratio), SSIM (Structural Similarity Index Measure), FSIM(Features Similarity Index Measure), SAM(Spectral Angle Mapper), and SRE(Signal to Reconstruction Error ratio). An MSE of 8.164, PSNR of 39.037, SSIM of 0.9654, SAM of 81.729, SRE of 63.842, and FSIM of 0.6128 are obtained, on average for a set of 10 test images which were previously degraded with Gaussian noise and Gaussian blur. Post-classification accuracy improved from 88% to 95% as and when the proposed preprocessing is followed by the classification task.
{"title":"Chromosome Image Enhancement for Efficient Karyotyping","authors":"R. Remya, H. Prasad, S. Hariharan, C. Gopakumar","doi":"10.1109/ICITIIT54346.2022.9744195","DOIUrl":"https://doi.org/10.1109/ICITIIT54346.2022.9744195","url":null,"abstract":"Chromosome images are susceptible to sensor and staining noises, inhomogeneity, and blurring which prevent efficient karyotyping. In this research work, image processing methods are systematically extended for the preprocessing of chromosome images, and a novel approach for denoising and enhancing the chromosome images is proposed. The proposed approach is mathematically modeled and evaluated with subjective and objective measures. Promising results are obtained which are further substantiated with the post-classification of the segmented chromosomes from the preprocessed input image. Performance of the proposed method is quantified in terms of MSE (Mean Squared Error), PSNR (Peak Signal to Noise Ratio), SSIM (Structural Similarity Index Measure), FSIM(Features Similarity Index Measure), SAM(Spectral Angle Mapper), and SRE(Signal to Reconstruction Error ratio). An MSE of 8.164, PSNR of 39.037, SSIM of 0.9654, SAM of 81.729, SRE of 63.842, and FSIM of 0.6128 are obtained, on average for a set of 10 test images which were previously degraded with Gaussian noise and Gaussian blur. Post-classification accuracy improved from 88% to 95% as and when the proposed preprocessing is followed by the classification task.","PeriodicalId":184353,"journal":{"name":"2022 International Conference on Innovative Trends in Information Technology (ICITIIT)","volume":"26 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"117040807","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 : 2022-02-12DOI: 10.1109/ICITIIT54346.2022.9744156
Amreen Rashik, C. V. Priya
A biometric system takes an individual’s physiological, behavioural, or both features as input, analyses them, and determines whether or not the user is genuine. This paper proposes a user authentication, data security, and access control system based on DNA QR code, which is quick, secure, and less foreseeable, followed by a multi-modal biometric system that relies on the face and fingerprint in order to assure the safety of the cyber-physical system. The alignment-based elastic technique is used to match fingerprints. Local binary patterns (LBP) are utilized to improve facial feature extraction. The proposed idea enables the login of the user using his user ID and password to the main domain only when punched IN through the access control system.
{"title":"A Secure Application of Multi-Biometric Recognition and QR Coding System","authors":"Amreen Rashik, C. V. Priya","doi":"10.1109/ICITIIT54346.2022.9744156","DOIUrl":"https://doi.org/10.1109/ICITIIT54346.2022.9744156","url":null,"abstract":"A biometric system takes an individual’s physiological, behavioural, or both features as input, analyses them, and determines whether or not the user is genuine. This paper proposes a user authentication, data security, and access control system based on DNA QR code, which is quick, secure, and less foreseeable, followed by a multi-modal biometric system that relies on the face and fingerprint in order to assure the safety of the cyber-physical system. The alignment-based elastic technique is used to match fingerprints. Local binary patterns (LBP) are utilized to improve facial feature extraction. The proposed idea enables the login of the user using his user ID and password to the main domain only when punched IN through the access control system.","PeriodicalId":184353,"journal":{"name":"2022 International Conference on Innovative Trends in Information Technology (ICITIIT)","volume":"81 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"128719774","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 : 2022-02-12DOI: 10.1109/ICITIIT54346.2022.9744245
Anupama V, M. Elayidom
It is critical to predict students' success in topics related to high study, along with deep learning as well as its connection to educational information. Recommending student performance aids in course selection and the creation of appropriate future study plans for students. It assists teachers and supervisors in monitoring pupils in order to give assistance and combining training programmes to obtain the best outcomes, in addition to recommending student performance. One of the benefits of student recommendation will be that it eliminates authorized alerting indicators while also restricting students from being ejected due to inefficiencies. Recommendation helps students by assisting them in selecting courses and study schedules that are suited for their ability. The proposed approach made suggestions using a deep neural network by obtaining relevant information as characteristic and giving weights to it. Feed forwarding and back propagation information have been used to modify the frequency of nodes and hidden layers, and the neural network is constructed automatically utilizing many modified hidden layers. The training phase was often employed to train the system utilizing labelled information from the datasets, whereas the testing phase is being utilized to assess it. With precision, the suggested technique was developed utilizing Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE). Presented has demonstrated its performance relevance by producing best recommendation outcomes in MAE (0.593) and RMSE (0.785).
{"title":"Recommendation system using deep learning to predict suitable academic path for higher secondary students","authors":"Anupama V, M. Elayidom","doi":"10.1109/ICITIIT54346.2022.9744245","DOIUrl":"https://doi.org/10.1109/ICITIIT54346.2022.9744245","url":null,"abstract":"It is critical to predict students' success in topics related to high study, along with deep learning as well as its connection to educational information. Recommending student performance aids in course selection and the creation of appropriate future study plans for students. It assists teachers and supervisors in monitoring pupils in order to give assistance and combining training programmes to obtain the best outcomes, in addition to recommending student performance. One of the benefits of student recommendation will be that it eliminates authorized alerting indicators while also restricting students from being ejected due to inefficiencies. Recommendation helps students by assisting them in selecting courses and study schedules that are suited for their ability. The proposed approach made suggestions using a deep neural network by obtaining relevant information as characteristic and giving weights to it. Feed forwarding and back propagation information have been used to modify the frequency of nodes and hidden layers, and the neural network is constructed automatically utilizing many modified hidden layers. The training phase was often employed to train the system utilizing labelled information from the datasets, whereas the testing phase is being utilized to assess it. With precision, the suggested technique was developed utilizing Mean Absolute Error (MAE) and Root Mean Squared Error (RMSE). Presented has demonstrated its performance relevance by producing best recommendation outcomes in MAE (0.593) and RMSE (0.785).","PeriodicalId":184353,"journal":{"name":"2022 International Conference on Innovative Trends in Information Technology (ICITIIT)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"123905637","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 : 2022-02-12DOI: 10.1109/ICITIIT54346.2022.9744222
K. T, S. S, Rakshith Vuppala
Plants are an important part of everyone’s life on this planet. Today there are many species of plant are present on earth, classifying them has become a challenge because some plants look similar but are not same and there are many species of plant on earth to remember. Identifying the plant is easy for those who study about plants but for a common human, it is difficult. Thus, this research paper proposes a deep learning model to classify the plant leaf which has the capability to automatically extract features from images. The input is given to two architectures Xception and ResNet50v2 and the features extracted from these architectures are concatenated and given to fully connected network also known as transfer learning. The concatenated network gives comprehensive understanding about the dataset which would help it to perform well. The concatenated model shows an accuracy of 96.38% training accuracy and 89.36% validation accuracy on One-hundred plant species dataset and 95% training accuracy and 91.6% validation accuracy on Leafsnap dataset. The results of proposed model are compared with Xception model and ResNet50v2 model.
{"title":"Hybrid transfer learning model for identification of plant species","authors":"K. T, S. S, Rakshith Vuppala","doi":"10.1109/ICITIIT54346.2022.9744222","DOIUrl":"https://doi.org/10.1109/ICITIIT54346.2022.9744222","url":null,"abstract":"Plants are an important part of everyone’s life on this planet. Today there are many species of plant are present on earth, classifying them has become a challenge because some plants look similar but are not same and there are many species of plant on earth to remember. Identifying the plant is easy for those who study about plants but for a common human, it is difficult. Thus, this research paper proposes a deep learning model to classify the plant leaf which has the capability to automatically extract features from images. The input is given to two architectures Xception and ResNet50v2 and the features extracted from these architectures are concatenated and given to fully connected network also known as transfer learning. The concatenated network gives comprehensive understanding about the dataset which would help it to perform well. The concatenated model shows an accuracy of 96.38% training accuracy and 89.36% validation accuracy on One-hundred plant species dataset and 95% training accuracy and 91.6% validation accuracy on Leafsnap dataset. The results of proposed model are compared with Xception model and ResNet50v2 model.","PeriodicalId":184353,"journal":{"name":"2022 International Conference on Innovative Trends in Information Technology (ICITIIT)","volume":"96 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"132221365","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 : 2022-02-12DOI: 10.1109/ICITIIT54346.2022.9744214
M. Saravanan, Mannathu Gopikrishnan
This paper examines the implementation of Topological Data Analysis methods based on Persistent Homology to meet the requirements of the telecommunication industry. Persistent Homology based methods are especially useful in detecting anomalies in time series data and show good prospects of being useful in network alarm systems. Of crucial importance to this method is a metric called the Wasserstein Distance, which measures how much two Persistence Diagrams differ from one another. This metric can be formulated as a minimum weight maximum matching problem on a bipartite graph. We here solve the combinatorial optimization problem of finding the Wasserstein Distance by applying the Quantum Approximate Optimization Algorithm (QAOA) using gate-based quantum computing methods. This technique can then be applied to detect anomalies in time series datasets involving network traffic/throughput data in telecommunication systems. The methodology stands to provide a significant technological advantage to service providers who adopt this, once practical gate-based quantum computers become ubiquitous.
{"title":"The Wasserstein Distance Using QAOA: A Quantum Augmented Approach to Topological Data Analysis","authors":"M. Saravanan, Mannathu Gopikrishnan","doi":"10.1109/ICITIIT54346.2022.9744214","DOIUrl":"https://doi.org/10.1109/ICITIIT54346.2022.9744214","url":null,"abstract":"This paper examines the implementation of Topological Data Analysis methods based on Persistent Homology to meet the requirements of the telecommunication industry. Persistent Homology based methods are especially useful in detecting anomalies in time series data and show good prospects of being useful in network alarm systems. Of crucial importance to this method is a metric called the Wasserstein Distance, which measures how much two Persistence Diagrams differ from one another. This metric can be formulated as a minimum weight maximum matching problem on a bipartite graph. We here solve the combinatorial optimization problem of finding the Wasserstein Distance by applying the Quantum Approximate Optimization Algorithm (QAOA) using gate-based quantum computing methods. This technique can then be applied to detect anomalies in time series datasets involving network traffic/throughput data in telecommunication systems. The methodology stands to provide a significant technological advantage to service providers who adopt this, once practical gate-based quantum computers become ubiquitous.","PeriodicalId":184353,"journal":{"name":"2022 International Conference on Innovative Trends in Information Technology (ICITIIT)","volume":"12 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"131615109","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 : 2022-02-12DOI: 10.1109/ICITIIT54346.2022.9744142
Palaniappan M, Sowmia K R, A. S
One of the most important challenges confronting the world today is the rise in road accidents. Improper and inattentive driving is the leading cause of road accidents. This study’s main goal is to develop a non-intrusive system that can detect human fatigue and provide an early warning. Drivers who do not stop frequently when driving long distances are at risk of becoming drowsy, which they sometimes do not realise until it is too late. The driver’s drowsiness or lack of concentration is regarded to be a primary factor in such incidents. Driver sleepiness monitoring research could aid in the reduction of accidents. According to expert research, about a quarter of serious highway accidents can be attributed to sleepy drivers who need to rest, which means that sleepy drivers cause more traffic accidents than drink-driving. The technology will employ a camera to follow and monitor drivers’ eyes, and by building a Landmarks algorithm, we will be able to detect sleepiness symptoms in drivers early enough to avoid accidents. As a result, this research will assist in detecting a driver’s tiredness in advance and providing warning output in the form of alarms and pop-up windows. Furthermore, rather than being disabled automatically, the warning will be disabled manually. This will identify tiredness or fatigue and can be used to automatically slow the vehicle down.
{"title":"Real Time Fatigue Detection Using Shape Predictor 68 Face Landmarks Algorithm","authors":"Palaniappan M, Sowmia K R, A. S","doi":"10.1109/ICITIIT54346.2022.9744142","DOIUrl":"https://doi.org/10.1109/ICITIIT54346.2022.9744142","url":null,"abstract":"One of the most important challenges confronting the world today is the rise in road accidents. Improper and inattentive driving is the leading cause of road accidents. This study’s main goal is to develop a non-intrusive system that can detect human fatigue and provide an early warning. Drivers who do not stop frequently when driving long distances are at risk of becoming drowsy, which they sometimes do not realise until it is too late. The driver’s drowsiness or lack of concentration is regarded to be a primary factor in such incidents. Driver sleepiness monitoring research could aid in the reduction of accidents. According to expert research, about a quarter of serious highway accidents can be attributed to sleepy drivers who need to rest, which means that sleepy drivers cause more traffic accidents than drink-driving. The technology will employ a camera to follow and monitor drivers’ eyes, and by building a Landmarks algorithm, we will be able to detect sleepiness symptoms in drivers early enough to avoid accidents. As a result, this research will assist in detecting a driver’s tiredness in advance and providing warning output in the form of alarms and pop-up windows. Furthermore, rather than being disabled automatically, the warning will be disabled manually. This will identify tiredness or fatigue and can be used to automatically slow the vehicle down.","PeriodicalId":184353,"journal":{"name":"2022 International Conference on Innovative Trends in Information Technology (ICITIIT)","volume":"76 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"133027123","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 : 2022-02-12DOI: 10.1109/ICITIIT54346.2022.9744207
Pratham Nayak, Aprameya Dash, Suyash Chintawar, M A.
As a substitute for conventional energy sources, Solar energy is quickly becoming a popular source of renewable energy. Various entities ranging from small households and businesses to large firms and MNCs are currently making plans on investing resources in the generation of solar energy. Thus, accurate prediction of solar radiation has become a necessity in the present scenario. Due to limitations like the unavailability of proper measuring equipment and a small number of meteorological departments, accurate prediction of solar radiation is not possible in many places around the world. This paper focuses on forecasting solar radiation using machine learning techniques. Solar radiation depends upon various natural factors, which are easier to measure, and these factors can help forecast solar radiation. This paper explores the available data to identify the various factors which affect solar radiation. Based on these factors, the paper investigates the performance of different standard regression models based on solar radiation prediction. Next, multi-level statistical models are proposed, which stack multiple standard models into layers, and the R2 scores of these custom models is compared with the R2 scores of the standard models.
{"title":"Multi-Level Statistical Model for Forecasting Solar Radiation","authors":"Pratham Nayak, Aprameya Dash, Suyash Chintawar, M A.","doi":"10.1109/ICITIIT54346.2022.9744207","DOIUrl":"https://doi.org/10.1109/ICITIIT54346.2022.9744207","url":null,"abstract":"As a substitute for conventional energy sources, Solar energy is quickly becoming a popular source of renewable energy. Various entities ranging from small households and businesses to large firms and MNCs are currently making plans on investing resources in the generation of solar energy. Thus, accurate prediction of solar radiation has become a necessity in the present scenario. Due to limitations like the unavailability of proper measuring equipment and a small number of meteorological departments, accurate prediction of solar radiation is not possible in many places around the world. This paper focuses on forecasting solar radiation using machine learning techniques. Solar radiation depends upon various natural factors, which are easier to measure, and these factors can help forecast solar radiation. This paper explores the available data to identify the various factors which affect solar radiation. Based on these factors, the paper investigates the performance of different standard regression models based on solar radiation prediction. Next, multi-level statistical models are proposed, which stack multiple standard models into layers, and the R2 scores of these custom models is compared with the R2 scores of the standard models.","PeriodicalId":184353,"journal":{"name":"2022 International Conference on Innovative Trends in Information Technology (ICITIIT)","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"129065003","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 : 2022-02-12DOI: 10.1109/ICITIIT54346.2022.9744191
Jiten Sidhpura, Rudresh Veerkhare, Parshwa Shah, S. Dholay
Body Mass Index (BMI) is a measure of how healthy a person is with respect to their body weight. BMI has shown a correlation with various factors like physical health, mental health, popularity. BMI calculation often requires accurate height and weight, which would take manual work to measure. Largescale automation of BMI calculation can be utilized for analyzing various aspects of society and can be used by governments and companies to make effective decisions. Previous works have used only geometric facial features discarding other information, or a data-driven deep learning-based approach in which the amount of data becomes a bottleneck. We used the state of the art pre-trained models such as Inception-v3, VGG-Faces, VGG19, Xception and fine-tuned them on the comparatively large public dataset with discriminative learning. We used the larger Illinois DOC labeled faces dataset for training and Arrest Records, VIP_attribute for evaluation purposes.
{"title":"Face To BMI: A Deep Learning Based Approach for Computing BMI from Face","authors":"Jiten Sidhpura, Rudresh Veerkhare, Parshwa Shah, S. Dholay","doi":"10.1109/ICITIIT54346.2022.9744191","DOIUrl":"https://doi.org/10.1109/ICITIIT54346.2022.9744191","url":null,"abstract":"Body Mass Index (BMI) is a measure of how healthy a person is with respect to their body weight. BMI has shown a correlation with various factors like physical health, mental health, popularity. BMI calculation often requires accurate height and weight, which would take manual work to measure. Largescale automation of BMI calculation can be utilized for analyzing various aspects of society and can be used by governments and companies to make effective decisions. Previous works have used only geometric facial features discarding other information, or a data-driven deep learning-based approach in which the amount of data becomes a bottleneck. We used the state of the art pre-trained models such as Inception-v3, VGG-Faces, VGG19, Xception and fine-tuned them on the comparatively large public dataset with discriminative learning. We used the larger Illinois DOC labeled faces dataset for training and Arrest Records, VIP_attribute for evaluation purposes.","PeriodicalId":184353,"journal":{"name":"2022 International Conference on Innovative Trends in Information Technology (ICITIIT)","volume":"35 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126655204","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 : 2022-02-12DOI: 10.1109/ICITIIT54346.2022.9744249
Shaiju Panchikkil, V. Manikandan
Technological advancements and digitalization have made the life of humankind simple but at the same time imposing many challenges. As information started bursting across the internet, information management and security became major concerns. Recently, researchers have been focusing on a hot topic called reversible data hiding (RDH). RHD secures the data by covering it within another medium. It allows the recovery of the medium and hidden information on the receiver side without any loss. This work discloses a high capacity RDH scheme in the encrypted image with a Fibonacci transform image scrambling algorithm for data hiding and a convolutional neural network (CNN) based recovery. It follows a block-wise embedding process, embedding (n + 1) bits within a block of size 2n while n > 1. The proposed scheme is tested on the USC-SIPI image data set from the University of Southern California and has resulted in an improved embedding rate compared to the existing Arnold transform-based RDH and many other well-acknowledged RDH schemes.
{"title":"A Machine Learning based Reversible Data Hiding Scheme in Encrypted Images using Fibonacci Transform","authors":"Shaiju Panchikkil, V. Manikandan","doi":"10.1109/ICITIIT54346.2022.9744249","DOIUrl":"https://doi.org/10.1109/ICITIIT54346.2022.9744249","url":null,"abstract":"Technological advancements and digitalization have made the life of humankind simple but at the same time imposing many challenges. As information started bursting across the internet, information management and security became major concerns. Recently, researchers have been focusing on a hot topic called reversible data hiding (RDH). RHD secures the data by covering it within another medium. It allows the recovery of the medium and hidden information on the receiver side without any loss. This work discloses a high capacity RDH scheme in the encrypted image with a Fibonacci transform image scrambling algorithm for data hiding and a convolutional neural network (CNN) based recovery. It follows a block-wise embedding process, embedding (n + 1) bits within a block of size 2n while n > 1. The proposed scheme is tested on the USC-SIPI image data set from the University of Southern California and has resulted in an improved embedding rate compared to the existing Arnold transform-based RDH and many other well-acknowledged RDH schemes.","PeriodicalId":184353,"journal":{"name":"2022 International Conference on Innovative Trends in Information Technology (ICITIIT)","volume":"35 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124922186","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}
Clustering, to group similar objects, is an important problem. Recently deep learning-based methods like Deep Embedded Clustering (DEC) [6] and its semi-supervised version called Semi-supervised Deep Embedded Clustering (SDEC) [12], where partially labeled data or data with constraints is available, are shown to give promising results. Both DEC and SDEC learn a latent space where similar objects are closer and dissimilar are away. While promising results are shown, the information present in constraints or a labeled subset of the data is not fully utilized. This paper proposes to use priorities for constraints so that important constraints are given more weightage than unimportant ones. Those constraints with points that are far away, but should be clustered into a group, gets more weight than other labeled points. Similarly, those in different groups which are very close get more weightage. The appropriate loss function is used in the learning process. The proposed method is called Prioritized Semi-supervised Deep Embedded Clustering (PSDEC). The results are compared using a few standard data sets against recent and classical similar methods. PSDEC is found to achieve a better result than un-prioritized constraints.
{"title":"Prioritized Semi-supervised Deep Embedded Clustering","authors":"Pranita Saladi, Rishi Manudeep Guntupalli, Sudheer Kumar Puppala, Viswanath Pulabaigari","doi":"10.1109/ICITIIT54346.2022.9744240","DOIUrl":"https://doi.org/10.1109/ICITIIT54346.2022.9744240","url":null,"abstract":"Clustering, to group similar objects, is an important problem. Recently deep learning-based methods like Deep Embedded Clustering (DEC) [6] and its semi-supervised version called Semi-supervised Deep Embedded Clustering (SDEC) [12], where partially labeled data or data with constraints is available, are shown to give promising results. Both DEC and SDEC learn a latent space where similar objects are closer and dissimilar are away. While promising results are shown, the information present in constraints or a labeled subset of the data is not fully utilized. This paper proposes to use priorities for constraints so that important constraints are given more weightage than unimportant ones. Those constraints with points that are far away, but should be clustered into a group, gets more weight than other labeled points. Similarly, those in different groups which are very close get more weightage. The appropriate loss function is used in the learning process. The proposed method is called Prioritized Semi-supervised Deep Embedded Clustering (PSDEC). The results are compared using a few standard data sets against recent and classical similar methods. PSDEC is found to achieve a better result than un-prioritized constraints.","PeriodicalId":184353,"journal":{"name":"2022 International Conference on Innovative Trends in Information Technology (ICITIIT)","volume":"24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2022-02-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122126361","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：