J. N. V. R. Swarup Kumar
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
J. N. V. R. Swarup Kumar has a total of 35 co-authors. Here are the top co-authors by number of shared publications:
- Devaganugula N. V. S. L. S. Indira
Seshadri Rao Gudlavalleru Engineering college
2 shared publications Dnvsls Indira
SR Gudlavalleru Engineering College,Dept. of IT,Gudlavalleru,India
2 shared publicationsAkuri Eswar Sravanth
SR Gudlavalleru Engineering College,Dept. of CSE,Gudlavalleru,Andhra Pradesh,India,521356
1 shared publication- Anil Kumar Bisht
MJP Rohilkhand University, India
1 shared publication Atmuri Trinadh Kumar
SR Gudlavalleru Engineering College,Dept. of CSE,Gudlavalleru,Andhra Pradesh,India,521356
1 shared publication- Balajee Jeyakumar
Department of CSE(AI), Mother Theresa Institute of Engineering and Technology, Chittoor, Andhra Pradesh, India
1 shared publication
Latest Publications
- Open Access
- 24 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 195436-195449
A Correlation-Aware Feature Selection and Explainable AI Framework for Avian Influenza Outbreak Prediction
Timely and accurate prediction of infectious disease outbreaks is essential for effective public health response and resource allocation. Machine learning (ML) models have shown promise in forecasting such outbreaks by uncovering complex patterns in surveillance data. However, many existing models suffer from diminished interpretability and reduced performance due to the presence of redundant or highly correlated features. This issue not only affects model generalizability but also hinders domain experts from gaining actionable insights. Furthermore, most prior work in disease prediction has overlooked the combined impact of feature correlation and post-hoc model explainability. To address this gap, this study proposes a two-phase framework that integrates correlation-aware feature selection with explainable artificial intelligence (XAI). In the first phase, statistical techniques including Pearson correlation, mutual information, and variance inflation factor (VIF) are used to identify and remove redundant features. In the second phase, models are trained on both full and refined feature sets, followed by SHAP (SHapley Additive Explanations) analysis to assess feature contributions and model transparency. The framework is evaluated using avian influenza outbreak data, where refined feature subsets consistently led to improved F1-scores, AUC-PR, accuracy, and execution time. SHAP plots confirmed that retained features were more interpretable and biologically relevant, while statistical validation via paired t-tests and effect size analysis showed that the improvements were significant (p < 0.05). Sensitivity analysis further validated the robustness of the chosen correlation threshold. The proposed approach enhances both performance and explainability, making it a robust and generalizable solution for epidemiological modeling. This work contributes a reproducible methodology that bridges statistical feature evaluation with interpretable ML, advancing practical deployment in disease surveillance systems.
- 24 October 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 195436-195449
Open AccessA Correlation-Aware Feature Selection and Explainable AI Framework for Avian Influenza Outbreak Prediction
Timely and accurate prediction of infectious disease outbreaks is essential for effective public health response and resource allocation. Machine learning (ML) models have shown promise in forecasting such outbreaks by uncovering complex patterns in surveillance data. However, many existing models suffer from diminished interpretability and reduced performance due to the presence of redundant or highly correlated features. This issue not only affects model generalizability but also hinders domain experts from gaining actionable insights. Furthermore, most prior work in disease prediction has overlooked the combined impact of feature correlation and post-hoc model explainability. To address this gap, this study proposes a two-phase framework that integrates correlation-aware feature selection with explainable artificial intelligence (XAI). In the first phase, statistical techniques including Pearson correlation, mutual information, and variance inflation factor (VIF) are used to identify and remove redundant features. In the second phase, models are trained on both full and refined feature sets, followed by SHAP (SHapley Additive Explanations) analysis to assess feature contributions and model transparency. The framework is evaluated using avian influenza outbreak data, where refined feature subsets consistently led to improved F1-scores, AUC-PR, accuracy, and execution time. SHAP plots confirmed that retained features were more interpretable and biologically relevant, while statistical validation via paired t-tests and effect size analysis showed that the improvements were significant (p < 0.05). Sensitivity analysis further validated the robustness of the chosen correlation threshold. The proposed approach enhances both performance and explainability, making it a robust and generalizable solution for epidemiological modeling. This work contributes a reproducible methodology that bridges statistical feature evaluation with interpretable ML, advancing practical deployment in disease surveillance systems.
- 26 February 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 25 (7) , 11955-11962
Enhanced Connectivity for Non-Critical In-Vehicle Systems Using EnRF24L01
There has been a significant paradigm shift from wired to wireless technology in in-vehicular networks. This shift is driven by the need for greater scalability, cost-effectiveness, and flexibility. In the automotive industry, traditional wired protocols such as the Local Interconnect Network (LIN) and Media Oriented Systems Transport (MOST) for non-critical systems add complexity to installation and maintenance, incur higher material costs, and offer limited scalability and mobility. Non-critical systems, such as infotainment and weather forecast systems, do not require low latency and do not impair vehicle function when unavailable. This paper presents an advanced methodology for enhancing connectivity in non-critical in-vehicular networks using Nordic Semiconductor’s enhanced- nRF24L01 (EnRF24L01) module. The EnRF24L01 module is the nRF24L01 module that incorporated the Sensor-Medium Access Control (S-MAC) algorithm for energy-efficient communication. The proposed method enables seamless communication between non-critical systems using a tree-based master and slave architecture, where the master is the actuator and the slave is the sensor node. To optimize energy efficiency using synchronized sleep/wake schedules, reduce power consumption, and enhance scalability, the S-MAC protocol was incorporated. Comprehensive experiments were conducted in simulated environments using OPNET and Proteus Circuit Simulators, analyzing critical performance metrics: latency, jitter, throughput, packet delivery ratio, and energy efficiency. The results indicate that the proposed method supports a greater number of nodes with enhanced data transmission rates and operates at lower voltages, thereby extending the communication range and reducing overall power consumption. Additionally, hardware simulation results demonstrate the successful integration of EnRF24L01 modules with Arduino for wireless data transmission, showing significant improvements in scalability, energy efficiency, and adaptability, as well as architectural and operational costs and maintenance efficiency.
- 26 February 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 25 (7) , 11955-11962
Enhanced Connectivity for Non-Critical In-Vehicle Systems Using EnRF24L01
There has been a significant paradigm shift from wired to wireless technology in in-vehicular networks. This shift is driven by the need for greater scalability, cost-effectiveness, and flexibility. In the automotive industry, traditional wired protocols such as the Local Interconnect Network (LIN) and Media Oriented Systems Transport (MOST) for non-critical systems add complexity to installation and maintenance, incur higher material costs, and offer limited scalability and mobility. Non-critical systems, such as infotainment and weather forecast systems, do not require low latency and do not impair vehicle function when unavailable. This paper presents an advanced methodology for enhancing connectivity in non-critical in-vehicular networks using Nordic Semiconductor’s enhanced- nRF24L01 (EnRF24L01) module. The EnRF24L01 module is the nRF24L01 module that incorporated the Sensor-Medium Access Control (S-MAC) algorithm for energy-efficient communication. The proposed method enables seamless communication between non-critical systems using a tree-based master and slave architecture, where the master is the actuator and the slave is the sensor node. To optimize energy efficiency using synchronized sleep/wake schedules, reduce power consumption, and enhance scalability, the S-MAC protocol was incorporated. Comprehensive experiments were conducted in simulated environments using OPNET and Proteus Circuit Simulators, analyzing critical performance metrics: latency, jitter, throughput, packet delivery ratio, and energy efficiency. The results indicate that the proposed method supports a greater number of nodes with enhanced data transmission rates and operates at lower voltages, thereby extending the communication range and reducing overall power consumption. Additionally, hardware simulation results demonstrate the successful integration of EnRF24L01 modules with Arduino for wireless data transmission, showing significant improvements in scalability, energy efficiency, and adaptability, as well as architectural and operational costs and maintenance efficiency.
- 14 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Consumer Electronics
- Vol. 71 (1) , 273-280
An Innovative Secure and Privacy-Preserving Federated Learning-Based Hybrid Deep Learning Model for Intrusion Detection in Internet-Enabled Wireless Sensor Networks
Cyberspace faces numerous security challenges, necessitating advanced research in intrusion detection systems (IDS) to mitigate vulnerabilities. Wireless Sensor Networks (WSNs) connected to the Internet are particularly vulnerable, requiring robust protection mechanisms. Traditional IDS struggle with identifying unknown attacks and maintaining data privacy, especially in WSNs. This study proposes a novel approach integrating Stacked Convolutional Neural Networks (SCNN), Bidirectional Long Short Term Memory (Bi-LSTM), and the African Vulture Optimization Algorithm (AVOA) within a framework of Federated Learning (FL). The integrated model, SCNN-Bi-LSTM-AVOA-FL, aims to enhance intrusion detection efficacy while preserving data privacy. A tailored AVOA optimization method fine-tunes SCNN-Bi-LSTM hyperparameters, leveraging specialized datasets (WSN-DS, CIC-IDS-2017, and WSN-BFSF) for attack detection and categorization. Evaluations compare variants with and without FL techniques (proposed-1 and proposed-2) across metrics such as accuracy, precision, recall, and F1-Score. Results demonstrate significant improvements in prediction performance, validating the efficacy of the proposed approach in enhancing IDS capabilities for WSNs. This research contributes a comprehensive framework for addressing security challenges in WSNs through advanced machine learning and optimization techniques.
- 14 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Consumer Electronics
- Vol. 71 (1) , 273-280
An Innovative Secure and Privacy-Preserving Federated Learning-Based Hybrid Deep Learning Model for Intrusion Detection in Internet-Enabled Wireless Sensor Networks
Cyberspace faces numerous security challenges, necessitating advanced research in intrusion detection systems (IDS) to mitigate vulnerabilities. Wireless Sensor Networks (WSNs) connected to the Internet are particularly vulnerable, requiring robust protection mechanisms. Traditional IDS struggle with identifying unknown attacks and maintaining data privacy, especially in WSNs. This study proposes a novel approach integrating Stacked Convolutional Neural Networks (SCNN), Bidirectional Long Short Term Memory (Bi-LSTM), and the African Vulture Optimization Algorithm (AVOA) within a framework of Federated Learning (FL). The integrated model, SCNN-Bi-LSTM-AVOA-FL, aims to enhance intrusion detection efficacy while preserving data privacy. A tailored AVOA optimization method fine-tunes SCNN-Bi-LSTM hyperparameters, leveraging specialized datasets (WSN-DS, CIC-IDS-2017, and WSN-BFSF) for attack detection and categorization. Evaluations compare variants with and without FL techniques (proposed-1 and proposed-2) across metrics such as accuracy, precision, recall, and F1-Score. Results demonstrate significant improvements in prediction performance, validating the efficacy of the proposed approach in enhancing IDS capabilities for WSNs. This research contributes a comprehensive framework for addressing security challenges in WSNs through advanced machine learning and optimization techniques.
- 21 September 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1073-1080
Personalized Fuel Assistance Recommendation System based on Artificial Neural Networks
Everybody wants a smooth and uninterrupted journey to get to their destination, but there are a lot of problems that could make a journey exhausting and tiresome. One of the major problems is fuel. Unless their fuel gauge indicates a warning, most people don't check their fuel tanks while driving. At times like this, there may or may not be a fuel station around their location. If the person is driving in a city, there is a possibility that there can be a fuel station every 300 meters. But when it comes to riding in the city outskirts, the chances are low. Here's the question: What's the likelihood of there being a fuel station there? Does that location have the possibility of not having a fuel station? In another scenario, what if there are three fuel stations at a particular location and the person has no idea which station to choose? People tend to pick the best option out of several available options when they have choices. When the customer has access to the station's details and customer reviews, it will be easier to choose from the three fuel stations. This paper describes a software web application that displays a list of all the fuel stations in a given area when a location is entered. To access this application, the user must input his or her location and destination. If the fuel tank reaches its limit, an e-mail is sent. The user can now use this software application to locate fuel stations near their location. The application shows all the fuel stations nearby and provides information about each station such as the cost of fuel, ratings, and customer reviews. In this way, the user can select the best option for their trip and avoid any hassles.
- 21 September 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1073-1080
Personalized Fuel Assistance Recommendation System based on Artificial Neural Networks
Everybody wants a smooth and uninterrupted journey to get to their destination, but there are a lot of problems that could make a journey exhausting and tiresome. One of the major problems is fuel. Unless their fuel gauge indicates a warning, most people don't check their fuel tanks while driving. At times like this, there may or may not be a fuel station around their location. If the person is driving in a city, there is a possibility that there can be a fuel station every 300 meters. But when it comes to riding in the city outskirts, the chances are low. Here's the question: What's the likelihood of there being a fuel station there? Does that location have the possibility of not having a fuel station? In another scenario, what if there are three fuel stations at a particular location and the person has no idea which station to choose? People tend to pick the best option out of several available options when they have choices. When the customer has access to the station's details and customer reviews, it will be easier to choose from the three fuel stations. This paper describes a software web application that displays a list of all the fuel stations in a given area when a location is entered. To access this application, the user must input his or her location and destination. If the fuel tank reaches its limit, an e-mail is sent. The user can now use this software application to locate fuel stations near their location. The application shows all the fuel stations nearby and provides information about each station such as the cost of fuel, ratings, and customer reviews. In this way, the user can select the best option for their trip and avoid any hassles.
- 21 September 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 656-663
Secured Cloud Application for Detection of Brain Tumor using Deep Learning Algorithms
Brain tumor segmentation is one of the most challenging and essential problems in medical image processing since human-assisted manual classification can lead to inaccurate prognosis and diagnosis. Having a huge amount of data to analyse makes the problem exponentially more complex. As a result of the great variety of tumor appearances and the close resemblance between normal and tumor tissues, it is difficult to isolate tumor regions from images. By using convolutional neural networks, this study was able to extract 2D MR brain pictures and identify tumors in the brain (MRI). A real-time dataset with a wide range of tumor sizes, locations, shapes, and intensities is used for performing system analysis. A Convolutional NeuralNetwork (CNN) built with Keras and Tensor Flow is used in this research work since it outperforms the ordinary neural networks. An online cloud application is used to determine whether an image contains a tumor or not, and it employs the secure AES256 approach for transmitting an MRI image to the cloud server. According to the obtained results, CNN produces an accuracy rate of 97.87%. The primary goal of this study is to differentiate the normal and abnormal pixels in order to determine whether or not the patient is at risk of developing a tumor.
- 21 September 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 656-663
Secured Cloud Application for Detection of Brain Tumor using Deep Learning Algorithms
Brain tumor segmentation is one of the most challenging and essential problems in medical image processing since human-assisted manual classification can lead to inaccurate prognosis and diagnosis. Having a huge amount of data to analyse makes the problem exponentially more complex. As a result of the great variety of tumor appearances and the close resemblance between normal and tumor tissues, it is difficult to isolate tumor regions from images. By using convolutional neural networks, this study was able to extract 2D MR brain pictures and identify tumors in the brain (MRI). A real-time dataset with a wide range of tumor sizes, locations, shapes, and intensities is used for performing system analysis. A Convolutional NeuralNetwork (CNN) built with Keras and Tensor Flow is used in this research work since it outperforms the ordinary neural networks. An online cloud application is used to determine whether an image contains a tumor or not, and it employs the secure AES256 approach for transmitting an MRI image to the cloud server. According to the obtained results, CNN produces an accuracy rate of 97.87%. The primary goal of this study is to differentiate the normal and abnormal pixels in order to determine whether or not the patient is at risk of developing a tumor.
- RetractedOpen Access
- 22 August 2022
- Published by Wiley in BioMed Research International
- Vol. 2022 (1) , 1755460
[Retracted] Lung Cancer Classification and Prediction Using Machine Learning and Image Processing
Lung cancer is a potentially lethal illness. Cancer detection continues to be a challenge for medical professionals. The true cause of cancer and its complete treatment have still not been discovered. Cancer that is caught early enough can be treated. Image processing methods such as noise reduction, feature extraction, identification of damaged regions, and maybe a comparison with data on the medical history of lung cancer are used to locate portions of the lung that have been impacted by cancer. This research shows an accurate classification and prediction of lung cancer using technology that is enabled by machine learning and image processing. To begin, photos need to be gathered. In the experimental investigation, 83 CT scans from 70 distinct patients were utilized as the dataset. The geometric mean filter is used during picture preprocessing. As a consequence, image quality is enhanced. The K‐means technique is then used to segment the images. The part of the image may be found using this segmentation. Then, classification methods using machine learning are used. For the classification, ANN, KNN, and RF are some of the machine learning techniques that were used. It is found that the ANN model is producing more accurate results for predicting lung cancer.
- 22 August 2022
- Published by Wiley in BioMed Research International
- Vol. 2022 (1) , 1755460
RetractedOpen Access[Retracted] Lung Cancer Classification and Prediction Using Machine Learning and Image Processing
Lung cancer is a potentially lethal illness. Cancer detection continues to be a challenge for medical professionals. The true cause of cancer and its complete treatment have still not been discovered. Cancer that is caught early enough can be treated. Image processing methods such as noise reduction, feature extraction, identification of damaged regions, and maybe a comparison with data on the medical history of lung cancer are used to locate portions of the lung that have been impacted by cancer. This research shows an accurate classification and prediction of lung cancer using technology that is enabled by machine learning and image processing. To begin, photos need to be gathered. In the experimental investigation, 83 CT scans from 70 distinct patients were utilized as the dataset. The geometric mean filter is used during picture preprocessing. As a consequence, image quality is enhanced. The K‐means technique is then used to segment the images. The part of the image may be found using this segmentation. Then, classification methods using machine learning are used. For the classification, ANN, KNN, and RF are some of the machine learning techniques that were used. It is found that the ANN model is producing more accurate results for predicting lung cancer.
- 29 July 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-7
Self-Attention Architecture for Ingredients Generation from Food Images
Food filming is becoming more popular among food connoisseurs. Each meal has a narrative that is detailed in a lengthy recipe, and sadly, just looking at a dish provides no insight into the preparation of food. There are various websites that aid in recognizing a meal by its components; however, no method has been successful at forecasting the ingredients in a dish. As a consequence, this research proposes a method for automatically generating the dish's recipe. This approach estimates the image's title and ingredients and then generates the image's featured meal's specific cooking instructions. This investigation looked at a range of Indian cuisines, including lunch and breakfast. To enhance user-friendly, this paper provides a web application that displays the recipe with a photograph of the dish.
- 29 July 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-7
Self-Attention Architecture for Ingredients Generation from Food Images
Food filming is becoming more popular among food connoisseurs. Each meal has a narrative that is detailed in a lengthy recipe, and sadly, just looking at a dish provides no insight into the preparation of food. There are various websites that aid in recognizing a meal by its components; however, no method has been successful at forecasting the ingredients in a dish. As a consequence, this research proposes a method for automatically generating the dish's recipe. This approach estimates the image's title and ingredients and then generates the image's featured meal's specific cooking instructions. This investigation looked at a range of Indian cuisines, including lunch and breakfast. To enhance user-friendly, this paper provides a web application that displays the recipe with a photograph of the dish.
- 13 March 2022
- Published by Springer Nature
- p. 271-281
Virtual Assistant Based on Facial Emotions Using Convolution Neural Networks
- 13 March 2022
- Published by Springer Nature
- p. 271-281
Virtual Assistant Based on Facial Emotions Using Convolution Neural Networks
- 17 July 2021
- Published by Elsevier in Materials Today: Proceedings
- Vol. 80, 2945-2950
A novel cluster based highway mobility model (CHMM) for low power vehicular networks
- 17 July 2021
- Published by Elsevier in Materials Today: Proceedings
- Vol. 80, 2945-2950
A novel cluster based highway mobility model (CHMM) for low power vehicular networks