Ashok Murugesan
Ashok Murugesan has a total of 54 co-authors. Here are the top co-authors by number of shared publications:
- Kumar Ramasamy
Muthayammal Engineering College, Rasipuram, Namakkal, India
2 shared publications Abbas Thajeel Rhaif Alsahlanee
University of Thi-Qar
1 shared publication- Aslam B. Nandya
Alva’s Institute of Engineering and Technology
1 shared publication B. Sandhiya Rajeshwari
Rajalakshmi Institute of Technology, India
1 shared publication- C. Karthikeyan
Department of Computer Science & Engineering, Saveetha School of Engineering, Saveetha Institute of Medical and Technical Sciences, Chennai, India
1 shared publication - D. Ramya
Department of Electrical and Electronics Engineering, Sathyabama Institute of Science and Technology, Chennai, India
1 shared publication
Latest Publications
- 7 October 2025
- Published by Springer Nature
- p. 709-720
Technical Language Processing—Generalization of Machine Language
- 7 October 2025
- Published by Springer Nature
- p. 709-720
Technical Language Processing—Generalization of Machine Language
- Open Access
- 3 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in TechRxiv
Cybersecurity Incident Response Automation Using Artificial Intelligence
In today's digital world, the speed and sophistication of cyber attacks are growing at a breakneck speed, and this calls for instant and intelligent response mechanisms. This project proposes an AI-powered solution for automating cybersecurity incident response, namely WhatsApp surveillance against abusive content. The system employs artificial intelligence models like Google's Gemma and DeepAI APIs to detect malicious messages, images, and videos in real-time. Upon detection, it captures screenshots using ADB, extracts vital metadata (sender, receiver, timestamp), and sends automated email alerts with complete evidence. Compared to traditional methods, this solution operates in the background seamlessly, is multidevice compatible, and is data privacy legislation compliant. The project addresses the requirement for standardized, efficient, and scalable cybersecurity response systems for modern communication platforms.
- 3 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in TechRxiv
Open AccessCybersecurity Incident Response Automation Using Artificial Intelligence
In today's digital world, the speed and sophistication of cyber attacks are growing at a breakneck speed, and this calls for instant and intelligent response mechanisms. This project proposes an AI-powered solution for automating cybersecurity incident response, namely WhatsApp surveillance against abusive content. The system employs artificial intelligence models like Google's Gemma and DeepAI APIs to detect malicious messages, images, and videos in real-time. Upon detection, it captures screenshots using ADB, extracts vital metadata (sender, receiver, timestamp), and sends automated email alerts with complete evidence. Compared to traditional methods, this solution operates in the background seamlessly, is multidevice compatible, and is data privacy legislation compliant. The project addresses the requirement for standardized, efficient, and scalable cybersecurity response systems for modern communication platforms.
- Open Access
- 3 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in TechRxiv
ML FOR ANALYSING HPC LOG FILES: A real-time AI-powered dashboard that analyses HPC log files to visualize resource usage and categorize users for efficient system monitoring and management
The rapid expansion of High-Performance Computing (HPC) systems has created enormous log files that capture system events, resource utilization, and user activity. It is time-consuming and error-prone to manually analyze these logs. The project offers a Machine Learning (ML)-based approach of automatically processing HPC log files to obtain actionable insights, anomalies, and maximize resource utilization. Using clustering, classification, and time-series analysis, the system can detect patterns, predict system crashes, and analyze user behavior. A real-time dashboard is offered to graphically display resource utilization, top users, and anomaly alarms. The solution enhances system reliability, performance monitoring, and decisionmaking for administrators such that they can proactively manage advanced HPC environments.
- 3 August 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in TechRxiv
Open AccessML FOR ANALYSING HPC LOG FILES: A real-time AI-powered dashboard that analyses HPC log files to visualize resource usage and categorize users for efficient system monitoring and management
The rapid expansion of High-Performance Computing (HPC) systems has created enormous log files that capture system events, resource utilization, and user activity. It is time-consuming and error-prone to manually analyze these logs. The project offers a Machine Learning (ML)-based approach of automatically processing HPC log files to obtain actionable insights, anomalies, and maximize resource utilization. Using clustering, classification, and time-series analysis, the system can detect patterns, predict system crashes, and analyze user behavior. A real-time dashboard is offered to graphically display resource utilization, top users, and anomaly alarms. The solution enhances system reliability, performance monitoring, and decisionmaking for administrators such that they can proactively manage advanced HPC environments.
- 28 July 2025
- Published by IGI Global Scientific Publishing
- p. 77-108
Ensuring Fairness in AI-Driven University Assessments
The growing concern over bias in AI systems, particularly in high-impact areas like healthcare, hiring, criminal justice and education emphasizes the need for ethical implementations to ensure fairness. To address bias, indicators such as demographic parity, equalized odds, calibration, and disparate impact measurement are crucial in monitoring and reducing biased outcomes across different demographic groups. In a scenario of university students taking online courses, where the exams were evaluated by an AI system, raising concerns about potential bias in the evaluation process. The key problem identified is ensuring that AI-driven assessments fairly evaluate students from diverse backgrounds, without disproportionately favoring or penalizing certain groups. Bias detection and mitigation efforts are essential to foster trust, fairness, and consistency in AI-based evaluations.
- 28 July 2025
- Published by IGI Global Scientific Publishing
- p. 77-108
Ensuring Fairness in AI-Driven University Assessments
The growing concern over bias in AI systems, particularly in high-impact areas like healthcare, hiring, criminal justice and education emphasizes the need for ethical implementations to ensure fairness. To address bias, indicators such as demographic parity, equalized odds, calibration, and disparate impact measurement are crucial in monitoring and reducing biased outcomes across different demographic groups. In a scenario of university students taking online courses, where the exams were evaluated by an AI system, raising concerns about potential bias in the evaluation process. The key problem identified is ensuring that AI-driven assessments fairly evaluate students from diverse backgrounds, without disproportionately favoring or penalizing certain groups. Bias detection and mitigation efforts are essential to foster trust, fairness, and consistency in AI-based evaluations.
- Open Access
- 24 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in TechRxiv
A Data-Driven Framework for Predicting and Visualizing Atmospheric Patterns through Sensor Integration and Machine Learning Correlation Analysis
Using real-time environmental data from Internet of Things-based weather sensors, this project offers a smart weather forecasting system. The Meenambakkam region's temperature, humidity, pressure, wind speed, and solar radiation are all included in the dataset. We used a number of machine learning models, such as K-Means Clustering, LSTM (Long Short-Term Memory), Random Forest Regressor, Decision Tree Regressor, and Linear Regression, to increase forecasting accuracy. We used metrics such as the accuracy percentage and R2 score to assess each model's performance. The best accuracy in temperature forecasting was attained by the LSTM model, which is perfect for time series prediction. K-Means While traditional models provided accurate and timely forecasts, clustering revealed hidden patterns in weather behaviour This system demonstrates how effective and automated weather forecasting is made possible by fusing IoT and machine learning.
- 24 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in TechRxiv
Open AccessA Data-Driven Framework for Predicting and Visualizing Atmospheric Patterns through Sensor Integration and Machine Learning Correlation Analysis
Using real-time environmental data from Internet of Things-based weather sensors, this project offers a smart weather forecasting system. The Meenambakkam region's temperature, humidity, pressure, wind speed, and solar radiation are all included in the dataset. We used a number of machine learning models, such as K-Means Clustering, LSTM (Long Short-Term Memory), Random Forest Regressor, Decision Tree Regressor, and Linear Regression, to increase forecasting accuracy. We used metrics such as the accuracy percentage and R2 score to assess each model's performance. The best accuracy in temperature forecasting was attained by the LSTM model, which is perfect for time series prediction. K-Means While traditional models provided accurate and timely forecasts, clustering revealed hidden patterns in weather behaviour This system demonstrates how effective and automated weather forecasting is made possible by fusing IoT and machine learning.
- 5 July 2025
- Published by Springer Nature
- p. 147-159
A Hybrid Evolutionary Deep Learning Model Integrating Multi-modal Data for Optimizing Ovarian Cancer Diagnosis
- 5 July 2025
- Published by Springer Nature
- p. 147-159
A Hybrid Evolutionary Deep Learning Model Integrating Multi-modal Data for Optimizing Ovarian Cancer Diagnosis
- Open Access
- 19 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in TechRxiv
Predicting Resource Bottlenecks in HPC Environments Using AI: A Data-Driven Method for Identifying and Visualizing Compute and Memory Limitations via Correlation Analysis
The use of artificial intelligence (AI), especially correlation analysis and feature influence scoring, shows strong potential for identifying performance bottlenecks in High-Performance Computing (HPC) environments. Yet, the lack of standardized ways to visualize and interpret these bottlenecks limits reliable diagnostics. This paper introduces a new framework that detects memory and compute bottlenecks using AI-based methods without predictive classifiers. It analyzes how hardware and workload features-such as memory cycle speed, bus width, memory channels, overclocking, accelerator bandwidth, and workload type-affect system performance. By generating influence scores and visualizing results through log-scaled bar plots and heatmaps, the framework offers a clear, reproducible approach to identifying bottlenecks. This work improves interpretability in HPC monitoring and sets the stage for scalable, explainable AI tools that optimize workload distribution and resource use in modern HPC systems.
- 19 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in TechRxiv
Open AccessPredicting Resource Bottlenecks in HPC Environments Using AI: A Data-Driven Method for Identifying and Visualizing Compute and Memory Limitations via Correlation Analysis
The use of artificial intelligence (AI), especially correlation analysis and feature influence scoring, shows strong potential for identifying performance bottlenecks in High-Performance Computing (HPC) environments. Yet, the lack of standardized ways to visualize and interpret these bottlenecks limits reliable diagnostics. This paper introduces a new framework that detects memory and compute bottlenecks using AI-based methods without predictive classifiers. It analyzes how hardware and workload features-such as memory cycle speed, bus width, memory channels, overclocking, accelerator bandwidth, and workload type-affect system performance. By generating influence scores and visualizing results through log-scaled bar plots and heatmaps, the framework offers a clear, reproducible approach to identifying bottlenecks. This work improves interpretability in HPC monitoring and sets the stage for scalable, explainable AI tools that optimize workload distribution and resource use in modern HPC systems.
- Open Access
- 28 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in TechRxiv
Predicting 1P/ Halley's Comet Pathway Using Linear Regression: Addressing the Lack of Standardization in Machine Learning Model Evaluation for Orbit Prediction Interoperability
The integration of machine learning (ML) especially Linear Regression into orbit determination has shown significant promise for enhancing the accuracy of celestial body trajectory predictions, including cometary pathways. However, the lack of standardized benchmarks for evaluating ML models in orbit prediction poses a critical challenge to interoperability among space mission systems. This paper introduces a novel framework for predicting cometary pathways using Linear regression, addressing the standardization gap by proposing a unified evaluation methodology tailored to ML-driven orbit prediction. Hence, this paper aims to analyze the lack of standardized evaluation criteria, create a methodology for predicting cometary pathway and evaluate its error.
- 28 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in TechRxiv
Open AccessPredicting 1P/ Halley's Comet Pathway Using Linear Regression: Addressing the Lack of Standardization in Machine Learning Model Evaluation for Orbit Prediction Interoperability
The integration of machine learning (ML) especially Linear Regression into orbit determination has shown significant promise for enhancing the accuracy of celestial body trajectory predictions, including cometary pathways. However, the lack of standardized benchmarks for evaluating ML models in orbit prediction poses a critical challenge to interoperability among space mission systems. This paper introduces a novel framework for predicting cometary pathways using Linear regression, addressing the standardization gap by proposing a unified evaluation methodology tailored to ML-driven orbit prediction. Hence, this paper aims to analyze the lack of standardized evaluation criteria, create a methodology for predicting cometary pathway and evaluate its error.
- 18 April 2025
- Published by Springer Nature
- p. 155-167
An Early Detection of Tomato Plant Disease with Deep Reinforcement Learning and CNN
- 18 April 2025
- Published by Springer Nature
- p. 155-167
An Early Detection of Tomato Plant Disease with Deep Reinforcement Learning and CNN
- Open Access
- 1 January 2024
- Published by Institute of Electronics and Computer in Journal of Artificial Intelligence and Systems
- Vol. 6 (1) , 59-75
Mitigating Carbon Footprint through Green Software: A Comprehensive Framework for Energy-Efficient Data Centers and User-Friendly Carbon Footprint Calculation
The carbon footprint of the world is increasing day by day by the activities caused by every citizen. This is because the demand for energy increases and due to the lack of awareness.The average carbon emission is around 4.72 tonnes per person. There are some free online tools which are available for calculating the carbon footprint of each citizen. But those work as a datacentre in which data must be entered manually by the user which makes it quite inefficient. The existing methods do not meet the requirement of ease of handling the data.This software and the datacentres were developed which includes the availability of a user-friendly interface.Data centers also play a major role in carbon footprint, they emit CO2 and consume more energy. Data centers and new technology adoptions are mainly causing this carbon emission.These Data Centres use cloud energy to serve the user-generated requests and this energy consumption is the basic cause of carbon emission. Therefore, the energy efficiency of the data center must be governed. To reduce it, the concept of green software is introduced. Though hardware is seen as the main culprit for the consumption of energy, software plays a tremendous role in determining the efficiency of the hardware. Therefore, software must be energy efficient to handle the data center in low cost, low consumption, and low pressure.This led to the development of GREEN Software. The implementation of green software involves various factors.To reduce it, data center simulators are being used. This simulator calculates the amount of energy consumed and the amount of CO2 emitted, which as a result computes the change in carbon footprint due to it. This green software design framework provides a separate track session on energy consumption by the data center and cloud service provider along with the acquired energy efficiency level by the specific center.
- 1 January 2024
- Published by Institute of Electronics and Computer in Journal of Artificial Intelligence and Systems
- Vol. 6 (1) , 59-75
Open AccessMitigating Carbon Footprint through Green Software: A Comprehensive Framework for Energy-Efficient Data Centers and User-Friendly Carbon Footprint Calculation
The carbon footprint of the world is increasing day by day by the activities caused by every citizen. This is because the demand for energy increases and due to the lack of awareness.The average carbon emission is around 4.72 tonnes per person. There are some free online tools which are available for calculating the carbon footprint of each citizen. But those work as a datacentre in which data must be entered manually by the user which makes it quite inefficient. The existing methods do not meet the requirement of ease of handling the data.This software and the datacentres were developed which includes the availability of a user-friendly interface.Data centers also play a major role in carbon footprint, they emit CO2 and consume more energy. Data centers and new technology adoptions are mainly causing this carbon emission.These Data Centres use cloud energy to serve the user-generated requests and this energy consumption is the basic cause of carbon emission. Therefore, the energy efficiency of the data center must be governed. To reduce it, the concept of green software is introduced. Though hardware is seen as the main culprit for the consumption of energy, software plays a tremendous role in determining the efficiency of the hardware. Therefore, software must be energy efficient to handle the data center in low cost, low consumption, and low pressure.This led to the development of GREEN Software. The implementation of green software involves various factors.To reduce it, data center simulators are being used. This simulator calculates the amount of energy consumed and the amount of CO2 emitted, which as a result computes the change in carbon footprint due to it. This green software design framework provides a separate track session on energy consumption by the data center and cloud service provider along with the acquired energy efficiency level by the specific center.