Cloud Computing, Edge Computing, Artificial Intelligence and NLP
15
Scopus Publications
Scopus Publications
Enhanced sentiment analysis on social media using BERT and multimodal attention-based fusion A. Amirtha Saravanan, Mani Deepak Choudhry, S. Jeevanandham, S. Ramasamy, K. Suvalakshmi, D. Karthikeswaran Automatika, 2026 Sentiment analysis on social media is increasingly challenging due to the growing prevalence of multimedia posts to convey complex emotions. Traditional sentiment analysis methods, which rely solely on textual or visual information, often fail to capture the nuanced interplay between modalities, leading to suboptimal predictions. Motivated by the limitations of unimodal approaches in handling sarcasm, this study proposes a novel multimodal sentiment analysis model that integrates textual and visual features for enhanced understanding. The proposed model leverages BERT and BiLSTM networks for capturing contextual and sequential dependencies in text, while CNNs extract rich spatial features from images. An attention-based fusion mechanism further enhances the integration of these modalities by focusing on the most informative elements in both text and pictures. Pretrained CNNs are fine-tuned through transfer learning to improve visual feature extraction, and the combined multimodal representation is used for sentiment prediction. Experimental results demonstrate that the proposed model achieves 92% accuracy, 89.5% precision, 89% recall, and an F1-score of 89.2%, outperforming conventional unimodal and existing multimodal approaches. This work lays the foundation for future extensions to include other modalities such as audio or video, enabling more comprehensive emotion understanding in real-world applications.
Optimized Architecture and Strategies for High Performance Computing in Cloud and Hybrid Environments S. Shanthi, R. Karthik, R. Madhuramya, Vimit Varghese, G.M.Banu Priya, S. Ramasamy Proceedings of 8th International Conference on Intelligent Sustainable Systems Iciss 2026, 2026 High-performance computing has traditionally been based on large on-site supercomputers that provide powerful but costly and rigid performance. HPC workloads may now benefit from flexible on-demand access to computing resources because to the expansion of cloud computing. Yet issues like inconsistent performance slow networks and complex data that processing persist when HPC is used to cloud or hybrid environments. A practical method for enhancing the performance of HPC applications in the cloud and hybrid settings is presented in this study. The framework leverages hardware that the acceleration and intelligent scheduling for containerized programs to balance the cost of performance and resource use. To lower communication latency, it also uses the software-defined networking adaptive task transfer and data-aware scheduling. When compared to conventional cloud-based HPC systems that experimental testing using the scientific and industrial workloads demonstrates up to 17% quicker performance. An average 10- 14% reduction in cloud compute cost, primarily due to reduced billed execution time on cloud resources. The results demonstrate that while retaining scalability to security and sustainability the suggested hybrid HPC architecture can manage the demanding applications such as genetic research, climate modeling, and AI simulations.
A WSN-Based WbCNF Algorithm to Enable Cloud Computing and Big Data Analytics in Healthcare S. Ramasamy, V. Baby Vennila International Journal of Communication Systems, 2025 Cloud computing is a technology that enables the internet‐based delivery of computer resources and services, giving customers access to on‐demand infrastructure, platforms, and applications without the need for local hardware or direct control. Cloud computing in healthcare is the use of internet‐based platforms and services for storing, managing, and processing healthcare data. Cloud computing in big data is the use of cloud‐based infrastructure and services to store, manage, and analyze large datasets. This paper proposed a WSN‐based Weighted Boolean Conjunctive Normal Form (WbCNF) method. This approach in cloud computing refers to the use of an optimization technique to handle large‐scale, distributed problems such as resource allocation, scheduling, and decision‐making in cloud environments. Cloud computing environments generate complex and dynamic constraints that can be modeled using weighted CNF formulas to indicate priorities or costs. The work allocation mechanism can be greatly improved, and job execution times can be lowered. This can be performed by employing a revolutionary whale‐based convolutional neural framework technique. The Python framework is used in the proposed approach. The experimental results show that the number of tasks required for the experiment has decreased due to the computing time.
Smart Healthcare and Wellness Systems for Real-Time Monitoring P.Sathish Kumar, T.Manasa Veena, S.Saranya, S.Ramasamy, S.Aruna Deepthi, K. Radha Proceedings of the 7th International Conference on Innovative Data Communication Technologies and Application Icidca 2025, 2025 Innovations in intelligent healthcare and the Internet of Medical Things (IoMT) are transforming medical practice, moving it from traditional reactive treatment toward preventive, real-time, and personalized care. This research presents an Edge–Cloud Healthcare Framework that brings together information from a number of sources, including wearable sensors, biomedical imaging equipment, and environmental monitoring systems. Fundamentally, the framework employs a Hybrid Deep Neural Network (HDNN) that integrates convolutional layers for image-based medical Recurrent layer analysis for interpreting time-series physiological signals. Early adoption is in favor this design. anomaly detection and effective risk prediction. In order to balance the computational burdens, a Dynamic Priority Scheduling Algorithm (DPSA) is applied, To maintain equilibrium among the computational demands and cloud resources based on latency requirements and the criticality of patient data, ensuring fast responses alongside accurate results. evaluation of an experiment utilizing 5G-enabled ECG and SpO<inf xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</inf> wearable Devices' inference latency was average of 14.7 ms and improved diagnostic accuracy by 9.3% over conventional telemedicine models. Moreover, smart home simulations demonstrated the task that is adaptive orchestration less energy consumption by 18% while remaining receptive. In general, the results show that combining edge computing with intelligent arranging mechanisms is able to provide scalable, reliable, and patient-focused healthcare solutions, enabling both ongoing observation and personalized wellness interventions in real-world environments.
An Energy-Efficient Edge-Cloud Computing Framework for Sustainable IoT Infrastructures in Smart Cities A.Nishanandhini, C.Narmada, S.Vijaya Lakshmi, U.Prakash, V.Baby Vennila, S.Ramasamy Proceedings of 6th International Conference on Iot Based Control Networks and Intelligent Systems Icicnis 2025, 2025 This study examines the potential benefits of integrating the edge-cloud computing with the Internet-connected equipment for enhancing environmental sustainability. To cut down the energy use and dependence on the centralized data centers, the suggested architecture are makes the advantage of edge computing. This strategy are backs up the previous studies that show the hybrid edge-cloud systems can reduce the energy consumption for specific workload categories by up to 50%. The suggested system includes a smart resource management strategies like shared resource pooling, machine learning-based energy optimization and two-phase absorption cooling which are based on the current concepts for sustainable edge computing. These techniques are increase scalability, dependability and energy efficiency while maintaining the system stability under changing the conditions. In comparison to conventional setups a simulations are conducted in smart city and IoT contexts show the framework could achieve the better energy usage tracking a lower carbon emissions and higher quality of service. In order to make a future edge-cloud systems both effective and environmentally conscious, the paper are concludes by highlighting the potential research objectives are such as creating a sustainable infrastructure for design and adaptive workload allocation algorithms.
Emerging Paradigms in 6G Wireless Communication for Future Technologies N.Mookhambika, Arun R, G.Arulkumaran, Pavithra S, S.Ramasamy, Maheshkumar V Proceedings of the 7th International Conference on Innovative Data Communication Technologies and Application Icidca 2025, 2025 The growing demand for ultra-reliable, high-speed, and intelligent connectivity has exposed the limitations of existing 5G networks, particularly in meeting the requirements of immersive communication, large-scale automation, and global coverage. To address these challenges, sixth-generation (6G) wireless communication is being designed as a trans-formative framework that integrates terahertz spectrum utilization, artificial intelligence-driven network optimization, intelligent reflecting surfaces, and satellite–terrestrial convergence. This paper examines the enabling technologies and system-level innovations that will underpin 6G, focusing on their capacity to deliver data rates approaching 1 Tbps, sub-millisecond latency, and ubiquitous connectivity. Key findings highlight the potential of 6G to support disruptive applications such as holographic communication, autonomous transportation, extended reality, and AI-powered healthcare services. The contribution of this work lies in synthesizing emerging 6G technologies with real-world application domains, providing insights into how 6G can serve as the backbone of an intelligent, sustainable, and hyper-connected society.
IntelliFuzz: An Advanced Fuzzy Logic Framework for Dynamic Evaluation of Student Performance in Open-Ended Learning Tasks S. Shankar, N. Padmashri, N. Shanmugapriya, S. Ramasamy, P.S. Sruthi International Journal of Computational and Experimental Science and Engineering, 2024 This study presents IntelliFuzz, an advanced fuzzy logic-based assessment system designed for the dynamic evaluation of student performance in open-ended tasks. The proposed system leverages fuzzy logic to address the inherent subjectivity and ambiguity in evaluating tasks such as essays, project work, and case studies. IntelliFuzz incorporates multiple evaluation criteria, including task relevance, critical thinking, creativity, and presentation quality, to generate a comprehensive performance score. Experimental results on a dataset of 500 student submissions demonstrate the effectiveness of IntelliFuzz. The system achieved a 95% accuracy in aligning with expert assessments and reduced evaluation time by 30% compared to traditional manual grading methods. The fuzzy inference system was calibrated using 150 expert feedback samples, yielding an average correlation coefficient of 0.92 between system-generated scores and expert evaluations. Furthermore, IntelliFuzz was rated 85% satisfactory by instructors for its ability to provide consistent and fair evaluations.The study highlights the potential of fuzzy logic in educational assessment, offering a scalable and efficient solution for evaluating subjective student tasks. Future research will focus on integrating machine learning to further enhance the adaptability and precision of the system.
Deep Learning Based Brain Tumor Analysis with Manual Layer Selection International Journal of Intelligent Systems and Applications in Engineering, 2024
Meta-learning through ensemble approach: Bagging, boosting, and random forest strategies S. Ramasamy, H. C. Kantharaju, N. Bindu Madhavi, M. P. Haripriya Toward Artificial General Intelligence Deep Learning Neural Networks Generative AI, 2023 Meta-learning through ensemble approaches is an intriguing subfield of machine learning research. With this method, a more comprehensive learning model is created by combining many machine learning methods, including neural networks and support vector machines. By using an ensemble of models, meta-learning techniques are able to produce more robust results than individual algorithms alone. In addition, ensemble techniques are advantageous because they can easily be expanded to accommodate additional data sources or algorithms. This approach can also embed more knowledge from the data into a more powerful meta-model, which allows the system to generalize better and discover patterns more accurately. In short, metalearning through ensemble approaches is an effective and useful technique for tackling challenging problems in machine learning.
Embedded PID Controller Design Based Self Adjusting Robot S. Ramasamy, N. Senthilkumar, S. Karpagam, U. Ramani, B. Swetha Proceedings of the 2nd International Conference on Artificial Intelligence and Smart Energy Icais 2022, 2022
