P. Senthil Kumar

Dr. P. Senthil Kumar

orcid

Research Focus & Metrics

Academic Career Length

15Years (2012-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
14 February 2024
Verified
Book Chapter
Closed access
PDF file
RKR. Dinesh Kumar
R. Dinesh Kumar
GPG. Prudhviraj
G. Prudhviraj
KVK. Vijay
K. Vijay
PKP. Senthil Kumar
P. Senthil Kumar
PPPhilipp Plugmann
Philipp Plugmann
In the current scenario, artificial intelligence (AI) and machine learning (ML) models play a major role in classification and prediction in business intelligence and early prediction of diseases. Machine learning plays a vital role in the medical industry, particularly when using medical databases to diagnose diseases. Much research is in progress using these techniques for the early prediction of diseases and to enhance medical diagnostics. In order to treat patients more effectively, predictive analysis with the aid of competent machine learning algorithms makes it possible to detect disease with more accuracy. Furthermore, implementing machine learning algorithms can result in quick and accurate disease prediction. The major goal of this research is to analyze how well different supervised machine learning algorithms perform in predicting COVID-19 using patient data. The purpose of this study is to provide an overview of machine learning techniques including naive Bayes, support vector machine, k-nearest neighbor, decision tree, and random forest that facilitate the categorization and prediction of various diseases. Generally, supervised machine learning prediction algorithms are created by learning a dataset where the label has been previously determined, making it possible to predict the result of instances that have not yet been assigned a label. This work focuses on machine learning algorithms used for disease prediction based on various performance metrics accuracy, F-score, precision, and recall. In general, any classification and prediction in business intelligence or any disease prediction in medical diagnosis is an essential and interesting work in recent years <!----> <!----> <!----> <!---->

1 January 2025 in SSRN Electronic Journal
Verified
Preprint
Open access
PDF file
BFBaseer Fatima
Baseer Fatima
PKP. Senthil Kumar
P. Senthil Kumar
The utilization of Zero-Knowledge Proofs (ZKPs) in blockchain technology enhances privacy while simultaneously preserving transparency. Given that blockchain networks frequently elicit privacy concerns owing to the inherently public nature of transaction data, ZKPs present a viable solution by enabling parties to authenticate transactions without disclosing sensitive information. This study primarily concentrates on zk-SNARKs and zk-STARKs, which represent advanced iterations of ZKPs that enhance both privacy and scalability. By analyzing established blockchain protocols, such as Zcash and Ethereum, this research illustrates that ZKPs can effectively safeguard privacy while also facilitating scalability through mechanisms such as zkrollups, which consolidate multiple transactions into a single proof, thereby alleviating congestion on the blockchain. Additionally, ZKPs enhance the verification efficiency, thereby reducing the computational burden on blockchain networks and promoting expedited transactions. However, challenges such as computational overheads and regulatory hurdles persist, hindering the widespread implementation of ZKPs. Future research endeavors should focus on overcoming these challenges by developing more efficient algorithms and collaborating with regulatory authorities to establish clear guidelines for ZKP-based systems. The potential implications of ZKPs extend beyond blockchain technology, offering substantial advantages to sectors such as finance, healthcare, and identity management, in which secure and confidential transactions are paramount. In summary, although ZKPs possess the capacity to transform privacy within decentralized networks, further advancements are required to fully harness their potential and ensure their extensive adoption.

1 January 2025 in International Journal on Smart Sensing and Intelligent Systems
Verified
Journal Article
Open access
PDF file
KDKusumakumari Daram
Kusumakumari Daram
PKP. Senthil Kumar
P. Senthil Kumar
As a crucial component of the modern, sophisticated Internet backbone, Cloudflare offers exhaustive web performance and security solutions. This research study examines the complex task of balancing high-level security with optimal performance by introducing an artificial intelligence (AI)-driven security orchestration system. The proposed system utilizes sophisticated machine learning (ML) models to rapidly optimize Web Application Firewall (WAF) access rules and detect anomalies in real-time applied scenarios, resulting in adaptable and scalable cyber threat mitigation. The importance of this investigation is highlighted by its ability to increase cyber threat response precision by 92% while decreasing latency by 18%, thus providing uninterrupted user interaction. The societal impact includes enhanced digital reliability and mitigated cybersecurity threats for both businesses and users. The study’s methodology utilizes supervised and unsupervised learning methods to implement effective and well-organized security measures. This innovative approach incorporates AI driven into the Cloudflare ecosystem, enabling resilient content delivery networks (CDNs). Future research initiatives involve exploring advanced deep learning methods to further enhance scalability and performance across distinct digital landscapes.

1 January 2025 in International Journal on Smart Sensing and Intelligent Systems
Verified
Journal Article
Open access
PDF file
SDSai Kiranmai Dornala
Sai Kiranmai Dornala
PKP. Senthil Kumar
P. Senthil Kumar
The rapid growth of digital technologies has led to increasingly sophisticated cyber threats, requiring innovative solutions for instantaneous intrusion detection system, classification, and network traffic analysis. This investigation presents a data-centric deep learning framework that merges advanced security system concepts with an optimized fully connected neural network for behavioral analysis and vulnerability evaluation. The proposed system, Optimized Fully Connected Neural Network for Cyber Threat Detection, incorporates innovative data preprocessing techniques, leverages the Agile Correlation-Based Filter for feature selection, and tackles class imbalance using the artificial minority data oversampling method. Moreover, it bolsters the detection framework with adaptive security system mechanisms to efficiently filter malicious traffic and thwart unauthorized access. Empirical assessment demonstrates the framework’s ability to achieve 96.35% accuracy, exceeding current innovative methods. The proposed model can identify and categorize multiple network intrusions, ensure high accuracy, recall minimizing false positives rate, and enable comprehensive vulnerability assessments. The proposed solution enhances real-time monitoring, decision-making, and threat mitigation capabilities by integrating sophisticated security system techniques. This contribution safeguards digital landscapes, enabling the wider objective of fostering secure, reliable, and sustainable technological advancements that drive economic growth and social progress.

1 January 2025 in International Journal on Smart Sensing and Intelligent Systems
Verified
Journal Article
Open access
PDF file
SRSarojini Rani
Sarojini Rani
PKP. Senthil Kumar
P. Senthil Kumar
The convergence of big data and machine learning (ML) heralds transformative capabilities in healthcare, addressing obstacles in utilizing extensive, sophisticated datasets from electronic health records (EHRs), medical imaging, and wearable technology. This research aims to improve symptomatic accuracy, personalize treatments, and optimize operational efficiency through prognosis ML models while addressing critical concerns such as data security, scalability, and interpretation. By integrating a range of datasets and reconstruct sophisticated algorithms like graph neural networks (GNNs) and reinforcement learning (RL) with healthcare restraints, the research study achieved disease prediction accuracy prodigious 90% and a 25% boost in patient adherence to treatment regimen. Privacy defense methods, including federated learning (FL), ensured compliance with protected health information regulations, while foremost execution showed a 30% improvement in the operational efficiency of the healthcare system. The outcomes emphasize the crucial integration of technological advancements with clinical pertinence, establishing a foundation for secure, scalable, and patient-oriented healthcare system solutions that enhance outcomes and drive standardized improvements.

20 January 2012 in International Journal of Research Studies in Computing
Verified
Journal Article
Closed access
PKP. Senthil Kumar
P. Senthil Kumar
ASArumugam S.
Arumugam S.