IoT-Driven Regressive Ensemble LSTM Framework for Scalable Paddy Yield Optimization P Rohini, S Mohana 2nd Asian Conference on Intelligent Technologies Acoit 2025, 2025 Paddy production needs proper resource management and proper prediction in yield to guarantee sustainable food production. The current smart farming technologies are typically restricted in the area of scalability, real-time decision-making, and heterogeneous sensor information scraping. The current paper would suggest a data-driven approach, otherwise defining a smart farming system which utilises IoT sensors, a predictive modelling framework, alongside optimisation solutions to enhance resource efficiency and productivity of paddy. The framework is based on the Kaggle Smart Agriculture Dataset that offers soil moisture, temperature, humidity, and pH data and uses machine learning and sophisticated data preprocessing and feature engineering methods. It is experimentally proven that the offered methodology gets a Root Mean Squared Error (RMSE) equal to 3.92, R2 equal to 0.95, and Water-Use Efficiency (WUE) equal to 2.95 kg/m3, which is lower than baseline models such as Linear Regression, Random Forest, and Support Vector Regression. The Scalability analysis represents reduced training (30.2 s) with less prediction latency (9.1 ms). Their findings indicate that the framework is feasible, scalable, and has the capacity to provide real-time actionable information to control the accuracy of paddy farming to allow enhanced yield, cost-effective resource use, and an environmentally sustainable agricultural activity.
Enhancing GAN Resilience through Adversarial-Aware Architecture and Latent-Space Defense R. Roshan Joshua, S. Mohana Proceedings 2025 International Conference on Recent Innovation in Science Engineering and Technology Icriset 2025, 2025 Generative Adversarial Networks (GANs) have emerged as powerful generative models capable of producing highly realistic data. Nonetheless, they are susceptible to adversarial attacks (i.e., perturbation in inputs, manipulation in latent space, and disrupting the discriminator) that compromise the predictability and integrity of their products. This paper introduces a powerful GAN architecture that is resistant to different adversarial attacks based on a multi-layered defense to consist of adversarial training, spectral normalization, latent space denoising Autoencoders, ensemble discriminators. The model combines StyleGAN-like hybrid architecture, with self-supervised learning and attention-directed noise suppression modules. Basically, the work was done on the MNIST dataset with main adversarial attacks being FGSM, PGD, and latent noise injection and model inversion. Robustness was determined using the evaluation metrics like Fréchet Inception Distance (FID), Inception Score (IS), Adversarial Success Rate (ASR), and Latent Space Consistency Score (LSCS). On an average, the FID of the proposed model was 16.23 under adversarial settings, where it was 25.65 on the baseline model. It also had a high score of Inception of 8.74, which was better than other settings. The Adversarial Success Rate was decreased by more than 40% and the LSCS stayed well above 0.80 suggesting sound latent space mapping. Besides substantial advancement on adversarial resistance, this study offers a basis on the use of safe GANs in privacy-sensitive tasks like synthesizing biometrics, medical imaging, and data generation.
Clustering-assisted privacy perseveration model for data mining S. Mohana, T.M. Nithya, Sardar Khan Nikkath Bushra, Ramakrishnan Vasanthi, K.S. Guruprakash, Sudha Rajesh International Journal of Ad Hoc and Ubiquitous Computing, 2024 Data mining techniques are used to examine the data in order to reveal hidden patterns. While preserving the privacy of individual records, privacy preserving data mining (PPDM) technology enables us to extract meaningful information from massive volumes of data. This paper proposes the two stages of the privacy preservation are data sanitisation and data restoration. The clustering, key generation and key pruning elements of the data sanitisation process are all carried out in a distributed environment. The key is pruned using the deep maxout model to make any last modifications after being formed using the hybrid optimisation, Tasmanian updated Namib beetle optimisation (TUNBO), which combines the Tasmanian devil optimisation (TDO) and Namib beetle optimisation (NBO) algorithms. In the data restoration step, which is the reverse of sanitisation, the sanitised data is also retrieved. In the meantime, the correlation coefficients are 85.64%, 88.76%, 75.94%, 74.67%, and 82.67%, compared to other models.
Hybrid Grasshopper and Improved Cat Swarm Optimization Algorithm-based clustering for guaranteeing energy stability and network lifetime in WSN Palaniappan Rajarajeswari, Chandrasekaran Shyamala, Shivashankar Mohana International Journal of Communication Systems, 2023 Summary Wireless sensor networks (WSNs) plays an indispensable role in the human life by supporting a diversified number of applications that includes military, environment monitoring, manufacturing, education, agriculture, etc. However, the sensor node batteries cannot be replaced under its deployment in an unattended or remote area due to their wireless existence. Cluster‐based routing is significant in handling the issue of energy stability and network lifetime. The meta‐heuristic algorithms‐based cluster head (CH) selection is determined to be highly promising for attaining the objective of CH selection that results in acquiring an optimal network performance. In this paper, a Hybrid Grasshopper and Improved Cat Swarm Optimization Algorithm (HGICSOA)‐based clustering scheme is proposed for attaining potential CH selection and guarantee significant sink mobility‐based data transmission. The capability of GHOA that controls the rate of exploitation and exploration degree is utilized for CH selection. It specifically adopted OBL‐based GHOA for optimal CH selection based on the objective function, which is formulated using node density, residual energy, and distance between sensor node and sink. It incorporated new CSOA for mobility‐based data transmission for increasing population diversity. It also utilized the benefits of ICSOA with a predominant local search strategy for achieving better sink mobility‐based data transmission. Simulation and statistical results confirmed that the proposed HGICSOA is better in attaining maximum energy stability by 17.21% and improved network lifetime by 23.82%, compared to the benchmarked schemes used for investigation. Moreover, the prevention rate of worst sensor nodes selected as CH is improved by 21.38%, better than baseline approaches.
Multi-Tier Kernel for Disease Prediction using Texture Analysis with MR Images M. Mohan, Anuradha Patil, S. Mohana, P. Subhashini, Sumit Kushwaha, S Manikandaprabu Pandian International Conference on Edge Computing and Applications Icecaa 2022 Proceedings, 2022 Denoising magnetic resonance images are traditionally done individually, introducing undesired artefacts like blurring. To solve this issue, this paper offers a unique adaptive interpolation architecture that simultaneously allows for image data augmentation, noise removal, and detail augmentation. The multi-tier kernel (MTK) algorithm adjusts the extrapolation weights based on mathematical features in magnetic resonance (MR) data. The MTK weight matrix is then adaptively sharpened, and a Rician bias corrective is used to reduce Rician noise and improve small details. After processing, the noise eliminates the bias produced by the asymmetric sources. The adaptive MTK, in this way, extends the zero ordering estimating methodology to higher regression power facilitating edge maintenance and detail restoration. Investigation outcomes using genuine and MR images (with/without noise) evidenced that the algorithm delivered good restoration outcomes than conventional deep-learning-based approaches but with fewer requirements and calculation burden.
Privacy preservation of data using modified rider optimization algorithm: Optimal data sanitization and restoration model Mohana Shivashankar, Sahaaya Arul Mary Expert Systems, 2021 Data preservation is the mechanism of protecting and safeguarding the confidentiality and integrity of data. Data stored in huge databases may contain metadata, elements that may be imprecise and unstable, It may include sensitive data, personal profiles and so on, which is vulnerable to third parties such as hackers or attackers. They may misuse the data and as a consequence of this the confidentiality and privacy of the data gets lost. There is a need to conserve the data and make it available for reuse when needed. Hence, it needs a proficient method to maintain and protect individuals' data privacy regarding confidentiality and reliability. This paper intends to develop an advanced model for privacy preservation of huge data with the accomplishment of two stages, namely data sanitization and data restoration. Data sanitization process preserves the safety of sensitive data stored in huge databases, by means of hiding those sensitive data from unauthorized users. Data restoration is the process of recovering or restoring of data that is sanitized at the sender side. Concerning the secrecy, there is a need for an optimal key to hide the sensitive data at sender as well as receiver side. Subsequent to the data sanitization, it requires the same key to restore the sanitized data. Thus, the optimal key generation plays a vital role to maintain privacy preservation. In order to choose an optimal key, a modified Rider optimization Algorithm (ROA) named as Randomized ROA (RROA) model is implemented in this work. Furthermore, the efficiency of the proposed work is compared over the state‐of‐the‐arts models by concerning the sanitization as well as restoration efficiency.
Heuristics for privacy preserving data mining: An evaluation S. Mohana, S. A. Sahaaya Arul Mary 2017 International Conference on Algorithms Methodology Models and Applications in Emerging Technologies Icammaet 2017, 2017 Availability of information in profusion in the internet and databases is common knowledge. It has to be viewed in the backdrop of chances for disclosure of such information by a third party. Privacy Preserving Data Mining (PPDM) is in use for maintaining the privacy of individuals. Numerous updated methods are available for the purpose. Evolutionary Algorithms (EA's) are able to provide effective solutions for real-world optimization problems. They find use in business practice too. This work has a proposal for the implementation of an EA using K-Anonymization; particle swam optimization (PSO), Ant colony optimization (ACO) and a Genetic Algorithm (GA). We herein propose Genetic algorithm and particle swam optimization work with the same data. The use of generalization of the original dataset is meant for achieving K-anonymity. A collection of people called “chromosomes” frame the populace which shows an aggregate solution for a characterized issue in the proposed GA. The achievement of good accuracy is obtained by GA optimization, recall and precision in comparison with K-Anonymization, PSO and ACO methods.
A comparitive framework for feature selction in privacy preserving data mining techniques using PSO and K-anonumization Iioab Journal, 2016
RECENT SCHOLAR PUBLICATIONS
CNN with ISNet-based healthcare monitoring system for heart disease prediction M Ambika, S Mohana, R Sheeba Biomedical Signal Processing and Control 110, 108339 , 2025 2025.0 Citations: 2
IoT-Driven Regressive Ensemble LSTM Framework for Scalable Paddy Yield Optimization P Rohini, S Mohana 2025 2nd Asian Conference on Intelligent Technologies (ACOIT), 1-6 , 2025 2025.0
Hybrid deep architecture for software defect prediction with improved feature set C Shyamala, S Mohana, M Ambika, K Gomathi Multimedia Tools and Applications 83 (31), 76551-76586 , 2024 2024.0 Citations: 9
Clustering-assisted privacy perseveration model for data mining S Mohana, TM Nithya, SKN Bushra, R Vasanthi, KS Guruprakash, ... International Journal of Ad Hoc and Ubiquitous Computing 47 (2), 108-125 , 2024 2024.0 Citations: 1
Preserving sensitive data with deep learning assisted sanitisation process S Mohana, C Shyamala, ES Rani, M Ambika Journal of Experimental & Theoretical Artificial Intelligence 35 (4), 589-616 , 2023 2023.0 Citations: 1
Hybrid Grasshopper and Improved Cat Swarm Optimization Algorithm‐based clustering for guaranteeing energy stability and network lifetime in WSN P Rajarajeswari, C Shyamala, S Mohana International Journal of Communication Systems 36 (6), e5444 , 2023 2023.0 Citations: 7
Geographical Information System-Aided Landmark Recognition System Using Machine Learning SA Sahaaya Arul Mary, LK Narayanan, S Mohana, R Senthamil Selvi, ... Computer Networks and Inventive Communication Technologies: Proceedings of … , 2022 2022.0
Geographical Information System-Aided Landmark Recognition System Using SASA Mary, LK Narayanan, S Mohana Computer Networks and Inventive Communication Technologies: Proceedings of … , 2022 2022.0
Privacy preservation of data using modified rider optimization algorithm: Optimal data sanitization and restoration model M Shivashankar, SA Mary Expert Systems 38 (3), e12663 , 2021 2021.0 Citations: 18
A survey-plant leaf disease in horticulture using classification algorithm N Ramya, S Mohana Annals of the Romanian Society for Cell Biology 25 (4), 10367-10372 , 2021 2021.0 Citations: 5
Detection of unwanted messages and fraudulent user identification on social network R Senthamilselvi, S Mohana, SU Maheswari, R Pushpalakshmi, ... Annals of the Romanian Society for Cell Biology 25 (2), 1041-1048 , 2021 2021.0 Citations: 1
Fuzzy based Reliable Data Gathering Routing in Mobile Ad hoc Networks KV Mohana S Test Engineering and Management 83 (March-April 2020), 14539-14544 , 2020 2020.0
Heuristics for privacy preserving data mining: An evaluation S Mohana, SASA Mary 2017 International Conference on Algorithms, Methodology, Models and … , 2017 2017.0 Citations: 14
A comparitive framework for feature selction in privacy preserving data mining techniques using pso and k-anonumization S Mohana, SA Sahaaya, A Mary Iioab Journal 7 (9), 804-811 , 2016 2016.0 Citations: 18
Preserving privacy in health care information: a memetic approach S Mohana, SA Mary Journal of Medical Imaging and Health Informatics 6 (3), 779-783 , 2016 2016.0 Citations: 8
Sentiment Analysis For Two Sides Of Reviews Using Dual Prediction Algorithm GG Ranjitham, S Mohana, B Vinothini All Rights Reserved , 2016 2016.0 Citations: 3
A survey on privacy preservation recent approaches and techniques K Dhivakar, S Mohana International Journal of Innovative Research in Computer and Communication … , 2014 2014.0 Citations: 11
A Survey on Various Candidate Generator Methods for Efficient String Transformation MP Malarvizhi, MS Mohana COMPUSOFT, An International Journal of Advanced Computer Technology 3 (2 … , 2014 2014.0 Citations: 1
Data preserving techniques for collaborative data publishing R Indhumathi, S Mohana Cancer 5 (6), 7 , 2013 2013.0 Citations: 5
SafeShe: IoT‑Enabled Women Security and Well‑Being A Joyce, S Mohanalakshmi, P Rohini, S Mohana Proceedings Copyright 155, 160 , 0
MOST CITED SCHOLAR PUBLICATIONS
Privacy preservation of data using modified rider optimization algorithm: Optimal data sanitization and restoration model M Shivashankar, SA Mary Expert Systems 38 (3), e12663 , 2021 2021.0 Citations: 18
A comparitive framework for feature selction in privacy preserving data mining techniques using pso and k-anonumization S Mohana, SA Sahaaya, A Mary Iioab Journal 7 (9), 804-811 , 2016 2016.0 Citations: 18
Heuristics for privacy preserving data mining: An evaluation S Mohana, SASA Mary 2017 International Conference on Algorithms, Methodology, Models and … , 2017 2017.0 Citations: 14
A survey on privacy preservation recent approaches and techniques K Dhivakar, S Mohana International Journal of Innovative Research in Computer and Communication … , 2014 2014.0 Citations: 11
Hybrid deep architecture for software defect prediction with improved feature set C Shyamala, S Mohana, M Ambika, K Gomathi Multimedia Tools and Applications 83 (31), 76551-76586 , 2024 2024.0 Citations: 9
Preserving privacy in health care information: a memetic approach S Mohana, SA Mary Journal of Medical Imaging and Health Informatics 6 (3), 779-783 , 2016 2016.0 Citations: 8
Hybrid Grasshopper and Improved Cat Swarm Optimization Algorithm‐based clustering for guaranteeing energy stability and network lifetime in WSN P Rajarajeswari, C Shyamala, S Mohana International Journal of Communication Systems 36 (6), e5444 , 2023 2023.0 Citations: 7
A survey-plant leaf disease in horticulture using classification algorithm N Ramya, S Mohana Annals of the Romanian Society for Cell Biology 25 (4), 10367-10372 , 2021 2021.0 Citations: 5
Data preserving techniques for collaborative data publishing R Indhumathi, S Mohana Cancer 5 (6), 7 , 2013 2013.0 Citations: 5
Sentiment Analysis For Two Sides Of Reviews Using Dual Prediction Algorithm GG Ranjitham, S Mohana, B Vinothini All Rights Reserved , 2016 2016.0 Citations: 3
CNN with ISNet-based healthcare monitoring system for heart disease prediction M Ambika, S Mohana, R Sheeba Biomedical Signal Processing and Control 110, 108339 , 2025 2025.0 Citations: 2
Clustering-assisted privacy perseveration model for data mining S Mohana, TM Nithya, SKN Bushra, R Vasanthi, KS Guruprakash, ... International Journal of Ad Hoc and Ubiquitous Computing 47 (2), 108-125 , 2024 2024.0 Citations: 1
Preserving sensitive data with deep learning assisted sanitisation process S Mohana, C Shyamala, ES Rani, M Ambika Journal of Experimental & Theoretical Artificial Intelligence 35 (4), 589-616 , 2023 2023.0 Citations: 1
Detection of unwanted messages and fraudulent user identification on social network R Senthamilselvi, S Mohana, SU Maheswari, R Pushpalakshmi, ... Annals of the Romanian Society for Cell Biology 25 (2), 1041-1048 , 2021 2021.0 Citations: 1
A Survey on Various Candidate Generator Methods for Efficient String Transformation MP Malarvizhi, MS Mohana COMPUSOFT, An International Journal of Advanced Computer Technology 3 (2 … , 2014 2014.0 Citations: 1
IoT-Driven Regressive Ensemble LSTM Framework for Scalable Paddy Yield Optimization P Rohini, S Mohana 2025 2nd Asian Conference on Intelligent Technologies (ACOIT), 1-6 , 2025 2025.0
Geographical Information System-Aided Landmark Recognition System Using Machine Learning SA Sahaaya Arul Mary, LK Narayanan, S Mohana, R Senthamil Selvi, ... Computer Networks and Inventive Communication Technologies: Proceedings of … , 2022 2022.0
Geographical Information System-Aided Landmark Recognition System Using SASA Mary, LK Narayanan, S Mohana Computer Networks and Inventive Communication Technologies: Proceedings of … , 2022 2022.0
Fuzzy based Reliable Data Gathering Routing in Mobile Ad hoc Networks KV Mohana S Test Engineering and Management 83 (March-April 2020), 14539-14544 , 2020 2020.0
SafeShe: IoT‑Enabled Women Security and Well‑Being A Joyce, S Mohanalakshmi, P Rohini, S Mohana Proceedings Copyright 155, 160 , 0
