Prema Arokia Mary G

EDUCATION

B.E, M.E, Ph.D.,

RESEARCH, TEACHING, or OTHER INTERESTS

Computer Science, Computer Engineering, Artificial Intelligence, Information Systems

Scopus Publications

Comparing Deep Reinforcement Learning Methods for HVAC System Optimization with Forecast Parameters Over Various Time Periods
Prema Arokia Mary G, Hema M S, S Bama, Nageswara Guptha
2025 IEEE International Conference on Interdisciplinary Approaches in Technology and Management for Social Innovation Iatmsi 2025, 2025
Buildings are significant contributors to global energy consumption. Maintaining comfortable indoor temperatures while reducing energy consumption are conflicting objectives. Deep Reinforcement Learning (DRL) is a promising area of research for building Heating, Ventilation and Air Conditioning (HVAC) system optimization. In this study an open-source framework Building Optimization Testing Framework (BOPTEST), which is a virtual testbed that help comparison different control strategies for evaluation of DRL control methods is used. A Proportional-Integral (PI) controller is used to benchmark the DRL methods. A single zone residential building of 192 m<sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> with a radial heating system and a heat pump in a climate zone with high heating requirement with dynamic electricity prices with prices varying every 15 min based on demand is chosen for implementing different control strategies. On comparing Deep Deterministic Policy Gradient (DDPG), Proximal Policy Optimization (PPO) and Twin Delay DDPG (TD3) based DRL controllers and the baseline controller, the DDPG based controller reduced energy consumption by 97.3 % and operating cost by 17.7 % during the peak heating period with reference to baseline method. Then on analyzing the impact of inclusion of forecast parameters occupancy, solar irradiance, and electricity prices over the period 3, 6 and 12 hours in DDPG based controller. The prediction for 3 hours gave the greatest reduction in thermal discomfort of 99.7 % and prediction for 12 hours gave maximum reduction in cost by 30.4 % but resulted in only 82% reduction in thermal comfort when compared with baseline method indicating that longer prediction horizon is not necessarily results in better performance.
Wild Animal Detection System
Prema Arokia Mary G, Nithesh P S, Nanthini V, Thebiksha G V
2nd International Conference on Advancements in Electrical Electronics Communication Computing and Automation Icaeca 2023, 2023
In forests and agricultural fields, human lives are in danger due to the conflict between human being and animals, which is a severe problem. As a result, individuals suffer losing their possessions, cattle, crops, and occasionally even their lives. This is due to the animals that cross roads without any conscious to the vehicles that pass the roads. Therefore, it is necessary to classify the different types of animals that are been found at roadsides. Manual identification of animals crossing roads are a difficult task. To overcome this difficulty the paper suggests a classification technique using DCNN deep learning technique and ResNet152V2 algorithm, which classifies the type of animal based on their images. After classification a notification is sent using a Wi-Fi enabled IOT device through a web application software called Blynk, which controls Arduino board with smart phone via internet. In addition to this, a warning message is displayed in the LCD display, for those who don’t have the application installed, which can be fixed at the Forest roads.
Detection of Parkinson’s Disease with Multiple Feature Extraction Models and Darknet CNN Classification
G. Prema Arokia Mary, N. Suganthi
Computer Systems Science and Engineering, 2022
Parkinson’s disease (PD) is a neurodegenerative disease in the central nervous system. Recently, more researches have been conducted in the determination of PD prediction which is really a challenging task. Due to the disorders in the central nervous system, the syndromes like off sleep, speech disorders, olfactory and autonomic dysfunction, sensory disorder symptoms will occur. The earliest diagnosing of PD is very challenging among the doctors community. There are techniques that are available in order to predict PD using symptoms and disorder measurement. It helps to save a million lives of future by early prediction. In this article, the early diagnosing of PD using machine learning techniques with feature selection is carried out. In the first stage, the data preprocessing is used for the preparation of Parkinson’s disease data. In the second stage, MFEA is used for extracting features. In the third stage, the feature selection is performed using multiple feature input with a principal component analysis (PCA) algorithm. Finally, a Darknet Convolutional Neural Network (DNetCNN) is used to classify the PD patients. The main advantage of using PCA- DNetCNN is that, it provides the best classification in the image dataset using YOLO. In addition to that, the results of various existing methods are compared and the proposed DNetCNN proves better accuracy, performance in detecting the PD at the initial stages. DNetCNN achieves 97.5 % of accuracy in detecting PD as early. Besides, the other performance metrics are compared in the result evaluation and it is proved that the proposed model outperforms all the other existing models.
Prediction of Parkinson Disease using Autoencoder Convolutional Neural Networks
M S Hema, R Maheshprabhu, M Nageswara Guptha, Prema Arokia Mary G, Aditi Sharma
International Interdisciplinary Humanitarian Conference for Sustainability Iihc 2022 Proceedings, 2022
Parkinson's Disease (PD) diagnosis is a challenging task for doctors because of the non-availability of separate testing and prediction methodology. PD is identified through various clinical tests, symptoms, and repeated clinical trials. Diagnosis of PD at an early stage is essential to improving patients' quality of life. An autoencoder feature extraction methodology is proposed for prediction of PD using Convolutional Neural Networks (CNNs). The feature extraction and de-noising of input data are performed using an autoencoder. CNN is used for classification and prediction. This algorithm has three layers: a convolutional layer, a pooling layer, and a dense layer. Feature extraction and image segmentation are automatically performed in the convolutional layer. A pooling layer performs downsizing. Finally, the dense layer is used for classification. The PPMI dataset is used for experimentation. The performance assessment is based on accuracy, precision, recall, and F1 measures.
A Machine Learning Approach to Detect Parkinson Disease Using Speech Signals
Prema Arokia Mary G, Hema M S, Nageswara Guptha M, Maheshprabhu R, Naveen Vignesh G, Aditi Sharma
International Interdisciplinary Humanitarian Conference for Sustainability Iihc 2022 Proceedings, 2022
One of the leading neuro-degenerative disorders is Parkinson's disease (PD), which affects a large number of people worldwide. Diagnostics of PD are complex because a detailed clinical review of the patient's medical record is necessary. One of the characteristic and predictive indicators of this disease is voice. Almost all such patients experience at least some vocal degeneration; therefore voice data is an effective way to diagnose the disease. This paper presents a method for analyzing voice impairments by means of a system design in people who have PD and in people who are not afflicted with the disease. Various methodologies including Artificial Neural Networks, Logistic Regression, Decision Trees, SVM, and Random Forests were applied in this study. To test how well these algorithms can discriminate between PD and healthy subjects, the best accuracy predictor model can be determined out of this implemented model. With an accuracy of 88.81%, neural network ended up being the best algorithm among all of the algorithms used during this study.
Sentimental Analysis of Twitter Data using Machine Learning Algorithms
G Prema Arokia Mary, M S Hema, R Maheshprabhu, M Nageswara Guptha
2021 International Conference on Forensics Analytics Big Data Security Fabs 2021, 2021
As social media is becoming a necessity for communication, a lot of data is available on this platform, which could be helpful for analysis. At a certain time, we use Twitter to tweet almost about similar topics, different emotions. In this article, sentimental analysis is proposed to get an idea about what people have in their minds or get people’s emotions. It segregates every tweet to its appropriate emotion. The emotion might be either positive or negative. The proposed methodology has two steps, namely preprocessing and classification. The corpus is created after all necessary preprocessing. The classification algorithms such as Logistic Regression, Linear SVC, Random Forest Classifier, Bernoulli NB, Decision Tree Classifier, Voting Classifier, and KNN Classifier are used for classification. Twitter 2020 and 2021 data has been taken for experimentation. The performance of Linear SVC shows a higher accuracy on training data, and Linear Regression shows higher accuracy in testing data.
A recapitalization on crypto jacking and end to end analysis of ransomware attacks
, G. Prema Arokia Mary, N. Suganthi, , M.S. Hema, and
International Journal of Engineering and Advanced Technology, 2019
The recent trend of today’s digital media is the usage of poisoned website to mine crypto currencies, these currencies are alternatives to traditional currencies which work based on decentralization, bit coin was the first currency to be establish in this way, crypto currencies are protected with block chain which can be simplified as growing chain. This block chain is managed by peer to peer network, based upon this blockchain network crypto jacking takes place, and hence cryptojacking is mining of one's digital currencies without their knowledge, hackers find cryptojacking more profitable because they are a lotcheaper and safe than compared to other digital thefts. Tracking and finding the cause of theft becomes very hard in this method because mining kits can be purchased at a very cheap cost. There are primarily two methods to be followed to get to the computer and to perform cryptojacking, one is to run a infected code on the host computer and the other is to make the user click the content with threat but widely both will be used for increased profit outcome. In this paper an overview of crypto currencies, method of decentralization, various mining techniques followed and different types of cybercrimes prevalent are discussed.
Information system for performance improvement of small and medium scale enterprises
International Journal of Recent Technology and Engineering, 2018

RECENT SCHOLAR PUBLICATIONS

Comparing Deep Reinforcement Learning Methods for HVAC System Optimization with Forecast Parameters Over Various Time Periods
PA Mary G, MS Hema, S Bama, N Guptha
2025 IEEE International Conference on Interdisciplinary Approaches in … , 2025
2025
Citations: 2
Machine Learning-Based Techniques for Predictive Diagnostics in Healthcare
S Sathyavathi, RK Kavitha, GPA Mary, KR Baskaran
Cybersecurity and Data Science Innovations for Sustainable Development of … , 2025
2025
Citations: 1
Wild animal detection system
PS Nithesh, V Nanthini, GV Thebiksha
2023 2nd international conference on advancements in electrical, electronics … , 2023
2023
Citations: 8
A machine learning approach to detect Parkinson disease using speech signals
MS Hema, N Guptha, R Maheshprabhu, A Sharma
2022 International Interdisciplinary Humanitarian Conference for … , 2022
2022
Citations: 1
Prediction of parkinson disease using autoencoder convolutional neural networks
MS Hema, R Maheshprabhu, MN Guptha, A Sharma
2022 International Interdisciplinary Humanitarian Conference for … , 2022
2022
Citations: 5
Detection of Parkinson's Disease with Multiple Feature Extraction Models and Darknet CNN Classification.
G Mary, N Suganthi
Computer Systems Science & Engineering 43 (1) , 2022
2022
Citations: 27
Parkinson Disease Prediction and Drug Personalization Using Machine Learning Techniques
MS Hema, K Meena, R Maheshprabhu, MN Guptha, GPA Mary
Industrial Internet of Things, 57-82 , 2022
2022
Sentimental analysis of twitter data using machine learning algorithms
GPA Mary, MS Hema, R Maheshprabhu, MN Guptha
2021 International Conference on Forensics, Analytics, Big Data, Security … , 2021
2021
Citations: 39
Early prediction of Parkinson’s disease from brain MRI images using convolutional neural network
G Prema Arokia Mary, N Suganthi, MS Hema
Journal of Medical Imaging and Health Informatics 11 (12), 3103-3109 , 2021
2021
Citations: 6
Various approaches of detecting parkinson disease using speech signals and drawing pattern
GPA Mary, GN Vignesh, N Suganthi
Naveen Vignesh, and N. Suganthi.“Various approaches of detecting parkinson … , 2021
2021
Citations: 2
Predicting metamorphic changes in parkinson’s disease patients using machine learning algorithms
MGP AROKIA
BIOSCIENCE 13 (11), 147-152 , 2020
2020
Citations: 6
Information system for performance improvement of small and medium scale enterprises
R Maheshprabhu, MS Hema, GPA Mary
Int. J. Recent Technol. Eng 7 (4), 407-411 , 2009
2009
Citations: 1

MOST CITED SCHOLAR PUBLICATIONS

Sentimental analysis of twitter data using machine learning algorithms
GPA Mary, MS Hema, R Maheshprabhu, MN Guptha
2021 International Conference on Forensics, Analytics, Big Data, Security … , 2021
2021
Citations: 39
Detection of Parkinson's Disease with Multiple Feature Extraction Models and Darknet CNN Classification.
G Mary, N Suganthi
Computer Systems Science & Engineering 43 (1) , 2022
2022
Citations: 27
Wild animal detection system
PS Nithesh, V Nanthini, GV Thebiksha
2023 2nd international conference on advancements in electrical, electronics … , 2023
2023
Citations: 8
Early prediction of Parkinson’s disease from brain MRI images using convolutional neural network
G Prema Arokia Mary, N Suganthi, MS Hema
Journal of Medical Imaging and Health Informatics 11 (12), 3103-3109 , 2021
2021
Citations: 6
Predicting metamorphic changes in parkinson’s disease patients using machine learning algorithms
MGP AROKIA
BIOSCIENCE 13 (11), 147-152 , 2020
2020
Citations: 6
Prediction of parkinson disease using autoencoder convolutional neural networks
MS Hema, R Maheshprabhu, MN Guptha, A Sharma
2022 International Interdisciplinary Humanitarian Conference for … , 2022
2022
Citations: 5
Comparing Deep Reinforcement Learning Methods for HVAC System Optimization with Forecast Parameters Over Various Time Periods
PA Mary G, MS Hema, S Bama, N Guptha
2025 IEEE International Conference on Interdisciplinary Approaches in … , 2025
2025
Citations: 2
Various approaches of detecting parkinson disease using speech signals and drawing pattern
GPA Mary, GN Vignesh, N Suganthi
Naveen Vignesh, and N. Suganthi.“Various approaches of detecting parkinson … , 2021
2021
Citations: 2
Machine Learning-Based Techniques for Predictive Diagnostics in Healthcare
S Sathyavathi, RK Kavitha, GPA Mary, KR Baskaran
Cybersecurity and Data Science Innovations for Sustainable Development of … , 2025
2025
Citations: 1
A machine learning approach to detect Parkinson disease using speech signals
MS Hema, N Guptha, R Maheshprabhu, A Sharma
2022 International Interdisciplinary Humanitarian Conference for … , 2022
2022
Citations: 1
Information system for performance improvement of small and medium scale enterprises
R Maheshprabhu, MS Hema, GPA Mary
Int. J. Recent Technol. Eng 7 (4), 407-411 , 2009
2009
Citations: 1
Parkinson Disease Prediction and Drug Personalization Using Machine Learning Techniques
MS Hema, K Meena, R Maheshprabhu, MN Guptha, GPA Mary
Industrial Internet of Things, 57-82 , 2022
2022

Publications

Prema Arokia Mary G, Gowthami K, Aishwarya S and Anusha CN, 2025, “Intelligent Online Chatbot System for Railway Ticketing: Enhancing Efficiency and User Experience”, IJCRT, Volume 13, Issue 1 January 2025, ISSN: 2320-2882.
Bhavana DK, Chaitra TR, Dinakar S and Prema Arokia Mary G, 2025, “SAR Image Colorization Compressive Insight Using Deep Learning”, IJRAR January 2025, Volume 12, Issue 1, E-ISSN 2348-1269.
Prema Arokia Mary G, Abhishek Y S, Dharshan R and Hariharan J 2025, “Intelligent Traffic Systems: Enhancing Emergency Vehicle Mobility with Real-Time Data”, IJNTI, Volume 3, Issue 1 January 2025 | ISSN: 2984-908X.
Abhishek Shetty B, Gagan Nayaka K.M, Dhanush B M and Prema Arokia Mary G, 2025, “Effective Career Building Using Machine Learning Algorithm”, IJCRT, Volume 13, Issue 1 January 2025, ISSN: 2320-2882.
Prema Arokia Mary G , Sathyavathi S , Dheesiga A , Krityaa S , Prithiv V, 2023, "Cryptocurrency price prediction", JETNR, Volume 1, Issue 12, pp. 204-211.
Shakthi Sabarinath S.V and Prema Arokia Mary, G, 2023, "A Review on Virtual Testbed Frameworks for Implementation of Various HVAC Control Strategies", IJARSCT, Volume 3, Issue 1, ISSN online 2581-9429.
Prema Arokia Mary G, Hariharan D, Saisaran K & HaniRahim 2022, "A Review on Machine and Deep Learning Classification Techniques for Monkeypox Detection", IJRAR, Volume 9, Issue 4, e-ISSN:2348-1269.
Prema Arokia Mary, G & Suganthi, N 2022, "Detection of Parkinson's Disease with Multiple Feature Extraction Models and

RESEARCH OUTPUTS (PATENTS, SOFTWARE, PUBLICATIONS, PRODUCTS)

Autonomous crop irrigation Robot Design No. : 449832-001, date of issue : 21.06.25
AI Enabled Device for Detection of Neurological Disorders Design No. : 452949-001, date of issue : 30.07.25

CONSULTANCY

Title of Project: Age Determination and Analysis using Orthopantomogram Images
Consultancy Name: Vihana Dental Care, Coimbatore, Tamlnadu.
Amount: Rs. 58,000
Duration : 3 months
Status : Completed
Title of Project: Invoice and Inventory Management System
Consultancy Name: Arul & Co, Coimbatore, Tamlnadu.
Amount: Rs. 50,000
Duration : 4 months
Status : Completed

Title of Project: Tea Leaf Classification
Consultancy Name: Rani Traders, Coimbatore, Tamlnadu.
Amount: Rs. 1,40,000
Duration : 6 months
Status : Completed

Industry, Institute, or Organisation Collaboration

S. Sathyavathi, R. K. Kavitha, G. Prema Arokia Mary, K. R. Baskaran (2025), Machine Learning-Based Techniques for Predictive Diagnostics in Healthcare, In Cybersecurity and Data Science Innovations for Sustainable Development of HEICC, Chapter 4, CRC Press, eBook ISBN9781032711300.
Hema, M. S., Meena, K., Maheshprabhu, R., Guptha, M. N., & Mary, G. P. A. (2022). Parkinson Disease Prediction and Drug Personalization Using Machine Learning Techniques. In Industrial Internet of Things (pp. 57-82). CRC Press.