Claim Missing Document
Check
Articles

Found 12 Documents
Search

Efficient Fruit Grading and Selection System Leveraging Computer Vision and Machine Learning Dewi, Deshinta Arrova; Kurniawan, Tri Basuki; Thinakaran, Rajermani; Batumalay, Malathy; Habib, Shabana; Islam, Muhammad
Journal of Applied Data Sciences Vol 5, No 4: DECEMBER 2024
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jads.v5i4.443

Abstract

Automated fruit grading is crucial to overcoming the time and accuracy challenges posed by manual methods, which are often limited by subjective human judgment. This study introduces an intelligent grading system leveraging computer vision and AI to improve speed and consistency in assessing fruit quality. Using high-resolution imaging and advanced feature extraction, including grayscale processing, binarization, and enhancement, the system achieves non-destructive, efficient sorting for fruits like apples, bananas, and oranges. Grayscale processing reduces image complexity while preserving essential details, binarization isolates the fruit from its background, and enhancement highlights critical features. Notably, the Edge Pixel method proved most effective, achieving 79.20% accuracy in grading, while the Grayscale Pixel method reached 93.94% accuracy for fruit types. Edge Pixel also achieved 80.32% in differentiating grading types, showcasing its ability to capture essential shapes and edges. Fruits are classified into four grades: Grade_01 (highest quality), Grade_02 (minor imperfections), Grade_03 (notable defects but consumable), and Grade_04 (unfit for consumption). A specialized dataset supports model training, ensuring practical real-world application. The study concludes that this automated system offers significant improvements over traditional grading, providing a scalable, objective, and reliable solution for the agricultural sector, ultimately enhancing productivity and quality assurance.
Recognizing Fake Documents by Instance-Based ML Algorithm Tuning with Neighborhood Size S., Prakash; B., Kalaiselvi; K., Sivachandar; Batumalay, Malathy
Journal of Applied Data Sciences Vol 6, No 2: MAY 2025
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jads.v6i2.654

Abstract

The primary objective of this research paper is to classify spam SMS messages for scamming threats as soon as they are received on a device. The study focuses on evaluating the performance of K-Nearest Neighbors (KNN) classifiers with different neighborhood sizes to determine the most effective machine learning technique for improving accuracy and predictions in SMS spam detection. SMS is a short text messages service that permits mobile phone users to exchange messages.  In today’s world, people are so much tending towards mobile phones and it has become easy to spread spam content through them. One can easily access any person’s details through these social networking websites. No information which is shared and stored in the device is not secure. Numerous anti-spam systems have been developed. In this paper, we compare the classification results against spam SMS data to estimate the effectiveness of the KNN classifiers at different k levels and the comparisons shown. An effective method of classifying spam SMS, based on the metrics like F-measures, Precision, and recall score is recognized from the experiment results. The best performance was achieved with K = 4, where the classifier provided a high accuracy of 94.78% and strong results across all key performance metrics. The research highlights that feature selection plays a crucial role in improving classification efficiency by eliminating irrelevant or redundant features. Although KNN is a simple and effective approach, its scalability and real-time processing limitations suggest that future work should explore deep learning, ensemble models, or heuristic-based optimization for further improvements and support process innovation.
Design of multiband antenna for full screen smartphone using ANSYS HFSS R., Karthick Manoj; K., Suresh Kumar; T., Ananth kumar; R., Nishanth; Joseph, Abin John; Batumalay, Malathy
Journal of Applied Data Sciences Vol 6, No 2: MAY 2025
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jads.v6i2.718

Abstract

The current scenario and the global trend are rapidly growing in terms of technology and connected through communication. The term communication is an extensive link and has an enormous number of inventions and technologies booming in this field. In mobile communication, numerous antennas are worked and fabricated in mobile phones. Here, in this paper, we focus on communication through mobile phones. It deals with simulating a multi-band slotted microstrip patch antenna for mobile phones with 6 GHz as the operating frequency using FR4 substrate material. This proposed antenna has resonating frequencies such as 2.732 GHz, 3.311 GHz,4.792 GHz,5.373 GHz, 6.462 GHz, 7.476 GHz, and 9.156 GHz. The parameters to be performed are VSWR, gain, directivity, return loss, and radiation pattern with the inputs as dimen-sion and frequency. The FR4 material might be used as the base substrate because flexi-bility is better, easily used for thin substrates and patch antennas, and the cost is meagre. The simulation is performed with the help of the ANSYS HFSS tool. When implemented in real-time in smartphones, the simulated output will be compatible in means of low power consumption and utilizes low area when fabricated. It is very efficient and has good performance. The antenna, thus simulated in this paper, will be proficient in a great way for the fifth-generation telecommunication field when designed further for process innovation
Power Quality Assessment in Grid-Connected Solar PV Systems Using Deep Learning Techniques S., Dhivya; S., Prakash; Batumalay, Malathy
Journal of Applied Data Sciences Vol 6, No 2: MAY 2025
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jads.v6i2.655

Abstract

To address challenges in stability, power quality, and computational demands while supporting sustainable energy goals in grid-connected solar PV systems, this research introduces a novel deep learning approach: Adaptive Graph-Aware Reinforced Autoencoder with Attention-Based Neural Architecture Search (AGRAAN). AGRAAN simplifies and accelerates the development of neural networks by automatically identifying optimal architectures through Neural Architecture Search (NAS), enabling efficient learning from limited data using Few-Shot Learning, and enhancing performance through attention mechanisms for time-series forecasting. This integrated approach reduces manual tuning and adapts effectively to various tasks. High levels of solar PV integration in power grids introduce variability due to weather conditions and limited forecasting, often resulting in high operational costs. To address this, the AGRAAN model enhances real-time solar variability prediction, improving adaptability, cost-efficiency, and grid stability. NAS supports architectural optimization, Few-Shot Learning improves adaptability with minimal data, and attention mechanisms enhance forecasting accuracy. Additionally, high PV penetration causes voltage fluctuations and harmonic distortions in diverse grid environments. To mitigate these effects, a complementary system named Graph-Aware Reinforced Autoencoder Control System (GRAACS) is proposed. GRAACS detects and manages power quality issues using Autoencoders for anomaly detection, Graph Convolutional Networks (GCNs) for spatial prediction, and Reinforcement Learning for adaptive real-time control. The combined AGRAAN and GRAACS models significantly enhance performance, achieving a high efficiency score of 0.98, an F1-Score of 0.97, and a low Mean Absolute Error (MAE) of 0.11. These results demonstrate the effectiveness of the proposed AI-driven framework in optimizing solar PV grid integration for energy efficiency.
Automated Brain Tumor Analysis with Multimodal Fusion and Augmented Intelligence R., Karthick Manoj; S., Aasha Nandhini; Batumalay, Malathy
Journal of Applied Data Sciences Vol 6, No 2: MAY 2025
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jads.v6i2.719

Abstract

Brain tumor segmentation and classification are critical tasks in medical imaging, having a major impact on spotting and treating brain tumors. In the medical field, augmented intelligence has garnered a lot of attention lately since it emphasizes how human knowledge and artificial intelligence can be combined to enhance efficiency and decision-making in applications like brain tumor identification. This research concentrates on developing a novel approach utilizing Attention U-Net and Multimodal Transformers to assist doctors with precise tumor segmentation and classification while maintaining their critical clinical judgment. Attention U-Net is used to segment brain tumor because it efficiently collects detailed spatial data while focusing on key locations compared with traditional U-Net models. Multimodal Transformers provide reliable as well as effective feature extraction when utilized for early fusion to merge data from many modalities, such as T1, T2, and FLAIR This work utilizes CycleGAN-based data augmentation to supplement limited training data, thus improving the variety and quality of the dataset. The fused multimodal features are then utilized for the segmentation of the tumor and further classified as benign and malignant using hybrid transformer. The performance of the proposed system is assessed using standard metrics like accuracy for classification and Dice Similarity Coefficient and Intersection Over Union for segmentation. The proposed approach demonstrates high effectiveness in both segmentation and classification tasks, achieving 98 % accuracy showcasing its potential as a process innovation for clinical applications.
Modelling and Investigation of Solar Photovoltaic-Based Converter Configurations with Data Science Approach S., Prakash; S., Lakshmi; S., Priya; Batumalay, Malathy
Journal of Applied Data Sciences Vol 6, No 3: September 2025
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jads.v6i3.715

Abstract

Renewable energy sources, such as solar photovoltaic (PV) systems, typically produce low-voltage outputs, necessitating the use of high-gain direct current (DC) converters for efficient energy conversion. This study proposes a high-gain DC-DC converter for PV applications, designed with two MOSFET switches, two inductors, and two capacitors, offering a compact and efficient configuration. The converter achieves a high voltage gain of 6.8 and maintains a conversion efficiency of 97.7%, making it suitable for high-power applications. A data science-driven approach was employed to analyze the converter’s performance, integrating conventional simulation with machine learning techniques. Simulation results, conducted using MATLAB, confirmed the converter's superior performance, achieving an input ripple of 0.05% and an output ripple of 0.01%. Machine learning models, including Linear Regression, Decision Tree, Ridge Regression, and Support Vector Machine (SVM), provided deeper insights into the converter's behavior. Linear Regression accurately predicted output voltage, Ridge Regression minimized overfitting, and the Decision Tree model identified Duty Ratio and Input Voltage as the most critical factors affecting efficiency. SVM effectively classified operating conditions into high, moderate, and low efficiency. The Zero-Voltage Switching (ZVS) technique minimized switching losses, enhancing overall efficiency. This study demonstrates that integrating data science techniques with conventional analysis enhances the understanding and optimization of high-gain converters. The proposed converter provides a scalable and efficient solution for PV applications, offering insights for further optimization as part of process innovation.
Artificial raindrop algorithm for control of frequency in a networked power system Dhandapani, Lakshmi; Sreenivasan, Pushpa; Batumalay, Malathy
International Journal of Power Electronics and Drive Systems (IJPEDS) Vol 16, No 2: June 2025
Publisher : Institute of Advanced Engineering and Science

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.11591/ijpeds.v16.i2.pp1116-1123

Abstract

Load frequency control (LFC) evaluates the net changes in generation by continuously monitoring tie-line flows and system frequency required relying on load changes. It adjusts generator set points to minimize the area control error's (ACE) time-averaged value. ACE is regarded as a controlled output of LFC. Previous research focused on customary power systems like hydro-hydro, thermal-thermal, and hydro-thermal configurations. This current development study introduces the hybrid PV and dual thermal system interconnected systems for LFC analysis. The research evaluates LFC performance with different controllers, considering parameters such as maximum peak overshoot (Mp), maximum undershoot (Mu), settling time (Ts), and peak time (Tp). Controllers, including proportional integral (PI), anti-windup PI, fuzzy gain scheduling PI, and A cutting-edge algorithm generating fake raindrops are used for minimize ACE. The analysis introduces various load perturbations to observe controller performance in interconnected power systems. Both PV-thermal-thermal and thermal-thermal-thermal systems exemplify innovative approaches to energy management that bolster energy efficiency and sustainability. By integrating these advanced systems, we can make significant strides towards achieving global sustainability goals and promoting a cleaner and support energy efficiency for the future.
A Study of Unified Framework for Extremism Classification, Ideology Detection, Propaganda Analysis, and Flagged Data Detection Using Transformers Balajia, R S Lakshmi; Thiruvenkataswamy, C S; Batumalay, Malathy; Duraimutharasan, N.; Devadas, Amar Dev Thirukulam; Yingthawornsuk, Thaweesak
Journal of Applied Data Sciences Vol 6, No 3: September 2025
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jads.v6i3.702

Abstract

The rise of extremism and its rapid dissemination through propaganda channels have become pressing global challenges, threatening peace, security, and social cohesion. This study aligns with the United Nations Sustainable Development Goal 16 by proposing a unified framework leveraging advanced machine learning and large language models to combat extremism through extremism classification, ideology detection, propaganda analysis, and flagged word recognition. This framework introduces process innovation by integrating state-of-the-art transformer models such as BERT, RoBERTa, DistilBERT and XLNet to streamline the analysis process and overcome traditional limitations in extremism detection with exceptional performance: 90.00% accuracy for extremism classification, 98.82% accuracy for ideology detection, and 99.71% accuracy for flagged word recognition. While the proposed approach demonstrates high precision and recall, it faces challenges such as potential data bias, ethical concerns in dataset usage and the risk of false positives, which could lead to misclassification of benign content. The inclusion of multilingual capabilities broadens the applicability of the framework but variations in linguistic structures and cultural contexts introduce complexities in model generalization. Additionally, ethical considerations in handling extremist content, especially in social media data collection, necessitate stringent privacy safeguards to prevent unintended harm. By providing actionable insights, this research contributes to counter-extremism efforts in areas such as online content moderation, law enforcement and intelligence analysis, laying a foundation for future advancements in safeguarding global security which enhance the process innovation.
Modernizing Medicinal Plant Recognition: A Deep Learning Perspective with Data Augmentation and Hybrid Learning Nagrath, Preeti; Batumalay, Malathy; Saluja, Dhruv; Kukreja, Harsh; Tegwal, Devanshi; Saini, Akash
Journal of Applied Data Sciences Vol 6, No 4: December 2025
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jads.v6i4.795

Abstract

This study proposes a deep learning-based solution to address the longstanding challenge of accurately identifying Indian medicinal plants, which are vital to Ayurvedic pharmaceutics but often misidentified due to their morphological similarities. The objective is to develop a reliable, automated classification system using image processing and advanced neural network architectures. A dataset of 5,945 images representing 40 distinct medicinal plant species was sourced from Kaggle and augmented to 11,890 images using techniques such as flipping, rotation, and scaling to enhance diversity. The models tested include a baseline Convolutional Neural Network (CNN), transfer learning with DenseNet121, DenseNet169, and DenseNet201, a voting ensemble of these DenseNet variants, and a hybrid DenseNet201-LSTM architecture. Experimental results show that the CNN model achieved the lowest accuracy at 69.58%, while the hybrid DenseNet201-LSTM model reached the highest validation accuracy of 93.38%, with a precision of 94.74%, recall of 93.38%, and F1-score of 93.42%. These findings confirm the hybrid model’s superior ability to capture spatial and sequential dependencies in leaf features. The novelty of this work lies in the integration of DenseNet201 with LSTM for medicinal plant classification, which has not been widely explored in this domain. The study also acknowledges dataset scalability as a limitation and proposes future work involving dataset expansion through botanical collaborations, integration of environmental metadata, and deployment of a mobile application using TensorFlow Lite for real-time, low-resource implementation. Overall, the research contributes a robust and scalable framework for medicinal plant identification, promoting trust in traditional medicine, supporting conservation efforts, and enabling practical field-level applications in both rural and clinical settings.
Study of Machine Learning Techniques for Predicting Panic Attacks with EEG and Personalized Binaural Beat Frequencies Batumalay, Malathy; Lakshmi Balaji, R S; Yingthawornsuk, Thaweesak
Journal of Applied Data Sciences Vol 6, No 4: December 2025
Publisher : Bright Publisher

Show Abstract | Download Original | Original Source | Check in Google Scholar | DOI: 10.47738/jads.v6i4.759

Abstract

Panic attack detection and intervention remain critical challenges in mental health care due to their unpredictable nature and individual variability. This study proposes a machine learning-based framework for early detection of panic attacks using EEG-derived physiological signals, coupled with real-time personalized auditory intervention through binaural beat frequencies. Data were collected under controlled conditions using wearable biosensors to capture features such as heart rate variability, electrodermal activity, and skin temperature. A Gradient Boosting Classifier achieved 96% accuracy in detecting panic states, while an Isolation Forest algorithm effectively identified anomalous patterns preceding attacks. Based on physiological profiles, the system dynamically recommends individualized binaural beat frequencies to promote relaxation and emotional stabilization. The results demonstrate the feasibility of combining predictive modeling and neuroadaptive sound therapy to deliver scalable, non-invasive, and personalized mental health interventions. This approach aligns with global preventive health strategies, particularly those promoting digital therapeutics and early intervention for anxiety-related conditions.