Journal of Intelligent Data Analysis and Computational Statistics (p-ISSN: 3049-3056 e-ISSN: 3048-7080)

A Graph-based Bottom-up Salient Object Detection Framework Using Objectness Cues and Multilayer Cellular Automata

S. Vanitha Sivagami, G Ananthi — Mon, 02 Feb 2026 00:00:00 +0000

In the digital era, salient object detection (SOD), which focuses on locating and extracting visually prominent objects from an image, plays a crucial role in various computer vision applications, including image segmentation, object recognition, and scene understanding. In this proposed work, an effective saliency detection framework is presented by fusing multiple complementary cues to improve detection accuracy. Initially, saliency maps are generated using a bottom-up saliency detection approach combined with a graph-based salient object detection method and objectness cues, which help in highlighting potential object regions. A graph is constructed by computing the geodesic distances among all superpixels derived from the saliency map, enabling the modeling of global and local relationships between image regions. To further enhance the consistency and accuracy of the detected salient regions, saliency optimization is performed using a multilayer cellular automata mechanism. This optimization process refines the saliency values by propagating information across multiple layers, leading to well-defined object boundaries and improved foreground-background separation. Extensive experimental evaluations conducted on benchmark datasets demonstrate that the proposed method consistently outperforms several existing bottom-up saliency detection approaches in terms of precision, recall, and visual quality.

An Experimental Study on Supervised Learning and Random Forest Feature Selection for Edible and Toxic Mushroom Classification

G. Ravi Kumar, G. Thippanna, D. Mahaboob Basha — Wed, 11 Feb 2026 00:00:00 +0000

This paper presents an experimental study on data-driven classification using machine learning techniques with a focus on feature selection for improving predictive performance. The study utilizes a structured benchmark dataset consisting of 8124 instances with 23 categorical features to classify samples into edible and poisonous categories. To enhance model efficiency and accuracy, a random forest (RF)-based feature selection method was applied to identify and retain the most informative attributes. Three supervised learning algorithms, logistic regression (LR), naïve Bayes (NB), and decision tree (DT), were implemented and evaluated both before and after applying RF-based feature selection. Accuracy, precision, and recall were the standard criteria used to evaluate performance. The experimental results show that all classifiers achieved improved performance after feature selection, with the decision tree model obtaining the highest classification accuracy. The findings demonstrate that integrating ensemble-based feature selection with traditional classifiers significantly enhances classification performance and model robustness, particularly for high-dimensional categorical datasets. This study highlights the effectiveness of feature-driven learning frameworks for reliable and efficient data-driven classification systems.

Climate-aware Crop Yield Prediction across Indian Agro-climatic Zones Using Hybrid Deep Learning and Zone-specific Transfer Learning

Ravi Verma, Raj Kumar Sahu, Aashish Sharma — Mon, 16 Mar 2026 00:00:00 +0000

Predicting crop yield has remained difficult for India, primarily due to uneven agro-climatic conditions, where factors like rainfall, temperature, soil type, and crop management vary greatly and restrict the application of existing predictive models. Most machine-learning and statistical models take uniform climate conditions as a given, and thus their performance declines sharply when used in different agro-climatic zones. This study addresses this gap and proposes a climate-aware crop yield prediction model combining hybrid deep learning with zone-specific transfer learning to improve robustness and adaptability to different climate conditions. The model integrates agro-meteorological data with climate data and a constructed variable representation obtained from a combination of different sources, comprising weather data, soil data, and remote sensing data (satellite vegetation indices) over time. The first step involves training a global deep learning model on the national agricultural data to capture crop-climate relationships. In the second step, the global model is fine-tuned using zone-specific transfer learning for each agro-climatic zone with a few samples. Evaluation experiments on different crops show that the machine learning approaches’ prediction error is reduced by 12 to 25%, and combined with other zone-agnostic deep learning models by an additional 9 to 18%. The model demonstrates more stable learning, faster convergence, and lower inter-seasonal variability, indicating robust performance under variable climate conditions. The findings validating the integration of agro-climatic awareness and transfer learning as a means to achieve scalable and climate-resilient forecasting of crop yields further offer a sound basis for precision agriculture and agro-policy framework in India.

Data-driven Drought Monitoring Using Satellite-derived Indices: A Google Earth Engine and Python-based Framework

Vemu Sri Sai Tarun, Vemu Sreenivasulu — Wed, 18 Mar 2026 00:00:00 +0000

Droughts are among the most devastating natural hazards, affecting water resources, agriculture, and the livelihoods of millions. This study presents a framework for assessing meteorological drought using satellite data through Google Earth Engine (GEE) and Python. This study focuses on meteorological drought assessment in selected districts of Telangana, India, including Nizamabad, Khammam, Mahbubnagar, Karimnagar, and Jangaon. The standard precipitation index (SPI) and drought severity index (DSI) were employed for drought analysis. SPI values, computed based solely on precipitation data, and the DSI values, computed using GEE, were compared with historical drought records. GEE provided efficient data processing and analysis during 2000–2021. The findings revealed the persistence of drought in the study districts: SPI values indicated drought persistence of 8% for Jangaon, 7.6% for Karimnagar, 8.7% for Mahbubnagar, 8.33% for Nizamabad, and 11.36% for Khammam. DSI values indicated drought persistence of 13% for Jangaon, 14.5% for Karimnagar, 13.27% for Mahbubnagar, 12.88% for Nizamabad, and 11.7% for Khammam. This research highlights the utility of SPI and DSI as effective tools for meteorological drought monitoring and emphasizes the role of GEE in facilitating efficient and scalable drought assessment.

AI-Driven Innovations in Clinical Diagnosis and Personalized Treatment

Samiksha R. Patil, Ranjit R. Patil, G. B. Koravi — Mon, 30 Mar 2026 00:00:00 +0000

Artificial Intelligence (AI) has become a transformative force in modern healthcare, particularly in the areas of medical diagnosis and treatment planning. With the rapid digitization of healthcare systems, vast amounts of data are generated daily through electronic health records, laboratory reports, medical imaging, and wearable devices. Analyzing this complex and high-volume data using traditional methods can be time-consuming and prone to human error. AI technologies, including machine learning, deep learning, and natural language processing, provide advanced computational techniques that can identify patterns, detect abnormalities, and generate predictive insights with high accuracy. As a result, AI systems are increasingly integrated into clinical environments to support healthcare professionals in making faster, more reliable decisions. In medical diagnosis, AI has shown significant success in areas such as radiology, pathology, cardiology, and oncology. Intelligent algorithms can analyze X-rays, CT scans, MRI images, and histopathological slides to detect diseases at early stages, often with performance comparable to medical experts. Additionally, predictive models can assess patient risk factors and forecast the likelihood of developing chronic conditions such as diabetes or heart disease. In treatment planning, AI contributes to personalized medicine by recommending therapies tailored to individual patient profiles, including genetic information and previous treatment responses. This data-driven approach enhances precision, improves treatment effectiveness, and reduces potential side effects. Despite its many advantages, the implementation of AI in healthcare presents challenges related to data privacy, ethical considerations, transparency of algorithms, and the need for regulatory compliance. Ensuring the reliability and fairness of AI systems remains a critical concern, especially when clinical decisions directly affect patient safety. Therefore, successful integration requires collaboration between healthcare providers, technology developers, and policymakers. This paper explores the role of Artificial Intelligence in medical diagnosis and treatment planning, highlighting its technologies, applications, benefits, limitations, and future potential in advancing patient-centered healthcare systems.