CR Rao Advanced Institute of Mathematics, Statistics and Computer Science

ABSTRACT The existence of Bushfire is a significant problem. So not only are environmental threats to life and society a risk, but also the threatening of the safety of all. Thus it is extremely critical to detect these fires in time. This identification can assist in the successful application of fire management. Existing approaches might not be effective under conditions such as bad weather. A decrease in visible light also can affect accuracy. These techniques focus on the identification of fires during early stages. We introduce an innovative method in this study. The early detection of bushfires is achieved through UAVs. Each drone in the network has a thermal image camera onboard to provide real‐time surveillance of the assigned area. The drones are also equipped with sensors that detect smoke and other indications of potential fire. A control network receives up‐to‐the‐minute information from drones. Machine learning algorithms are used to interpret this data. The aim is to detect potential fires. The above detailed evaluations were captured using the eBee X and DJI Matrice 300 RTK. This study was carried out in a dedicated area provided by a National Park and involved the use of fixed and multirotor UAVs. State of the art sensor system could be successfully deployed on board the UAVs. The integrated technology was comprised of a Sequoia+ multispectral camera, H20T Quad Sensor and PGS 813 Gas Sensor. These were test burns, including controlled prescribed burning and fire simulations. They highlighted the system's impressive ability to detect bushfires in their early stages. The system has provided a quantum leap in preventive fire management. And it has heightened cost‐mindedness in real‐world settings. This enhancement is due to its high‐quality monitoring and precise detection of fire signs. The proposed method might be a useful and powerful tool for proactive fire fighting and early warning, such as fighting wildfire and related scenarios.

Phishing attacks remain among the most regular cybersecurity threats since they exploit human weaknesses to penetrate systems unauthorized. Security procedures that exist today are unable to identify advanced cyberattacks effectively which leads to a requirement for updated protection measures. The combination of machine learning with deep learning and natural language processing methods within artificial intelligence creates an effective tool for detecting phishing threats. Algorithms based on artificial intelligence enable the analysis of large datasets for detecting anomalous email patterns and identifying dangerous URLs while recognizing unhonest website characteristics at a level of high precision. This research investigates how artificial intelligence (AI) detects phishing through its superior capability to identify threats in real time while improving accuracy by learning from experience and establishing fewer incorrect alerts. The text addresses main obstacles like adversarial attacks, data privacy challenges along with the requirement of continuous model maintenance. The incorporation of AI-based phishing detection systems into existing cyber security frameworks leads to defense improvements that lower both data breach possibilities and financial losses for organizations.

Agriculture is under great threat from climate variability, resource mismanagement, and plant diseases, which will affect crop yields and food security. This project proposes an AI-based system that combines crop yield prediction and plant leaf disease detection to increase modern farming efficiency. The system uses Random Forest for the analysis of numerous environmental factors such as rainfall, temperature, soil quality to make accurate yield predictions. At the same time, YOLOv8, a cutting-edge object detection model, is utilized to detect diseased leaf areas with high accuracy even in adverse conditions such as low light and occlusions. By integrating these two models, the system will give an integrated decision-support platform that will assist farmers in optimizing resource allocation, avoiding crop losses, and also enhancing productivity. The real-time data provided by the system enables proactive disease management and enhanced strategic planning, which ultimately results in more sustainable farming. Through thorough testing, the system proves to have high accuracy in yield prediction and disease detection, indicating its potential in increasing efficiency in food production. This AI-powered facilitates precision farming, which improves farmers' capacity to control the risks also to maximize crop management based on insights informed by data.

Deepfake technology’s widespread use has sparked serious worries about the security and authenticity of digital media. Using several pre-trained convolutional neural networks (CNNs), this study offers a thorough ensemble learning method for deepfake image detection. In addition to a specially created 9-layer CNN model, we use four cutting-edge architectures: VGG16, VGG19, InceptionV3, and ResNet50. To improve model performance, our methodology combines hyperparameter optimization, augmentation techniques, and extensive data pre-processing. Better-performing models are given higher weights in the ensemble technique, which uses a weighted average approach to combine predictions from all models. The efficacy of our method is demonstrated experimentally on a dataset of 140,000 real and fake images, with detection accuracies of 95.27% (VGG16), 95.21% (VGG19), 95.02% (Custom Model), 94.0% (ResNet50), and 77.20% (InceptionV3).When compared to individual models, the ensemble approach greatly increases overall detection accuracy. Our research addresses important issues in cybersecurity and digital media forensics by advancing the creation of reliable deepfake detection systems. The open-source implementation facilitates reproducibility and community contributions to this important field of study.

The increasing prevalence of thyroid disorders necessitates an efficient and reliable system for early diagnosis and classification. Machine learning (ML) offers a promising approach to predict thyroid conditions by analyzing clinical data effectively. This study focuses on developing a robust thyroid classification system using four machine learning algorithms: Logistic Regression (LR), Random Forest (RF), Support Vector Machine (SVM), and K-Nearest Neighbors (KNN). The system leverages a publicly available thyroid dataset, consisting of various clinical parameters, for preprocessing and model training. The ML pipeline begins with data preprocessing techniques such as handling missing values, normalization, and feature selection to ensure the quality of input data. Logistic Regression, as a baseline model, provides interpretability and initial insights into the dataset. Random Forest, a robust ensemble algorithm, is employed for its capability to handle high-dimensional data and prevent over fitting. Support Vector Machine, with its kernel-based approach, demonstrates high performance in identifying non-linear relationships within the data. K-Nearest Neighbors is utilized for its simplicity and effectiveness in capturing local patterns. Extensive experimentation is conducted to evaluate the models using metrics such as accuracy, precision, recall, F1-score, and area under the ROC curve (AUC). Results reveal that Random Forest outperforms other algorithms in terms of accuracy and recall, demonstrating its suitability for thyroid classification. However, SVM also achieves competitive results, particularly for imbalanced data scenarios. The study highlights the importance of feature selection and hyper parameter tuning to optimize model performance.

Concurrently sophisticated online dangers have proven that conventional rule-based security approaches fail to identify current attacks effectively. A self-governing AI system with hybrid deep learning models for immediate security threat recognition functions as a solution to solve this issue. The designed system merges LSTM technology with Transformer-based system features to detect network traffic patterns for precise attack prediction. Evaluation conducted on the CICIDS2017 and NSL-KDD datasets demonstrated a 97.5% accuracy rate lower false positive rates by 30% and superior performance than traditional models with 5-7% improvement. The system proved its capability to boost security defenses due to its ability to spot zero-day assaults and develop threat defense capabilities. The implemented model delivers efficient and scalable proactive security capabilities which position it as an effective solution for delivering real-time cybersecurity applications.

We introduce a generalized mixed fractional Brownian motion (gmfBm) as a linear combination of n independent fractional Brownian motions with possibly different Hurst indices and investigate conditions under which the time-changed gmfBm exhibit long-range dependence when the time-change is induced by a tempered stable subordinator or a Gamma process. AMS Subject Classification (2020): Primary 60G22

In recent times, the pressing need to develop sustainable food systems has been rising amidst the already existing concerns over environmental degradation, resource depletion, and issues related to mental health. The issue of long-term food security and ecosystem health is driven by poverty through the traditional agricultural practices that are common across the world, including soil erosion, reduced fertility, and increased greenhouse gas emissions. To address these issues, this study brings forth the evaluation of sustainable agricultural methods to provide answers to what the most appropriate sustainable agricultural practices are to establish resilient food systems. Based on the impact to soil health and crop yields, the comparison is made between Organic Farming, Conservation Tillage, Agroforestry, Integrated Pest Management (IPM), Permaculture, and Regenerative Agriculture. We then propose our system, which integrates these methods to see what kind of benefit and what kind of limitation they have and identify the most promising strategy. Our results show that all methods have very valuable benefits, but Regenerative Agriculture is the best of the best. Compared to other methods, it is good for soil health, crop yield, and environmental sustainability. Regenerative Agriculture is superior concerning soil organic matter, erosion control, and resilience of the ecosystem.