Suriya Sundaramoorthy

The widespread dissemination of inaccurate information on digital platforms poses a threat to social trust, public safety, and democratic institutions. This work presents a novel and efficient model to mitigate the risk of identifying fake news that has three major components: context-aware text embeddings using DistilBERT, PCA for dimensionality reduction, and feature selection using a Genetic Algorithm (GA). The lightweight transformer model DistilBERT is utilized for the generation of 768-dimensional embeddings that provide deep contextual and semantic meaning of the text. To overcome the issues high-dimensional data poses regarding computational cost and overfitting, PCA is used to maintain 95% of data variance while utilizing significantly fewer features. For maximizing accuracy and model interpretability, an attribute selection procedure based on GA is subsequently utilized to select the most informative and discriminative attributes from the reduced feature space. This two-stage optimization (PCA followed by GA) is one of the paper's main contributions, distinguishing it from much of the prior work that primarily uses full embeddings or simple filters. For precision, a Logistic Regression classifier is employed for the final classification, even compromising on interpretability. The model attains a high accuracy of 98% when tested on a synthetically equalized set of fake reports. It also shows significant improvements in precision, recall, and F1-score when compared to other models. This system can identify fake news on various digital platforms in real time, quickly, and in scalable ways due to the combination of a high-quality language model, dimensionality reduction, and evolutionary optimization.

Oculang is a communication system that uses computer vision and machine learning techniques to enable individuals with neurodegenerative disorders to communicate using eye gestures. The application focuses on detecting facial landmarks, gaze estimation, and blink detection to analyze combinations of eye movements captured from video input and produce message outputs. Dlib’s shape predictor (68 facial landmarks) and OpenCV-based image processing methods are used to extract and process the features of the eye region. A decision-making algorithm maps the detected gestures and predefined keywords for message generation. The application supports both real-time video capturing and uploading video through a Django-based user interface. Experimental evaluation on recorded datasets of eye movements demonstrated robust performance in accurately recognizing blinks, winks, and gaze directions, validating the system’s reliability.

This study aims to research on immersive and interactive experience to users by using augmented reality and GPS technology. The system will be built using the Vuforia engine or ARcore support, and will be able to detect the user's location and device location using GPS. Once the system is initiated, it will check for GPS support in the user's device and ask for camera access. Then, using Vuforia's markerless tool and AR+GPS package, the system will detect the user's location and search for the rendered scene for the respective location. Each building will be labelled with AR objects for better understanding of the place and tracking where the user is. The AR objects will provide information about the building, such as its name, purpose, and history. This will enhance the user's experience and enable them to navigate the campus more easily. The application will create a scene that renders the real environment image and the rendered AR scene image, providing an interactive output. Users will be able to interact with the AR objects by clicking on them to access more information, such as photos and videos. Overall, the proposed system will be a useful tool for navigating the campus, providing users with a unique and immersive experience.

Type 2 diabetes is a persistent disorder that affects millions of individuals globally. It is characterised by the excessive levels of glucose within the blood due to insulin resistance or the incapability to supply insulin. Early detection and prediction of type 2 diabetes can improve patient outcomes. K-Nearest Neighbor (KNN) is used in the present model to predict type 2 diabetes. The KNN set of rules is a simple but powerful machine learning set of rules used for categorization and regression. It's far a non-parametric approach that makes predictions based totally on the nearest k-neighbours in a dataset. KNN is widely used in healthcare and scientific studies to expect and classify sicknesses primarily based on the affected person’s data. The intention of this work is to predict the threat of growing type 2 diabetes using the KNN set of rules. Data has been collected from electronic medical records of patients diagnosed with type 2 diabetes and healthy individuals. The dataset consists of various patient attributes, such as age, gender, body mass index, blood pressure, cholesterol levels, and glucose levels. Information has also been collected about lifestyle habits, such as physical activity, smoking status, and alcohol consumption. Data have been pre-processed by removing missing values and outliers, and normalization of the data has been done to ensure that all features have the same scale. Splitting the dataset into training and test sets, with training sets using 80% of the data and test sets using 20% of the data is performed. KNN algorithm have been used to classify the patients into two groups: those at high risk of developing type 2 diabetes and those at low risk. The model's performance has been assessed using a variety of metrics, including accuracy, precision, recall, and F1-score.

Accurate case predictions are essential for efficient public health management and resource allocation since the COVID-19 pandemic has had a substantial impact on economies and global health. Using polynomial regression, a machine learning technique that fits a polynomial function to the data, this research seeks to create a predictive model for future COVID-19 cases. The model takes into consideration the elements such as population density, healthcare facilities, and governmental initiatives using historical COVID-19 case data from India. In order to forecast the number of upcoming COVID-19 instances, the polynomial regression model is employed. The model's effectiveness is assessed using a number of measures, including mean squared error and R-squared. The outcomes demonstrate that the polynomial regression model can precisely forecast the trend of COVID-19 instances over time. This approach can be useful for forecasting the spread of the virus and informing public health policies. The limitations and future directions of the model are also discussed. Furthermore, the model's adaptability to changing trends and its ability to capture non-linear relationships between variables, make it a promising tool for forecasting future pandemics and other public health crises.

Agriculture is an important sector for both human survival and economic growth. It has to be managed efficiently. This can be done by the use of technology in order to minimize human effort. It can be managed efficiently by following crop management tasks. One such crop management task is the identification and removal of weeds. Weeds are considered to be plants which are not required to be grown with the agricultural crops, because the weeds also utilize the water and nutrients like the agricultural crop and cause impact on the growth of agricultural crops. In order to identify weeds, deep learning technology can be used. The proposed system helps to classify weeds using Convolutional Neural Networks. This system employs models like, ResNet50, MobileNetV2 and InceptionV3, which are used for better classification. The system is evaluated based on these models, and all the three models have resulted in better accuracy.

Issues regarding trespassing and intrusion of fishermen in the maritime boundary line is of great importance to be discussed nowadays. One of the main reasons still existing is transgression for better catch of fishes in foreign waters. Thus is a concern, and in order to prevent this issue from becoming a national security threat, it should be taken care of, by identifying the intruders as the first step to get a better view on the situation. Finally, in the hope to slim the chances of transgressions by marine fisher folk, a SVM model based on Automated Identification System that makes use of real-world data is implemented that will analyse the possibility of successful detection of intrusions of fisherman by categorising the vessel as normal or anomalous one. Convolution Neural Network model is used to find whether it is ship or not a ship, and if it is ship then it will categorize whether it belongs to anomalous or non-anomalous. The model's validation accuracy of 96% shows that it can correctly identify whether a ship is present in each image.