22505 research outputs found
Sort by
Procedurally Textured Gaussians
No full textGaussian Splatting is rapidly establishing itself as the leading methodology for Novel View Synthesis—the process of generating arbitrary views of a 3D scene from sparse 2D input images—and from its popularity, various geometric and texture-based improvements have emerged. We propose a novel texture-based modification to Gaussian Splatting that draws inspiration from the advent of procedural textures in classical rendering, superseding previous texel-based approaches by training a continuous texture per splat that improves view fidelity while also lending itself to be easily combined with geometric improvements to baseline Gaussian Splatting
A Morpho-Kinematic Model of Molecular Line Emission in Planetary Nebulae
No full textPlanetary nebulae (PNe), the late-stage remnants of low- to intermediate-mass stars, offer valuable insights into the final phases of stellar evolution and the enrichment of the interstellar medium via intermediate-mass star ejecta. Molecular emission from these objects serves as a powerful diagnostic for probing physical conditions, chemical composition, and kinematics of these ejecta. In this thesis, we describe the flexible, modular modeling framework we are developing to analyze high spatial and spectral resolution molecular emission-line data for planetary nebulae, as obtained by modern mm-wave interferometers such as the Submillimeter Array (SMA) and the Atacama Large Millimeter/submillimeter Array (ALMA). The Python- based coding routine uses morpho-kinematic modeling of key molecular species, like CO, to reconstruct the large-scale molecular morphologies and velocity structures of the nebulae. We demonstrate the model’s application and capabilities by constructing 3D morpho-kinematic models which reproduce the basic structures observed in velocity-resolved SMA and ALMA CO maps of the planetary nebulae NGC 6720, NGC 3132, and Hubble 5. The resulting models reveal complex three-dimensional structures and provide new constraints on nebular geometry, excitation conditions, and evolutionary pathways. These findings illustrate the potential our new modeling framework has to advance our understanding of mass loss during the late stages of stellar evolution
Supervised Learning for Predicting Mental Health and Burnout in Healthcare Workers
No full textThis study aims to predict and analyze burnout among healthcare workers using supervised machine learning techniques and Exploratory Data Analysis (EDA). Leveraging the Healthcare Workforce Mental Health Dataset, the research identifies key demographic, occupational, and psychological factors most strongly associated with burnout. The methodology involves data preprocessing, feature selection, and model training using algorithms such as logistic regression, decision trees, random forests, and gradient boosting. Model performance will be evaluated through standard metrics, including accuracy, precision, recall, and ROC-AUC. The expected outcome is a predictive framework that highlights high-impact burnout predictors and generates actionable insights to support early intervention and prevention strategies, thereby enhancing the overall well-being and resilience of the healthcare workforce
ChatGPT as a Mental Health Ally: A Study on College Students’ Adoption of AI for Therapy
No full textThis research investigates college students’ use of ChatGPT for mental health support, addressing a population with high unmet mental health needs due to barriers like accessibility and cost. Through a mixed-methods study, which included a survey of 126 students and sentiment analysis of 1,200+ social media posts, the research examined adoption prevalence, gender influences, and perceived benefits and limitations. Survey findings show 40.5% of students use ChatGPT for mental health, especially those with self-reported challenges. Female students reported higher adoption, linked to greater mental health needs and openness to supplementary support. Key benefits included 24/7 access, anonymity, and low cost, while major concerns were lack of empathy, privacy risks, and questionable reliability. The study concludes that ChatGPT serves as a supplement and not as a replacement to traditional mental health support. Recommendations include developing AI literacy programs for students and exploring hybrid care models. Future research should employ longitudinal and qualitative designs to further explore long-term impacts and user experiences, guiding the responsible integration of AI into student well-being frameworks
Beyond Detection: A Batch-Based AI Framework for Temporal and Event-Correlated Trend Analysis of Misinformation on Social Media
No full textThe rapid spread of misleading information on social media influences public behaviour and complicates crisis communication. Although transformer models such as BERT accurately detect misinformation at the post level, most studies analyse posts in isolation and overlook howmisinformation fluctuates over time or responds to major events. This study addresses that gap by developing an end-to-end analytical workflow that integrates BERT-based classification with temporal aggregation, topic clustering, anomaly detection, and event alignment. The analysis uses 10,700 COVID-19–related tweets (6,420 training, 2,140 validation, and 2,140 testing). Because timestamps were unavailable, synthetic timestamps were assigned using an evenly spaced date range between 1 January 2020 and 1 August 2020, and ISO week numbers were derived to support temporal exploration. A fine-tuned BERT-base model trained with early stopping achieved strong test performance (accuracy = 0.956; F1-scores ≈ 0.95– 0.96). Topic discovery using TF–IDF and K-means (with a five-cluster solution) revealed shifting thematic patterns. For anomaly detection, Isolation Forest and a simple peak-based approach were tested; the latter more reliably identified high-activity weeks. Assigning synthetic event labels to the detected spike weeks enabled structured interpretation of peak periods within a simplified timeline (e.g., early alerts, lockdown communications, and vaccine-trial updates), without implying real-world causality. Rather than treating detection as an end in itself, the workflow reframes classifier outputs as signals for temporal monitoring, narrative interpretation, and contextual prioritisation. Overall, the findings show that combining transformer-based classification with lightweight temporal and event-aware methods provides a clearer picture of how misinformation evolves. The framework is modular, interpretable, and suitable for extension to real temporal data, multimodal contexts, or early-warning applications
Predicting Airbnb Prices in Dubai using Machine Learning
Get PDFThe Airbnb rental market in Dubai is dynamic and full of varying factors like location, equipment, and demand time shortages, which makes it challenging to have fair pricing for the different levels of Airbnb properties in Dubai. The goal of the project was to construct a powerful machine-learning model for the forecast of the Airbnb rental price in Dubai The method used by the project was systematic from the beginning to the end stages, which included the data collection from Kaggle, the data preprocessing which involved the handling of missing values, outlier elimination, data type conversion, and exploratory data analysis (EDA) which helped in obtaining insights into the distribution of listing attributes, the correlation between variables, and the market segmentation based on property characteristics. Subsequently, feature engineering solutions were applied, which allowed selection and transformation into new features such as amenities and property attributes in order to increase the model\u27s precision. Out of the machine learning algorithms, namely Linear Regression, Random Forest Regression, Gradient Boosting Regression, and ensemble model trained and evaluated on the pre-processed data, it was the Random Forest Regression model that emerged as the best among the Hyperparameter tuning further improved the model\u27s performance, So, it was clear that the adoption of optimization approaches boosted the performance of the model very much. The research report provides a very significant source of information for the proprietors, property managers, and investors in the Airbnb rentals market in Dubai, maintaining their decision-making rationale data-driven and strategy-based regarding rents
Towards Characterizing and Quantifying Interpretability of Knowledge Graph Embedding Models
Get PDFKnowledge graphs are structured representations of real-world information, where entities are connected by edges with labels known as predicates. These graphs contain logical patterns, known as inference patterns, such as symmetry, transitivity and composition, to name a few. To leverage the information contained within a knowledge graph, embedding methods have been developed to encode entities and predicates as numerical vectors in a multi-dimensional space. These embeddings aim to capture the semantic and structural information of the graph, including the inference patterns. However, the sub-symbolic nature of these embeddings present significant challenges in terms of interpretability. Additionally, each embedding model utilizes different embedding representations and scoring functions, adding to the complexity of interpretations. Therefore, it is important to develop interpretation methods that do not solely rely on the embeddings of the model, and can be applied to any knowledge graph embedding model. To address these challenges, this dissertation proposes model-agnostic methods for interpreting knowledge graph embeddings. By focusing on approaches that are independent of specific embedding architectures, we aim to provide a unified framework for analyzing and comparing model interpretability. Our methods leverage the link prediction evaluation task, a common protocol for assessing knowledge graph embedding models. During this task, we collect edges that are deemed plausible by the embedding model at hand into a separate knowledge graph, which serves as the basis for our interpretability analyses. The main contribution of this dissertation is twofold. First, we extract Horn rules from the knowledge graph of deemed-plausible edges to provide dataset-level (global) interpretations of a model\u27s behavior. We introduce the concept of interpretation accuracy, which quantifies how well these extracted rules represent the edges deemed plausible. Interpretation accuracy enables quantitative comparisons among diverse knowledge graph embedding models. Second, we propose a method to analyze the ability of embedding models in capturing inference patterns. We propose an overlap metric that assesses how effectively an embedding model captures inference patterns present in the original graph. Both interpretability methods allow us to gain insights into the strengths, weaknesses of different embedding models. By providing a clear understanding of the logical relationships captured by the embeddings, these methods help identify potential biases or gaps in the models under evaluation
Impact of Repeated Washing Cycles on Surface Integrity and Mechanical Properties of Glass Bottles
Get PDFGlass bottles are a top choice in the global beverage industry, but their inherent brittleness poses challenges in maintaining durability. This study focused on investigating the impact of numbers of washing processes from reusing on the surface quality and mechanical properties of glass bottles after each washing cycle. Contact angle technique was used to determine surface quality of glass bottles. It was found that mechanical integrity, as reflected in internal pressure resistance values, and surface quality of glass bottles exhibited an exponential decline with each washing cycle. Furthermore, a predictive model (R2 = 0.8882) was developed to estimate its internal pressure resistance of a bottle at a specific number of washing cycles. These findings have significant implications for the reuse of glass bottles. As washing cycles increase, the declining integrity could lead to a higher rate of bottle failures. Reusing glass bottles, while environmentally beneficial, requires careful consideration of the washing process to balance longevity with sustainability goals
Fake News Detection on X (Twitter)
Get PDFThe most important innovation globally is the internet, which is used by a huge number of people all around the world. People use the internet for different purposes. By using different social media platforms people share content and news. Thus, some people use these internet platforms to share false news or fake news among people. Propaganda against a person, group, organization, or political party may be present in this news. For an individual, it is difficult to distinguish between false and true news over the internet. Therefore, machine learning is the smart solution that can identify fake news on the internet automatically. The goal of this project was to stop misinformation from spreading on X (Twitter). Considering that X (Twitter) is a significant real-time information source as well as a haven for the quick spread of false information, this study attempts to address the threats fake news poses to democratic processes, public opinion, and social stability. The research aims to develop strong detection systems that can reliably discern between authentic news and fake information by utilizing machine learning algorithms, natural language processing methods, and network analysis. This study carries out fake news detection using machine learning algorithms. In conclusion, the project on fake news detection on X (Twitter) represents a significant step toward mitigating the pervasive issue of misinformation on social media platforms. By leveraging advanced machine learning techniques and natural language processing, the project aims to build robust models capable of identifying fake news with high accuracy. Through comprehensive data collection, rigorous preprocessing, and continuous model evaluation, the project addresses the challenges associated with the dynamic and diverse nature of content on X (Twitter)