Essays

• How a castle's construction is impacted by its location This project investigates how the construction and features of English castles were influenced by their geographic location, using natural language processing to extract architectural traits from official heritage records and mapping them across NUTS1 regions. The analysis reveals strong correlations between castle features—such as building materials, basements, tunnels, and stables—and regional geology, historical threats, and local craftsmanship, highlighting both the physical and socio-political factors that shaped medieval fortifications.
• Image Processing using JavaScript and Web Assembly This study evaluates the performance benefits of using Web Assembly over JavaScript for browser-based image processing tasks by implementing and benchmarking algorithms such as JPEG compression, Gaussian blur, brightness, and contrast. Results show that Web Assembly significantly outperforms JavaScript for complex computations—achieving up to 8× speed improvements and reduced memory usage—though native applications like ImageMagick remain superior, highlighting Web Assembly’s strengths and limitations in modern web development.
• Building a hybrid recommender system This paper presents a hybrid recommender system designed to suggest Italian restaurants using Yelp review data, combining content-based and collaborative filtering with a weighted approach. Through evaluation against baseline models and comparative studies, the hybrid model demonstrates improved precision, balanced recall, and lower root mean square error, validating the effectiveness of integrating multiple techniques in a real-world recommendation scenario.
• A report on Non-Local Means Denoising This report explores the Non-Local Means (NLM) denoising algorithm, which enhances image quality by averaging similar pixel patches across an entire image rather than relying solely on local neighborhoods. It analyzes the algorithm's underlying principles, implementation strategies, performance compared to other methods, parameter sensitivities, and adaptations for applications such as medical imaging and video denoising.
• To what extent could Quantum Computers Improve On Classical Computers? This essay explores the potential of quantum computers to surpass classical computers by examining their physical principles, computational methods, and algorithmic advantages. Through detailed comparisons—including hardware structures, quantum gates, and key algorithms like Grover’s and Shor’s—it concludes that while quantum computing offers significant theoretical performance gains in fields like encryption and large-scale search, practical limitations in hardware and accessibility mean classical systems remain dominant for most real-world applications today.