Projects
Some of the projects I have carried out prior to my current line of research.
Bachelor's Thesis - Tracking system for obtaining training data for detecting online misinformation
This work explores the use of supervised natural language processing (NLP) models to address online misinformation, with a focus on health-related content in search engine results. Given the lack of suitable datasets, I developed a custom web crawler to collect and label general-domain data by truthfulness. These datasets were used to train several NLP classifiers, which were then evaluated on health misinformation data from the 2021 TREC Health Misinformation Track.
Master's Thesis - GAN-based data augmentation for the classification of remote sensing multispectral images
Research on the application of Generative Adversarial Networks (GANs) for data augmentation in remote sensing image classification tasks. The work focuses on improving classification performance by generating synthetic multispectral images that maintain the spectral characteristics of the original data.
Master's Project - Comparative Analysis of Control Methods for Wall-Following in TurtleBot3
This Master's project explored wall-following behaviors for the TurtleBot3 Burger mobile robot, comparing rule-based, PID, and machine learning (ANN, Decision Tree) control approaches. Evaluated in Webots simulations, the PID controller demonstrated superior robustness and generalizability compared to rule-based and less effective ML methods. The work highlights the complexities of applying ML in robotics and favors PID for its explainability and adaptable precision.
Master's Project - Multi-Agent Book Auction System with JADE Framework
This project presents a carefully designed book auction system using the JADE framework. It features an auctioneer agent and reactive/deliberative bidding agents. The architecture includes class diagrams and swimlane diagrams to clarify structure and interactions. Agent behaviors are formally modeled with mathematical functions that define bidding logic based on interest, aggressiveness, and budget. Extensive experimentation evaluates different agent configurations, highlighting the impact of behavior types on auction dynamics.
Compiler Optimization Study
Study of different optimization levels used by the GCC compiler, with particular focus on loop unrolling optimization techniques. The project involved implementation, experimentation, and detailed explanation of compiler optimization mechanisms.
Multicore Programming and SIMD Extensions
Implementation of an algorithm with floating-point vectors using different optimization levels and study of the results. The project focused on parallel programming techniques and SIMD (Single Instruction Multiple Data) extensions.