Projects

This project was developed in the context of the curricular unit “Introduction to Robotics” during my Masters Degree. The project was divided into two parts. The first focuses on mobile robot localization, using Bayesian filters to manage uncertainties in perception and action. The second addresses path planning and following, applying Rapidly Exploring Random Tree (RRT) algorithms. The project was done in groups of 3, and it was attributed a grade of 18/20.

This project was developed in the context of the curricular unit “Interaction Design Studio” during my Masters Degree. The objective was to co-design something with someone from a minority group in order to improve their quality of life and make them feel seen, while also providing a solution to a problem they might be facing. The project was done in groups of 5, and it was attributed a grade of 19.2/20.

This project was developed in the context of the curricular unit “Ambient Intelligence” during my Master Degree. The smart pill organizer is able to detect the medication that is being taken and send the information to a cloud service. This information is then used to send reminders to the user and to keep track of the medication inventory. The objective is to alleviate the struggle of medication adherence by uniting a traditional pillbox with sensors and actuators. The project was done in groups of 3, and it was attributed a grade of 19/20.

This project was developed in the context of the curricular unit “Data Science” during my Master Degree. The project followed the KDD process and classify the data/forecast the future in four datasets - Covid-Pos and Credit Score classification; Covid deaths and traffic intensity forecasting. In both situations the models were trained from the available data, by adequately selecting and preparing it, followed by the assessment of the learnt models. The project contains an extensive report criticizing the results achieved and hypothesized the causes for better and worse performances by the learnt models. The goal was to understand the impact of the available options on the produced models’ performance. The project was done in groups of 4 and it was attributed a grade of with 17.09/20.

This project was developed in the context of the curricular unit “Deep Learning” during my Master Degree. The objective was to implement a medical image classification with linear classifiers and neural networks and autodiff toolkit using the language Python. In the implementation with linear classifiers and neural networks no machine learning library (such as scikit-learn) was used, and the solution relied solely on Linear Algebra, using the numpy library to do so. The classification results are covered in a detailed project encompassing its development topics. The project was done in groups of 2 and was classified on 20/20.

This project was developed in the context of the curricular unit “Natural Language Processing” during my Master Degree. The objective was to implement diverse classifiers such as Logistic Regression, Support Vector Classifier, and Naïve Bayes which are then compared with the objective of classifying hotel reviews into four categories - positive-deceptive, negative-deceptive positive-truthful, and negative-truthful. The results are covered in a detailed project encompassing its development topics such as the models used, the experimental setup, and discussion section where I analyzed the dataset and the obtained results in order to identify common errors in the classification and a future work section. The project was done in groups of 2 and was classified with 19/20 - 6/5 was attributed to the code (+1 point for achieving an accuracy bigger than 85%) and 13/15 for the report.

This project was developed in the context of the curricular unit “Natural Language Processing” during my Master Degree. The objective was to implement a date reading module using only transducers, which converts dates in specific format to the corresponding sentence as it would appear if transcribed by a human. The module is based on the idea of transducers, which are a type of data structure that can be used to process data in a streaming fashion. The project was done in groups of 2, and was graded in 19.5/20, achieving a grade of 17/17 in the code and 3/3 for the report. We received a penalty of -0.5 for the report filename not being in desired format.

This project was developed in the context of the curricular unit “Artificial Intelligence” during my Bachelor Degree. The objective was to solve a Takuzu problem using AI search algorithms - DFS, BFS, Greedy Search, and A* Search. A report which encompasses the description of the problem, the solution that was found and the compilation of the tests issued to compare all algorithms, is available. The project was done in groups of 2, and it was attributed a grade of 17.68/20 - 15/15 to the code and 2.68/5 for the report.

This project was developed in the context of the curricular unit “Human-Computer Interaction” during my Bachelor Degree. The objective was to conceptualize the design for a Bitcoin wallet application, which is aimed at helping users to manage their cryptocurrencies. The project was done in groups of 3 and was classified on 19/20.

This project was developed in the context of the curricular unit “Analysis and Synthesis of Algorithms” during my Bachelor Degree. The objective was to implement an algorithm which evaluates input nodes and outputs if they create a valid genealogy tree. The algorithm is based on the concept of a binary search tree and was implemented using C++. The project was done in groups of 2, and it was attributed a grade of 19.5/20 - 17/17 to the code and 2.5/3 for the report.