BSc and MSc theses

I look for motivated, curious students eager to tackle relevant, cutting-edge research on AI and media perception.

What I offer: computational resources, individualised supervision and a kind environment.

Modelling persuasion strategies in advertising

Multimodal AI · machine learning · Python · research

Every time we watch television or browse online we are exposed to ads designed to convince us to buy something. This project builds automatic systems that encourage a more responsible consumption of advertising by detecting the persuasion strategies each ad relies on. The approach is multimodal machine learning: models that process, understand and analyse several data sources at once.

Three modalities enter the model together; the persuasion strategies come out.

Tasks

Survey the state of the art in machine learning and neuromarketing.
Implement models that predict persuasion strategies in ads, using Python and its frameworks (PyTorch, Huggingface, ModelScope-Swift).
Evaluate the proposed solutions with objective metrics.

Requirements

Initiative and genuine interest in artificial intelligence and machine learning, and in building state-of-the-art neural solutions.
Comfortable writing scripts and small projects in Python.
Able to pick up new AI frameworks quickly.
Basic knowledge of image and video processing.

A plus

Grounding in the theory and practice of artificial intelligence and language models.
Experience with Python's AI libraries (pandas, scikit-learn, pytorch…).

Predicting multimedia memorability from brain signals

Biomedical signals · machine learning · Python · research

This project uses machine learning to better understand our brain response while watching a video, taking the electroencephalogram (EEG) as input and memory as the central question. Can we tell whether someone remembered a video from their EEG? What role do different brain areas and frequency bands play in our ability to remember it?

The EEG recorded during viewing feeds a model that predicts whether the video will be remembered.

Tasks

Survey the previous literature on predicting memorability from brain signals.
Analyse and preprocess the EEG data.
Implement computational models that predict video memorability from the processed signals.
Evaluate the implemented models with objective metrics.

Requirements

Initiative and genuine interest in the topics of the project, especially artificial intelligence and machine learning.
Able to read, understand and write code fluently.
Basic knowledge of biomedical signal processing, EEG in particular.

A plus

Grounding in the theory and practice of artificial intelligence and its application to biomedical signals.
Critical thinking, curiosity and creativity.
Experience with Python's AI and neurotech libraries (pandas, scikit-learn, pytorch, python-mne…).

Computational analysis of persuasive design in digital gambling

Multimodal AI · machine learning · Python · research

Digital gambling and betting — casinos, sports betting and micro-bets embedded in video games — are an increasingly common form of entertainment among young adults, and carry real risks to financial and social wellbeing when play stops being responsible. This project builds automatic tools that detect those risk factors in the interfaces and multimedia elements of digital gambling.

Interface elements are analysed to flag the risk factors they carry.

Tasks

Survey the previous literature on multimodal analysis of digital gambling.
Collect, analyse and preprocess data sources relevant to the task.
Implement computational models that predict risk features: persuasion, intermittent rewards, misleading promotions.
Evaluate the implemented models with objective metrics.

Requirements

Initiative and genuine interest in artificial intelligence and machine learning, and in building state-of-the-art neural solutions.
Comfortable writing scripts and small projects in Python.
Able to pick up new AI frameworks quickly.
Basic knowledge of image and video processing.

A plus

Grounding in the theory and practice of artificial intelligence and its application to the multimodal analysis of images, video and interfaces.
Critical thinking, curiosity and creativity.
Experience with Python's AI libraries (pandas, scikit-learn, pytorch…).