TFTs

Suggested final-degree project topics, adaptable to each student's interests.

These ideas are starting points: the scope and methods can be adapted to each student's interests and project level.

Mental health and emotional well-being

  • Detecting signals of stress and well-being in text

    Apply natural language processing and machine learning to identify emotional indicators in text, and study model robustness and explainability.

  • Agent-based simulation of health interventions

    Design an agent-based simulation to compare potential well-being interventions and policies, analysing their effects across different population groups.

  • Emotion-aware virtual assistants

    Explore how language models and environmental adaptation can be combined to create personalised support systems, and evaluate their behaviour in controlled scenarios.

  • Virtual reality for clinical training

    Prototype virtual reality experiences for training or controlled exposure, and analyse how interaction and context affect emotional experience.

Art and emotion

  • Recommending artworks in relation to mood

    Build a recommender connecting visual and emotional features of artworks with people's preferences, and compare personalisation strategies.

  • Can AI recognise emotional nuance in art?

    Compare vision and language models on their emotional interpretation of artworks, identifying where their judgements align with people and where they differ.

  • Music recommendation for emotional regulation

    Investigate music recommendation methods that account for a person's context and emotional goals.

Urban planning and mobility

  • Optimising shared-vehicle fleets

    Use mobility data and optimisation or machine-learning methods to improve shared-vehicle distribution as demand changes.

  • Simulating interventions in urban environments

    Build an agent-based simulation to study how different urban-planning decisions may affect mobility and access to services.

Sports and physical activity

  • Predicting performance from physical-activity data

    Analyse activity data to develop performance-prediction models and investigate which variables provide useful, interpretable information.

  • Personalised training plans

    Explore optimisation and machine-learning methods to adapt training plans to each person's progress, goals, and availability.

If a proposal interests you, or you have a related idea, feel free to get in touch. You can also browse previous GSI projects.