04August2025
10:00 Master's Defense Room 85 of IC2
Topic on
Synthetic Driving Conditions Data Generation Using Federated Generative Adversarial Networks
Student
David de Melo Almeida dos Reis
Advisor / Teacher
Allan Mariano de Souza
Brief summary
Road safety remains a global challenge, especially in scenarios where behavioral and environmental factors strongly influence drivers’ decision-making. Machine learning models play a crucial role in promoting safety and making informed decisions by learning effective actions based on traffic conditions. However, training these models requires access to user data, which can compromise drivers’ privacy and expose sensitive information. In order to overcome this problem, this study proposes a solution for generating synthetic driving condition data using a Federated Learning approach combined with Generative Adversarial Networks (GAN). This method allows training distributed models across multiple federated clients, preserving data privacy by avoiding direct data sharing. Using the Harmony dataset, similarity metrics such as Euclidean Distance and Kullback-Leibler Divergence were incorporated into the GAN loss function to improve the quality of the generated synthetic data. The results demonstrate that the proposed approach is capable of generating realistic data on driving conditions, enabling centralized model training while maintaining user privacy, highlighting its potential for road safety applications with a focus on privacy protection.
Examination Board
Headlines:
| Allan Mariano de Souza | IC / UNICAMP |
| Luiz Fernando Bittencourt | IC / UNICAMP |
| Joahannes Bruno Dias da Costa | ICT / UNIFESP |
Substitutes:
| Juliana Freitag Borin | IC / UNICAMP |
| Vinícius Fernandes Soares Mota | CT/UFES |