Research
Currently, my research focuses on the development of generative artificial intelligence models for data augmentation and diversification in remote sensing. I work on the development of generative models such as diffusion models (DDPM) and have worked on GANs, adapting them to contexts, data sparsity and data imbalance. In parallel, I am starting to explore structural analysis and interpretability approaches with the aim of better understanding how foundational models behave in complex geospatial environments, and thus facilitating their efficient and reliable adaptation to concrete applications. My medium-term goal is to build a research framework that combines synthetic data generation, dynamic analysis of neural architectures and quantitative explainability in the context of Earth observation.
Research Areas
Data augmentation for remote sensing
Techniques to generate synthetic data for remote sensing applications, with a focus on solving the data scarcity and data imbalance.
Model compression
Efficient adaptation of large foundation models through structured pruning and similarity analysis, enabling faster inference and training while maintaining performance.