I'm Valentina, a post-doctoral researcher in Machine Learning at INRIA and University College London, in the context of the INRIA-London Programme, working with Benjamin Guedj and Pascal Germain on theoretically grounded unsupervised learning.
My research interests cover representation learning, local learning and decentralized learning, adversarial robustness and ML applied to climate science.
I received my PhD in Computer Science from Jean Monnet University (Saint-Etienne, France), in the Data Intelligence team of Hubert Curien Lab., advised by Marc Sebban and Rémi Emonet. My PhD focused on kernel learning with theoretical guarantees on performance. In particular, studying new data-dependent ways for approximating the Gram matrix and for learning the kernel function as well, through the selection of representing inputs.
Since 2019, I am involved in FDL research accelerator, working on Deep Learning applied to climate science and Earth Observational data, in particular for performing cloud classification and weather forecasting at global scale, and ML on-board of satellites. An outcome of these projects was the release of CUMULO dataset.
In 2017, I worked as an intern at IBM Research, Dublin, in Mathieu Sinn's team, on studying and building Deep Learning architectures robust to adversarial examples. The library developped for this research work served as codebase of the initial release of Adversarial Robustness Toolbox.
Previously, I graduated in 2015 from INSA de Lyon, France, in Computer Science. I am originally from Italy.