Inria THOTH project team

Ongoing Ph.D on the topic of self-supervised and semi-supervised learning methodsapplied to dense visual recognition in remote sensing.Cifre collaboration with Inria Grenoble and GIPSA-lab.

Research internship in deep learning applied to audio, at LIP6 in the MLIA research team. Acoustic scene and sound event classification with convolutional neural networks.I developed deep F-CNNs architectures to classify urban sound scenes, and assessed them on datasets of the DCASE 2019 challenge. The datasets were of varying difficulties (closed-set, open-set classification, mismatched recording devices).I experimented with data augmentation techniques (mixup, between-class learning) to increase generalization of models, because of the multiple challenges of learning from sound data.

Tutor to first year computer science students to assist with introductory programming courses.