Graph-based Algorithms in Computer Vision and Machine Learning: Theory and Applications

Orateur : Jhony Giraldo
17 Mars 2022 à 13:30

Le prochain séminaire de l’équipe A3SI du LIGM aura lieu jeudi 17 mars à 13h30 en mode hybride. Jhony Giraldo nous présentera ses travaux intitulés Graph-based Algorithms in Computer Vision and Machine Learning: Theory and Applications (voir résumé ci-dessous). 

Le séminaire aura lieu :

– à ESIEE Paris dans l’amphithéâtre 160 ; et
– via https://meet.google.com/ioh-krdx-uqo , pour ceux qui ne pourront pas être présents.
La présentation sera en anglais.

Title: Graph-based Algorithms in Computer Vision and Machine Learning: Theory and Applications

Abstract: Graph representation learning and its applications have gained significant attention in recent years. Notably, Graph Neural Networks (CNN) and Graph Signal Processing (GSP) have been extensively studied. GNNs extend the concepts of Convolutional Neural Networks (CNNs) to non-Euclidean data modeled as graphs. Similarly, GSP extends the concepts of classical digital signal processing to signals supported on graphs. CNN and GSP have numerous applications such as semi-supervised learning, point cloud semantic segmentation, prediction of individual relations in social networks, modeling proteins for drug discovery, image, and video processing.

In this seminar, I will present our recent works on GSP and GNNs applied to some problems in computer vision and machine learning. The motivation of these studies is to leverage the structural information we can get from the data to effectively reduce the amount of labeled information required to solve the specific problem. I will focus on the applications of moving object segmentation, image classification, and weakly-supervised semantic segmentation. To that end, we map the particular problem to a graph, and subsequently, we apply GSP/GNN ideas to solve the learning problem. Our algorithms show competitive performance against state-of-the-art methods under the paradigm of learning with minimal supervision.  

Bio: Jhony H. Giraldo received a Bachelor in Electronics Engineering in 2016 from the Universidad de Antioquia, Colombia, and a Master of Science degree with honors at the same university in 2018. He spent 15 months at the University of Delaware, USA, between 2018 and 2019, working on Graph Signal Processing as a Research Assistant. He was visiting scholar at the Università degli Studi di Napoli Parthenope, Italy, working at the CVPR Lab “Alfredo Petrosino” in 2021. Currently, he is a final-year Ph.D. student in computer science at La Rochelle Université, Laboratoire MIA (Mathématiques, Image, et Applications), France. He is also a visiting PhD student at Centre de Vision Numérique (CVN), Inria OPIS at CentraleSupélec, Université Paris-Saclay, working on Deep Graph Neural Networks. His research interests include the fundamentals and applications of Graph Neural Networks, Computer Vision, Machine Learning, and Graph Signal Processing.