14/10/2024

USP event discusses advances in deep learning.

The event 'Understanding Deep Learning 2024' started this Monday (14) and will run until Friday (18), at the São Carlos (SP) campus of USP (University of the State of São Paulo). In its 2nd edition, the workshop discusses advances and main research in deep learning , an area of artificial intelligence that allows the resolution of highly complex problems. Registration is open on the initiative's website and the event will be broadcast on the Data and ICMC TV channels on YouTube.

The event was organized by the Data extension group of the Institute of Mathematical and Computer Sciences (ICMC) at USP. The program includes 12 lectures and is open to undergraduate and graduate students, as well as researchers and others interested in the field. Topics such as Multilingual LLMs and Multiresolution Image Representation using Neural Networks, among others, will be discussed.

IMPA doctoral student Hallison Paz is one of the participants in the event. Having previously worked at Meta (Facebook's parent company), Paz specializes in visual computing, graphics, and image processing. In an interview with IMPA, he shared part of his career path and the opportunity he had with the Meta group.

This Tuesday (15), Paz will give the lecture 'Representation of images in multiresolution', based on his doctoral thesis. The computer engineer will discuss the operation of a neural network architecture developed by him, MR-Net, focusing on applications of image representation in multiple resolutions. The work is the result of the ongoing research of Visgraf (Computer Graphics Laboratory of IMPA) on neural media.

For Paz, the USP event is an “excellent opportunity to share the trajectory of this project, discuss ideas with other experts, and explore new avenues of research.”

The list of speakers includes researchers from leading institutions around the world, such as Jeremy Bernstein and Hannes Stärk from the Massachusetts Institute of Technology (MIT), Petar Veličković from the University of Cambridge, and Mathilde Papillon from the University of California. The full program is available on the event website.

The event can be followed online through the Data and ICMC TV channels on YouTube. Certificates will be offered to those who register in advance and participate in at least 50% of the online lectures. Click here to register.