INVITED RESEARCHER: JAD MOUNAYER
-SUMARY: THIS TALK INCLUDES A FEW ADVANCED TOPICS IN ARTIFICIAL INTELLIGENCE.
TO BEGIN WITH, A FEW IMPORTANT TIPS ABOUT TRAINING NEURAL NETWORKS ARE
DISCUSSED. AFTERWARDS, THE MACHINE LEARNING LIBRARY JAX IS PRESENTED,
FOLLOWED BY SOME CODING LIVE THAT EXPLAINS HOW JIT COMPILATION IS DONE AND
WHY A DEVELOPER WOULD CHOOSE JAX OVER OTHERS (E.G. PYTORCH, TENSORFLOW).
THE TALK WILL ALSO INCLUDE NOVEL STATE OF THE ART ARCHITECTURES IN BOTH TIME
SERIES PREDICTIONS AND GENERATIVE DESIGN. FOR TIME SERIES, NEURAL CDES AND PU-
NODES ARE PRESENTED. FOR THE LATTER, RANK REDUCTION AUTOENCODERS ARE
EXPLAINED ALONG WITH HOW TO PERFORM INTERPOLATION, CLASSIFICATION, OR
GENERATIVE DESIGN USING THIS ARCHITECTURE.
WHEN:
If you are a student: September 26 from 12:30 p.m. to 1:30 p.m.
ETSIADI classroom S01
If you are a researcher: September 27 from 12:00 to 13:00 h.
Library of the Institute of Design for Manufacturing (IDF) CPI 8G. under