Modifié le
Séminaires
Gravity and Quantum Mechanics
Manu PARANJAPE, Professeur à l'Université de Montréal, Canada
Modifié le
Séminaire de Manu PARANJAPE, Professeur à l'Université de Montréal, Canada
This talk aims to emphasize the crucial role of cutting-edge data-driven modeling techniques in constructing compact and fast-to-evaluate surrogate models of electronic devices and circuits for stochastic analysis and optimization purposes. Initially, several modeling techniques, such as least-squares regressions, kernel regression, Artificial Neural Networks, and advanced solutions for vector-valued problems, will be briefly presented to highlight their advantages, capabilities, and limitations. Subsequently, in the second part of the presentation, the effectiveness and strength of these techniques will be showcased through applications in uncertainty quantification, optimization, and parametric modeling across various test cases.