Neural ODEs - From ResNet to Continuous-Time Machine Learning
Offered By: Steve Brunton via YouTube
Course Description
Overview
Explore Neural ODEs (NODEs), a powerful machine learning approach for learning ODEs from data, in this 25-minute video lecture. Delve into the background of ResNet, understand the transition from ResNet to ODE, and discover why ODEs outperform ResNet. Compare ODE and ResNet performance, and learn about ODE extensions such as Hamiltonian Neural Networks (HNNs) and Lagrangian Neural Networks (LNNs). Gain insights into the ODE algorithm overview, including ODEs and Adjoint Calculation. Produced at the University of Washington with funding support from the Boeing Company, this comprehensive lecture covers essential concepts in Physics Informed Machine Learning.
Syllabus
Intro
Background: ResNet
From ResNet to ODE
ODE Essential Insight/ Why ODE outperforms ResNet
// ODE Essential Insight Rephrase 1
// ODE Essential Insight Rephrase 2
ODE Performance vs ResNet Performance
ODE extension: HNNs
ODE extension: LNNs
ODE algorithm overview/ ODEs and Adjoint Calculation
Outro
Taught by
Steve Brunton
Related Courses
Dynamical Modeling Methods for Systems BiologyIcahn School of Medicine at Mount Sinai via Coursera Linear Differential Equations
Boston University via edX Equations différentielles : de Newton à nos jours
Sorbonne University via France Université Numerique Matlab Programming for Numerical Computation
Indian Institute of Technology Madras via Swayam Introduction to Dynamical Models in Biology
Indian Institute of Technology Guwahati via Swayam