Preprint on model reduction on nonlinear manifolds using deep convolutional autoencoders now available
Using deep learning to overcome Komolgorov-width limitation
My work with Kookjin Lee on performing model reduction on nonlinear manifolds using deep convolutional autoencoders is now available on the arXiv .
I'm particularly excited about this work, as it is—to our knowledge—the first method that demonstrates how dimensionality reduction using deep learning can be integrated in an optimal manner to reduce the dimensionality of nonlinear dynamical systems.
Kevin T. Carlberg
Director of AI Research Science
My research combines machine learning, computational physics, and high-performance computing. The objective is to discover structure in data to drastically reduce the cost of simulating nonlinear dynamical systems at extreme scale. I also work on technologies that enable the future of virtual and augmented reality.