Use of mathematical methods to inform NN parameter selection (width, depth, latent space dimension, etc.) and design of NNs intended to operate on inputs with mathematical structure (MIND)

PI: Sarah Scullen, National Security Directorate

MIND uses mathematical methods to inform NN parameter selection (width, depth, latent space dimension, etc.) and design of NNs intended to operate on inputs with mathematical structure (novel tokenization, complex valued data, subspace learning, etc.)

PNNL

Generative AI

Use of mathematical methods to inform NN parameter selection (width, depth, latent space dimension, etc.) and design of NNs intended to operate on inputs with mathematical structure (MIND)