Fitting smooth functions to data and theoretical understanding of neural nets
Abstract
The PI, Dr. Charles Fefferman, is on a five year plan for two goals- (1) to identify efficient algorithms for fitting smooth functions and manifolds to data, and to prove mathematical theorems underlying these algorithms that conform to classes of data, and (2) to understand mathematically why neural networks can be trained effectively. This proposal is of enormous value from the research results obtained, both the theoretical work for fitting functions to data (e.g., manifold learning in statistics learning) and the applied work identifying useful-useable algorithms in machine learning.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Feb 29, 2024
- Source ID
- FA95502310273
Entities
People
- Charles Fefferman
Organizations
- Air Force Office of Scientific Research
- Trustees of Princeton University
- United States Air Force