Math @ Duke
|
Publications [#357493] of Rong Ge
Papers Published
- Wang, X; Wu, C; Lee, JD; Ma, T; Ge, R, Beyond lazy training for over-parameterized tensor decomposition,
Advances in Neural Information Processing Systems, vol. 2020-December
(January, 2020)
(last updated on 2024/04/25)
Abstract: Over-parametrization is an important technique in training neural networks. In both theory and practice, training a larger network allows the optimization algorithm to avoid bad local optimal solutions. In this paper we study a closely related tensor decomposition problem: given an l-th order tensor in (Rd)?l of rank r (where r « d), can variants of gradient descent find a rank m decomposition where m > r? We show that in a lazy training regime (similar to the NTK regime for neural networks) one needs at least m = ?(dl-1), while a variant of gradient descent can find an approximate tensor when m = O*(r2.5l log d). Our results show that gradient descent on over-parametrized objective could go beyond the lazy training regime and utilize certain low-rank structure in the data.
|
|
dept@math.duke.edu
ph: 919.660.2800
fax: 919.660.2821
| |
Mathematics Department
Duke University, Box 90320
Durham, NC 27708-0320
|
|