Back to results
US 2026/0228308 A1Published application
Optimizing parameter estimation for training neural networks
AssigneeNVIDIA Corporation
InventorsDaguang Xu, Holger Reinhard Roth, Dong Yang, Andriy Myronenko, Ali Hatamizadeh, Vishwesh Nath, Anas Zainul Abidin
Life of this patent
Estimating the term…
Abstract
Apparatuses, systems, and techniques estimate parameters to train one or more neural networks based on uniqueuss of training data. In at least one embodiment, a subset of training data is selected and used to estimate parameters to train one or more neural networks, based on, for example, uniqueness of training data.
Claims21
-31. (canceled)
One or more processors comprising:
- circuitry to:
- use one or more first neural networks to estimate one or more hyperparameters for one or more second neural networks based, at least in part, on a subset of training data to be used by the one or more second neural networks;
- and
- update one or more parameters of the one or more second neural networks using the estimated one or more hyperparameters.
1 more element
A system, comprising:
- one or more processors to:
- use one or more first neural networks to estimate one or more hyperparameters for one or more second neural networks based, at least in part, on a subset of training data to be used by the one or more second neural networks;
- and
- update one or more parameters of the one or more second neural networks using the estimated one or more hyperparameters.
1 more element
A method, comprising:
- using one or more first neural networks to estimate one or more hyperparameters for one or more second neural networks based, at least in part, on a subset of training data to be used by the one or more second neural networks;
- and
- updating one or more parameters of the one or more second neural networks using the estimated one or more hyperparameters.
Dependencies are inferred from the claim text.