DEEP NEURAL NETWORK OPTIMIZATION SYSTEM FOR MACHINE LEARNING MODEL SCALING

Title:

DEEP NEURAL NETWORK OPTIMIZATION SYSTEM FOR MACHINE LEARNING MODEL SCALING

Document Type and Number:

Japanese Patent JP2023060820

Kind Code:

A

Abstract:

To provide an apparatus, method and program for optimizing machine learning models.SOLUTION: In a joint optimization framework 100 using sparse distillation, an apparatus 102 for sparse distillation of machine learning models comprises: a knowledge distillation mechanism configured to distill knowledge of a supernet into a subnet during a single machine learning training epoch; and a pruning mechanism configured to prune one or more parameters from the subnet during the single machine learning training epoch to produce a sparse distilled subnet.SELECTED DRAWING: Figure 1

More Like This:

WO/2020/054164	OPERATION ASSISTANCE SYSTEM AND METHOD, AUTOMATIC PLANNER, AND COMPUTER READABLE MEDIUM
JP7101287	Inference device, inference system, inference method, and program
WO/2024/085577	METHOD FOR LEARNING AND CORRECTING SCREEN PRINTING PATTERN USING ARTIFICIAL INTELLIGENCE, AND SYSTEM AND PROGRAM FOR PERFORMING SAME

Inventors:

SAIRAM SUNDARESAN
SOUVIK KUNDU

Application Number:

JP2022147630A

Publication Date:

April 28, 2023

Filing Date:

September 16, 2022

Export Citation:

Click for automatic bibliography generation Help

Assignee:

INTEL CORP

International Classes:

G06N20/00

Attorney, Agent or Firm:

Tadashige Ito
Tadahiko Ito
Osamu Miyazaki
Naoki Fujimura

Previous Patent: Internal combustion engine including variable compression ratio device

Next Patent: Bifocals and bifocals with oblique light