Patent · US Active

Filtering in trainable networks

US11132619B1 · kind B1 · utility

15Cited by

1References

18Claims

0Family size

Assignee

Cadence Design Systems, Inc. · US

Inventors

Raul Alejandro Casas · Doylestown, US
Samer Hijazi · Bethlehem, US
Piyush Kaul · New Delhi, IN
Rishi Kumar · New York, US
Xuehong Mao · San Jose, US
Christopher Rowen · Santa Cruz, US

Key dates

Filing date	Feb 24, 2017
Grant date	Sep 28, 2021
Priority date	—
Expiry date	Nov 17, 2039

Classification

Technology area (CPC G)Physics
CPC primaryG06N3/082
WIPO fieldComputer technology
WIPO sectorElectrical engineering

Abstract

Some embodiments perform, in a multi-layer neural network in a computing device, a convolution operation on input feature maps with multiple convolutional filters. The convolutional filters have multiple filter precisions. In other embodiments, electronic design automation (EDA) systems, methods, and computer-readable media are presented for adding such a multi-layer neural network into an integrated circuit (IC) design.

Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.