NEUROMORPHIC ARCHITECTURE DYNAMIC SELECTION METHOD FOR MODELING ON BASIS OF SNN MODEL PARAMETER, AND RECORDING MEDIUM AND DEVICE FOR PERFORMING SAME - FOUNDATION SOONGSIL UNIV INDUSTRY COOPERATION

Title:

NEUROMORPHIC ARCHITECTURE DYNAMIC SELECTION METHOD FOR MODELING ON BASIS OF SNN MODEL PARAMETER, AND RECORDING MEDIUM AND DEVICE FOR PERFORMING SAME

Document Type and Number:

WIPO Patent Application WO/2022/102912

Kind Code:

A1

Abstract:

A neuromorphic architecture dynamic selection method for modeling on the basis of SNN model parameters comprises the steps of: extracting information of an SNN model from an NPZ file received from an IoT application by means of a neuromorphic architecture abstraction (NAA) HOST; selecting at least one neuromorphic architecture capable of executing the SNN model, on the basis of the number of input dimensions, the number of output dimensions, and the maximum number of available neurons of the SNN model; calculating the time required for training of and inferences by the neuromorphic architecture if there are two or more selected neuromorphic architectures; and finally selecting a neuromorphic architecture having the shortest time required for training and making inferences. Accordingly, an optimal neuromorphic architecture capable of executing an SNN model can be dynamically selected.

Inventors:

HONG JIMAN (KR)
PARK KICHEOL (KR)
LEE SEUNGYEON (KR)
KIM BONGJAE (KR)

Application Number:

PCT/KR2021/010092

Publication Date:

May 19, 2022

Filing Date:

August 03, 2021

Export Citation:

Click for automatic bibliography generation Help

Assignee:

FOUNDATION SOONGSIL UNIV INDUSTRY COOPERATION (KR)

International Classes:

G06N3/063; G06N3/04; G06N3/08; G06N5/04

Foreign References:

US20200257961A1	2020-08-13
US20200302272A1	2020-09-24
US20190347548A1	2019-11-14
KR20200080296A	2020-07-06

Other References:

PARK KICHEOL, LEE YENA, HONG JIMAN, AN JAE-HOON, KIM BONGJAE: "Selecting a Proper Neuromorphic Platform for the Intelligent IoT", PROCEEDINGS OF THE INTERNATIONAL CONFERENCE ON RESEARCH IN ADAPTIVE AND CONVERGENT SYSTEMS, ACM, NEW YORK, NY, USA, 13 October 2020 (2020-10-13), New York, NY, USA, pages 255 - 260, XP055930572, ISBN: 978-1-4503-8025-6, DOI: 10.1145/3400286.3418264

Attorney, Agent or Firm:

YUN, Kuisang (KR)

Download PDF:

View/Download PDF PDF Help

Previous Patent: BEE-BOX-MOUNTABLE HORNET REPELLING DEVICE

Next Patent: WHITENING COSMETIC COMPOSITION COMPRISING EXOSOMES DERIVED FROM ROSE STEM CELLS AS ACTIVE INGREDIENT