Processing

Please wait...

PATENTSCOPE will be unavailable a few hours for maintenance reason on Tuesday 25.01.2022 at 12:00 PM CET
Settings

Settings

Goto Application

1. WO2021244912 - APPARATUS, METHOD AND COMPUTER PROGRAM FOR ACCELERATING GRID-OF-BEAMS OPTIMIZATION WITH TRANSFER LEARNING

Publication Number WO/2021/244912
Publication Date 09.12.2021
International Application No. PCT/EP2021/064010
International Filing Date 26.05.2021
IPC
G06N 3/04 2006.1
GPHYSICS
06COMPUTING; CALCULATING OR COUNTING
NCOMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
3Computer systems based on biological models
02using neural network models
04Architecture, e.g. interconnection topology
G06N 3/08 2006.1
GPHYSICS
06COMPUTING; CALCULATING OR COUNTING
NCOMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
3Computer systems based on biological models
02using neural network models
08Learning methods
H04B 7/0456 2017.1
HELECTRICITY
04ELECTRIC COMMUNICATION TECHNIQUE
BTRANSMISSION
7Radio transmission systems, i.e. using radiation field
02Diversity systems; Multi-antenna systems, i.e. transmission or reception using multiple antennas
04using two or more spaced independent antennas
0413MIMO systems
0456Selection of precoding matrices or codebooks, e.g. using matrices for antenna weighting
H04B 7/06 2006.1
HELECTRICITY
04ELECTRIC COMMUNICATION TECHNIQUE
BTRANSMISSION
7Radio transmission systems, i.e. using radiation field
02Diversity systems; Multi-antenna systems, i.e. transmission or reception using multiple antennas
04using two or more spaced independent antennas
06at the transmitting station
CPC
G06N 3/0454
GPHYSICS
06COMPUTING; CALCULATING; COUNTING
NCOMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
3Computer systems based on biological models
02using neural network models
04Architectures, e.g. interconnection topology
0454using a combination of multiple neural nets
G06N 3/08
GPHYSICS
06COMPUTING; CALCULATING; COUNTING
NCOMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS
3Computer systems based on biological models
02using neural network models
08Learning methods
H04B 7/0617
HELECTRICITY
04ELECTRIC COMMUNICATION TECHNIQUE
BTRANSMISSION
7Radio transmission systems, i.e. using radiation field
02Diversity systems; Multi-antenna system, i.e. transmission or reception using multiple antennas
04using two or more spaced independent antennas
06at the transmitting station
0613using simultaneous transmission
0615of weighted versions of same signal
0617for beam forming
Applicants
  • NOKIA TECHNOLOGIES OY [FI]/[FI]
Inventors
  • LIAO, Qi
  • SYED MUHAMMAD, Fahad
  • CAPDEVIELLE, Veronique
  • FEKI, Afef
  • KALYANASUNDARAM, Suresh
  • MALANCHINI, Ilaria
Agents
  • NOKIA EPO REPRESENTATIVES
Priority Data
2020556501.06.2020FI
Publication Language English (en)
Filing Language English (EN)
Designated States
Title
(EN) APPARATUS, METHOD AND COMPUTER PROGRAM FOR ACCELERATING GRID-OF-BEAMS OPTIMIZATION WITH TRANSFER LEARNING
(FR) APPAREIL, PROCÉDÉ ET PROGRAMME INFORMATIQUE DESTINÉS À ACCÉLÉRER UNE OPTIMISATION DE GRILLE DE FAISCEAUX AVEC APPRENTISSAGE PAR TRANSFERT
Abstract
(EN) A deep transfer reinforcement learning (DTRL) method based on transfer learning within a deep reinforcement learning (DRL) framework is provided to accelerate the GoB optimization decisions when experiencing environment changes in the same source radio network agent or when being applied from a source radio network agent to a target radio network agent. The transferability of the knowledge embedded in a pre-trained neural network model as a Q-approximator is exploited, and a mechanism to transfer parameters from a source agent to a target agent is provided, where the transferability criterion is based on the similarity measure between the source and target domain.
(FR) Un procédé d'apprentissage profond de renforcement par transfert (DTRL), sur la base d'un apprentissage par transfert dans un framework d'apprentissage profond par renforcement (DRL), est fourni pour accélérer des décisions d'optimisation GoB lorsqu'il rencontre des changements environnementaux dans le même agent de réseau radio source ou lorsqu'il est appliqué d'un agent de réseau radio source à un agent de réseau radio cible. La possibilité de transfert des connaissances intégrées dans un modèle de réseau de neurones artificiels pré-entraîné, en tant qu'approximateur Q, est exploitée et un mécanisme de transfert de paramètres, d'un agent source à un agent cible, est fourni, le critère de possibilité de transfert étant basé sur la mesure de similarité entre la source et le domaine cible.
Latest bibliographic data on file with the International Bureau