WO2022247878 - SYSTEM, METHOD, AND COMPUTER-READABLE MEDIA FOR LEAKAGE CORRECTION IN GRAPH NEURAL NETWORK BASED RECOMMENDER SYSTEMS
National phase entry is expected:
Publication Number
WO/2022/247878
Publication Date
01.12.2022
International Application No.
PCT/CN2022/095106
International Filing Date
26.05.2022
Title **
[English]
SYSTEM, METHOD, AND COMPUTER-READABLE MEDIA FOR LEAKAGE CORRECTION IN GRAPH NEURAL NETWORK BASED RECOMMENDER SYSTEMS
[French]
SYSTÈME, PROCÉDÉ ET SUPPORT LISIBLE PAR ORDINATEUR POUR UNE CORRECTION DE FUITE DANS DES SYSTÈMES DE RECOMMANDATION BASÉS SUR UN RÉSEAU NEURONAL GRAPHIQUE
Applicants **
HUAWEI TECHNOLOGIES CO., LTD.
Inventors
KUMAR, Ishaan
HU, Yaochen
ZHANG, Yingxue
Priority Data
17/824,556
25.05.2022
US
63/193,576
26.05.2021
US
Application details
| Total Number of Claims/PCT | * |
| Number of Independent Claims | * |
| Number of Priorities | * |
| Number of Multi-Dependent Claims | * |
| Number of Drawings | * |
| Pages for Publication | * |
| Number of Pages with Drawings | * |
| Pages of Specification | * |
| * | |
| Number of Office Actions | * |
| * | |
International Searching Authority |
CNIPA
* |
| Recordal of a Change of the Applicant's Name/Address |
Change of Applicant's Name and Address
* |
| Type of Assignment |
The Standard Agent's Assignment
* |
| Applicant's Legal Status |
Legal Entity
* |
| * | |
| * | |
| * | |
| * | |
| * | |
| Entry into National Phase under |
Chapter I
* |
| Patent Delivery |
Send the Letters Patent by Courier
* |
| Translation |
|
* The data is based on automatic recognition. Please verify and amend if necessary.
** IP-Coster compiles data from publicly available sources. If this data includes your personal information, you can contact us to request its removal.
Quotation for National Phase entry
| Country | Stages | Total | |
|---|---|---|---|
| China | Filing, Examination, Granting | 1966 | |
| EPO | Filing, Examination, Granting | 14228 | |
| Japan | Filing, Examination, Granting | 2307 | |
| South Korea | Filing, Examination, Granting | 2502 | |
| USA | Filing, Examination, Granting | 5340 |
Total:
26,343
The term for entry into the National Phase has expired. This quotation is for informational purposes only
Contact Us
Abstract[English]
Systems, methods, and computer-readable media provide a graph processing system that incorporates a graph neural network (GNN) based recommender system (RS), as well as a method for training a GNN based RS to address feature leakage that leads to overfitting of the trained GNN based RS. A message correction algorithm is used to modify a user node embedding and a positive item node embedding generated by the graph neural network when generating mini batches of training triples used to train the GNN based RS. The GNN message passing operations are performed on one graph only, in contrast to existing approaches which typically run GNN message passing operations on multiple adjusted input graphs constructed for multiple training triples.[French]
Des systèmes, des procédés et des supports lisibles par ordinateur fournissent un système de traitement graphique qui incorpore un système de recommandation (RS) basé sur un réseau neuronal graphique (GNN), ainsi qu'un procédé d'apprentissage d'un RS basé sur un GNN pour résoudre le problème de fuite de caractéristiques qui conduit à un surapprentissage du RS basé sur un GNN entraîné. Un algorithme de correction de message est utilisé pour modifier une incorporation de nœud d'utilisateur et une incorporation de nœud d'élément positif générées par le réseau neuronal graphique lors de la génération de mini-lots de triplets d'apprentissage utilisés pour entraîner le RS basé sur un GNN. Les opérations de passage de message GNN sont effectuées sur un graphique uniquement, contrairement aux approches existantes qui exécutent typiquement les opérations de passage de message GNN sur de multiples graphiques d'entrée ajustés construits pour de multiples triplets d'apprentissage.