公开(公告)号：US11120023B2

公开(公告)日：2021-09-14

申请号：US15929739

申请日：2020-05-19

Applicant: Huawei Technologies Co., Ltd.

Inventor： Yinglong Xia ,  Li Zhou ,  Ren Chen

IPC: G06F16/2455 ,  G06F16/2453 ,  G06F16/23 ,  G06F16/901 ,  G06F9/48 ,  G06F9/52

Abstract: A graph processing system for concurrent property graph queries of a property graph implemented in a distributed network computes on respective nodes a subgraph shard represented as edge-sets containing vertices within a certain range. Each node stores data for a subgraph shard that contains a range of local vertices that are a subset of all vertices of the property graph. Each subgraph shard also has boundary vertices having edges that connect the subgraph shard to boundary vertices of another subgraph shard. Upon receipt of concurrent queries of the property graph, a query of the subgraph shards is scheduled in accordance with an initial vertex for each concurrent user query. The property graph is traversed by traversing edge-sets within a subgraph shard on each node and during traversal messaging is used to send values of boundary vertices to at least one other node having another subgraph shard sharing the boundary vertices.

公开(公告)号：US20220121936A1

公开(公告)日：2022-04-21

申请号：US17434563

申请日：2019-02-27

Applicant: Huawei Technologies Co., Ltd.

Inventor： Zhicheng Sui ,  Li Zhou ,  Lei Zhao ,  Mohan Liu ,  Qinghua Yu ,  Hongrui Jiang

IPC: G06N3/08 ,  G06N3/04

Abstract: A neural network model processing method includes obtaining a first low-bit neural network model through training, where the model includes a first operation layer and a second operation layer. Each operation layer includes at least one operation. Values/a value of a parameter and/or data used for the operation are/is represented by using N bits, and N is a positive integer less than 8. The neural network model processing method further includes compressing the model to obtain a second low-bit neural network model, where the compressed model includes a third operation layer. The third operation layer is equivalent to the first operation layer and the second operation layer, and an operation layer other than the third operation layer in the at least one operation layer is the same as an operation layer other than the first operation layer and the second operation layer in the at least two operation layers.