Huawei Technologies Co., Limited

Tencent Technology (shenzhen) Company Limited

Shenzhen Jinzhicheng Plastic Technology Co., Ltd.

Tencent Technology Company Limited

Yong Cheng

Ping Fang

Gaokun Pang

Xiaoxian Li

Yangyu Tao

Fangcheng Fu

Gong Zhang

Yi Long

Qinghai Zeng

Qufang Huang

Chong Lou

Qiang Fan

Bo Li

Ning Wu

Yinghao Guo

Jie Jiang

Chen Hou

Yinghai He

Cen Ling

Xiaodong Bao

Zhan Zhang

Kaibin Huang

Pengfei Xia

Yong Qing Wang

Yang Liu

Qi Wang

Zhiming Ding

Cheng He

Yuhong Liu

Wei Wei

Bin Liu

Zhenguo Du

Zhaopeng Tu

Jingfei Lv

Kin Nang Lau

Yawei Zhang

Fandong Meng

Pin Xiao

Xiaoyong Yu

Junjie Zhai

Dayang Feng

Junwei Pan

Jianneng Liu

Jianping Xu

Zhicheng Cheng

Xiaoliang Fan

Huanran Xue

Yong Wang

Kin (Vincent) Nang Lau

Yinghai Hao

Embodiments of this application provide a training method, apparatus, and device for a federated neural network model, a computer program product, and a computer-readable storage medium. The method includes: performing forward computation processing of sample data through a first model, to obtain an output vector of the first model; performing forward computation processing through the interaction layer of the output vector of the first bottom model, at last one model parameter of an interaction layer, and at least one encrypted model to obtain an output vector of the interaction layer; performing forward computation processing on the output vector of the interaction layer through the second model, to obtain an output vector of the federated neural network model; back propagating the federated neural network model according to a loss result corresponding to the output vector of the federated neural network model, and updating the at least one model parameter of the interaction layer, a parameter of the first bottom model, and a parameter of the top model according to a back propagation processing result; and obtaining a trained federated neural network model based on the updated model parameters of the interaction layer, the first bottom model, and the second model.

Training method, apparatus, and device for federated neural network model, computer program product, and computer-readable storage medium

A data processing method includes: acquiring a first intersection set, acquiring a second intersection set, calculating an intersection between the first intersection set and the second intersection set to obtain an intersection result set that includes an intersecting portion of the first intersection data and the second intersection data, and obfuscating the intersection result set to obtain an obfuscation set that includes obfuscated data based on data in the second intersection set and an intersection data set based on the intersection result set.

Data processing method and apparatus, device, and computer-readable storage medium

A method for federated data query includes: using, by a first electronic device, a joint query statement corresponding to a federated data query to obtain at least one operator containing a joint query operator; transforming the joint query operator to a first joint security operator; determining first data for intersection on the first electronic device by executing the first joint security operator, receiving second data for intersection from a second electronic device involved in the federated data query, and performing joint computing on the first data for intersection and the second data for intersection in a form of a ciphertext to obtain a joint data table; and determining a federated data query result corresponding to the at least one operator based on the joint data table.

Method and apparatus for federated data query, electronic device, and computer-readable storage medium

This disclosure relates to a method for vertical federated learning. In multiple participation nodes deployed in a multi-way tree topology, an upper-layer participation node corresponds to k lower-layer participation nodes. After the upper-layer participation node and the k lower-layer participation nodes exchange public keys with each other, the upper-layer participation node performs secure two-party joint computation with the lower-layer participation nodes with a first public key and second public keys as encryption parameters to obtain k two-party joint outputs of a federated model. Further, the upper-layer participation node aggregates the k two-party joint outputs to obtain a first joint model output corresponding to the federated model. As such, a multi-way tree topology deployment-based vertical federated learning architecture is provided, improving the equality of each participation node in a vertical federated learning process.

Method and apparatus for vertical federated learning

A method for locating a clock fault includes that a network device locates a clock fault source based on whether first frequency offsets between a plurality of input clocks and a local clock exceed a frequency offset threshold, or based on whether second frequency offsets between the input clocks exceed a relative frequency offset threshold.

Method and network device for locating clock fault

A secure multi-party computation method and apparatus, a device, and a storage medium are provided, which belong to the field of data security technologies. The method includes: performing homomorphic encryption processing on first plaintext data to generate first ciphertext data; transmitting encrypted data containing the first ciphertext data to a second node device; receiving difference ciphertext data transmitted by the second node device; and decrypting the difference ciphertext data to obtain a positive and negative property of a difference between the first plaintext data and the second plaintext data. The foregoing method and apparatus, device, and storage medium are conducive to reducing the number of times of communications of secure multi-party computation, have low communication overhead and high computation efficiency, and enable magnitude comparison and equality testing to be performed simultaneously.

Secure multi-party computation method and apparatus, device, and storage medium

Educational toy

A random access method includes working, by a user equipment (UE), on a first bandwidth part (BWP), where the first BWP is a currently active BWP, when the UE needs to perform a random access procedure, the UE performs the random access procedure on an initial BWP or on the first BWP, and when the UE receives BWP switching indication information from a network device instructing the UE to use a second BWP when performing the random access procedure on the first BWP, continuing, by the UE, to perform the random access procedure on the first BWP, or stopping, by the UE, performing the random access procedure on the first BWP, and performing the random access procedure on the second BWP.

Random access method and apparatus

Embodiments of this application provide a beam configuration method and apparatus. The method includes: determining base station scheduling information rank, and determining a direction of arrival power spectrum; determining peak information based on the direction of arrival power spectrum; determining, based on the peak information, beam directions that meet power conditions; selecting, based on a preset filtering condition, at least one candidate beam direction from the beam directions that meet the power conditions; and determining a target beam direction of a terminal from the at least one candidate beam direction based on the base station scheduling information rank, a value of a channel quality indicator CQI, and the direction of arrival power spectrum. In this way, a beam configuration manner on a terminal side is more flexible, to meet requirements of different service scenarios, and improve stability of a communications link.

Beam configuration method and apparatus

A translation model training method for a computer device includes obtaining a training sample set, the training sample set including a plurality of training samples. Each training sample is a training sample pair having a training input sample in a first language and a training output sample in a second language. The method also includes determining a disturbance sample set corresponding to each training sample in the training sample set, the disturbance sample set comprising at least one disturbance sample, and a semantic similarity between the disturbance sample and the corresponding training sample being greater than a first preset value; and training an initial translation model by using the plurality of training samples and the disturbance sample set corresponding to each training sample to obtain a target translation model, such that the training output sample remains same for the disturbance sample and the corresponding training sample.

Translation model training method, sentence translation method, device, and storage medium

Disclosed is an impact-resistant and aging-resistant reflective plastic applied to an automobile, which is prepared by copolymerizing of polyether polyol, 4,4,4-triphenylmethane triisocyanate and a benzimidazole derivative modified ZnS-mesoporous silica composite, the benzimidazole derivative modified ZnS-mesoporous silica composite is a ZnS-mesoporous silica composite modified by an anti-aging agent and a silane coupling agent KH-560, the ZnS-mesoporous silica is obtained by mixing an amphoteric surfactant, zinc chloride, and &#x3b3;-aminopropyltrimethoxysilane uniformly, then reacting in ethyl orthosilicate, and calcining with hydrogen sulfide. In the present disclosure, by dispersing zinc sulfide in the pores of mesoporous carbon dioxide, chemical grafting 2-(2-hydroxy-5-aminophenyl)benzimidazole as a bridge onto isocyanate, and then polymerizing into polyurethane, which can improve the reflective performance and aging resistance of polyurethane, and improve the abrasion resistance and impact resistance.

Impact-resistant and aging-resistant reflective plastic applied to automobiles

A beam selection method includes determining, by a terminal, a first angular power spectrum of a low frequency channel transmitted between the terminal and an access network device, determining, by the terminal, a first high frequency beam scanning range based on the first angular power spectrum, and scanning, by the terminal, the first high frequency beam scanning range for a high frequency beam of the access network device.

Beam selection method and apparatus

A random access method includes determining a physical channel resource and a preamble sequence, where the physical channel resource or the preamble sequence is associated with a first beam determined by the terminal device and used by a network device for performing downlink transmission, sending the preamble sequence on the physical channel resource, receiving a random access response message and a paging message, where the random access response message includes a preamble identifier, the paging message includes first identification information of one or more to-be-paged terminal devices, and the paging message is sent on the first beam, sending a second identifier of the terminal device when the preamble identifier corresponds to the preamble sequence, and a first identifier of the terminal device matches the first identification information, receiving first indication information, and determining, based on the first indication information to complete a random access procedure.

Random access method, terminal device, and network device

A cell search method and a communications apparatus are provided. The method includes the following: before searching for a cell, a terminal device may determine, with reference to historical access information of another terminal device, whether there is a cell at the location, and adjust, based on a determining result, a value of a search parameter used to search for a cell. For example, if there is a cell, the terminal device may adjust the search parameter to accelerate a search. If there is no cell, the terminal device may perform no search, or adjust the search parameter to reduce an unnecessary cell search performed by the terminal device, so that power consumption of the terminal device can be reduced.

Cell search method and communications apparatus

Growing community of inventors

Shenzhen, China

Yong Cheng