Beijing Bytedance Network Technology Co., Ltd.

Lemon Inc.

Bytedance Inc.

Douyin Vision Co., Ltd.

Junyuan Xie

Di Wu

Chenliaohui Fang

Peng Zhao

Xiaobing Liu

Longyijia Li

Liangchao Wu

Cheng Chen

Chong Wang

Jiankai Sun

Yixiang Chen

Long Chang

Lizhe Zhang

Weihao Gao

Xin Yang

Long Huang

Jun Zhang

Lei Shi

Kan Liu

Yuanshun Yao

Hongyi Zhang

Ruoxing Huang

Shihao Shen

Lingyuan Zhang

Deliang Fan

Lele Yu

Aonan Zhang

Xiang Wu

A model construction method and an apparatus, and a medium and an electronic device are disclosed. The method is applied to a first participant platform, and includes: associating first configuration information pre-created by a first participant with second configuration information pre-created by a second participant; verifying the first configuration information; sending, to a second participant platform corresponding to the second participant, a second creation request for requesting the creation of the federated learning model, to cause the second participant platform to verify the second configuration information creating a first model task on the basis of a first parameter corresponding to the first configuration information; and performing co-training on the basis of the first model task and a second model task, to obtain the federated learning model.

Model construction method and apparatus, and medium and electronic device

The present disclosure relates to a data protection method, apparatus, medium and electronic device. The method comprises: obtaining a specified batch of reference samples of an active participant of a joint training model; determining generation gradient information of the first reference sample; determining target gradient information sent to the passive participant according to the generation gradient information, and sending the target gradient information to the passive participant, to update, by the passive participant, parameters of the joint training model according to the target gradient information. Through the above solution, the influence of the generated data on the training process and model performance of the joint training model is avoided as much as possible, and the privacy and security of data are improved.

Data protection method, apparatus, medium and electronic device

The present disclosure relates to a data processing method, non-transitory medium and electronic device. The method includes: acquiring first user data and second user data, and initializing a first time window corresponding to the first user data and first time information corresponding to the first time window, as well as a second time window corresponding to the second user data and second time information corresponding to the second time window; determining first data and the first time information corresponding to the first time window based on the first user data; determining second data and the second time information corresponding to the second time window based on the second user data; based on the first data and the second data, determining alignment data corresponding to a same user from the first user data and the second user data.

Data processing method, and non-transitory medium and electronic device

Provided are a method and an apparatus for generating prediction information, an electronic device, and a computer readable medium. The method includes: generating, based on first user characteristic information of a target user, anonymous user information of the target user (); sending the anonymous user information to a second processing end to enable the second processing end to generate prediction information based on the anonymous user information and second user characteristic information () of the target user. Data interaction and sharing are realized while ensuring data privacy, thereby improving accuracy of the prediction information.

Method and apparatus for generating prediction information, electronic device, and computer readable medium

There are proposed a method, device, apparatus, and medium for protecting sensitive data. In a method, to-be-processed data is received from a server device. A processing result of a user for the to-be-processed data is received, the processing result comprising sensitive data of the user for the processing of the to-be-processed data. A gradient for training a server model at the server device is determined based on a comparison between the processing result and a prediction result for the to-be-processed data. The gradient is updated in a change direction associated with the gradient so as to generate an updated gradient to be sent to the server device. Noise is added only in the change direction associated with the gradient. The corresponding overhead of processing noise in a plurality of directions can be reduced, and no excessive noise data interfering with training will be introduced to the updated gradient.

Method, apparatus, device and medium for protecting sensitive data

Provided are a method for processing model parameters, and an apparatus. The method comprises: a model parameter set to be sharded is obtained, wherein the model parameter set comprises a multi-dimensional array corresponding to a feature embedding; attribute information for a storage system used for storing the model parameter set to be sharded is obtained, wherein the storage system used for storing the model parameter set to be sharded differs from a system on which a model corresponding to the model parameter set to be sharded is located when operating; the model parameter set to be sharded is stored in the storage system according to the attribute information.

Method for processing model parameters, and apparatus

A network connection method and device for a training participant of a joint training model are provided, the training participant operates in a master-worker mode. The method includes: acquiring communication state information of a worker, the communication state information indicating a communication connection phase that the worker is in; acquiring communication state information of a target worker as target communication state information, where the target worker includes a peer node corresponding to the worker, and the peer node belongs to a different training participant of the joint training model; and resetting, in response to determining that the target communication state information does not match the communication state information of the worker, a communication connection phase that the worker is in.

Network connection method and device for training participant end of common training model

Disclosed are a data protection method and apparatus, and a server and a medium. A particular embodiment of the method comprises: acquiring gradient associated information, which respectively corresponds to a target sample that belongs to a binary classification sample set with unbalanced distribution and a reference sample that belongs to the same batch as the target sample; generating information of data noise to be added; according to the information of said data noise, correcting an initial gradient transfer value corresponding to the target sample, such that corrected gradient transfer information corresponding to samples in the sample set that belong to different types is consistent; and sending the gradient transfer information to a passive party of a joint training model. By means of the embodiment, there is no significant difference between corrected gradient transfer information corresponding to positive and negative samples, thereby effectively protecting the security of data.

Data protection method and apparatus, and server and medium

Growing community of inventors

Beijing, China

Junyuan Xie