Methods, apparatuses, and systems for collaboratively updating model by multiple parties for implementing privacy protection

A system for collaboratively updating a model by multiple parties for implementing privacy protection, comprising a server and n participants;

• wherein
• the server is configured to deliver an aggregation result of a t-th round of common samples to each participant i, wherein the aggregation result is obtained by the server by aggregating n first prediction results that are output by the n participants based on respective local models for the t-th round of common samples;
• each participant i is configured to perform first update on a local ith model thereof according to the t-th round of common samples and the aggregation result;
• each participant i is further configured to:
• perform second update on the ith model obtained after the first update based on a first private sample fixed in a local sample set and a sample label thereof;
• each participant i is further configured to:
• input a (t+1)th round of common samples that are used for a next round of iteration into the ith model obtained after the second update, and send an output second prediction result to the server;
• the server is configured to aggregate n second prediction results sent by the n participants for a next round of iteration;
• and
• each participant i is further configured to:
• after the multiple rounds of iterations, use the ith model thereof obtained after the second update as a model that is updated in collaboration with another participant.

A non-transitory computer readable storage medium having stored therein instructions that, when executed by a processor of a computing device, cause the processor to:

• deliver, by the server, an aggregation result of a t-th round of common samples to each participant i, wherein the aggregation result is obtained by the server by aggregating n first prediction results that are output by the n participants based on respective local models for the t-th round of common samples;
• perform, by each participant i, first update on a local ith model thereof according to the t-th round of common samples and the aggregation result;
• perform, by each participant i, second update on the ith model obtained after the first update based on a first private sample fixed in a local sample set and a sample label thereof;
• input, by each participant i, a (t+1)th round of common samples that are used for a next round of iteration into the ith model obtained after the second update, and send an output second prediction result to the server;
• aggregate, by the server, n second prediction results sent by the n participants for a next round of iteration;
• and
• after the multiple rounds of iterations, use, by each participant i, the ith model thereof obtained after the second update as a model that is updated in collaboration with another participant.