Training method and device of federated model based on neural network, and computer device

Abstract

The invention discloses a training method and device of a federated model based on a neural network, and a computer device. The method comprises the following steps: receiving a training request for carrying out iterative training on a preset federated model to obtain historical gradient information sent by each participant of the federated model to the federated model; screening the historical gradient information according to a preset LSTM model to obtain a plurality of participants which need to iteratively update the federal model at the current moment; and receiving gradient information generated after the local data of each participant in the plurality of participants train the corresponding local model so as to update the federal model. According to the invention, based on the neural network technology, the participants needing to participate in iteration training in each iteration process are obtained from all participants of the federated model by using the long-short-term memory artificial neural network, so that the calculation cost and the network communication transmission cost of the participants are reduced, the training speed of the federated model is accelerated, and meanwhile, the training precision of the federal model is improved.

Description

technical field [0001] The invention relates to the technical field of federated learning, in particular to a neural network-based federated model training method, device and computer equipment. Background technique [0002] In the field of federated learning technology, each participant participating in the federation jointly builds a federated model, and the federated model is jointly participated and contributed by all federated participants, which can effectively help multiple institutions meet the requirements of user privacy protection, data security, and government regulations Next, data usage and machine learning modeling. In the process of joint training of the federated model, the data itself will not leave each federated user, but the participants and the central server are generally not in the same network environment, and the operating status is also different. In addition, the data transmitted during the joint training process Model information is encrypted by...

AI Technical Summary

Problems solved by technology

However, the above method still has a large defect, which will inevitably cause some loss of precision in the process of quantization or compression; and random selection of participants for update introduces some randomness, and some participants may be less Participating in federated training may result in the model not necessarily being able to achieve the ideal convergence effect in ideal iteration rounds; overlapping computing and communication can only reduce the required time, but cannot reduce communication overhead

Images