Lightweight Actor-Critic generative adversarial network-based medical question and answer generation system

Abstract

The invention discloses a lightweight Actor-Critic generative adversarial network-based medical question and answer generation system, which comprises a generator and a discriminator of a lightweight Actor-Critic structure, and is characterized in that after the system inputs a medical question proposed by a user, the generator generates a medical diagnosis scheme in a coding-decoding mode; a known medical question and answer text serves as a data set and is input into a generator, a maximum likelihood estimation method is adopted for pre-training, data distribution generated by the pre-trained generator serves as a false sample, known data serves as a true sample, and the data distribution and the known data are input into a discriminator network for pre-training. The invention also includes pre-training a generator and a discriminator, multiplexing the generator as an Actor network and constructing a Critic network of which the structure is a long-short-term memory network, updating the network weight parameter of the generator by adopting an Actor-Critic algorithm, performing adversarial training with the discriminator, and performing lightweight processing on the network by adopting a multi-path multi-layer Actor and Critic network lightweight method based on a group MCP regular term.

Description

technical field [0001] The invention relates to the fields of reinforcement learning and natural language processing, in particular to a medical question-and-answer generation system based on a lightweight Actor-Critic generative confrontation network. Background technique [0002] At present, due to the shortage and uneven distribution of medical resources, the hospital is under severe operating pressure and the relationship between doctors and patients is tense. With the development of mobile Internet technology, the informatization of the medical industry has been valued by more and more enterprises and countries. The medical question answering system is widely used in the medical industry. It integrates medical resources in different regions through the network to obtain high-quality medical service efficiency and relieve doctors' work pressure. Question answer generation is an implementation of question answering system. It is a promising research direction in the fiel...

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is to provide a medical question-and-answer generation system based on a lightweight Actor-Critic generative confrontation network, so as to alleviate the severe operating pressure and tense relationship between doctors and patients caused by the shortage and uneven distribution of medical resources.

Images