Autonomous evolution intelligent dialogue method, system and device based on physical environment game

Abstract

The invention belongs to the field of artificial intelligence and visual dialogues, particularly relates to an autonomous evolution intelligent dialogue method, system and device based on a physical environment game, and aims to solve the problems of high calculation consumption, low convergence speed and low information processing accuracy of an intelligent system. The method comprises the stepsof obtaining a to-be-processed image and a corresponding problem text; Encoding the picture into a picture vector by adopting an optimized dialogue model, and encoding the question text into a question vector; Generating a state vector based on the picture vector and the question vector; And decoding the state vector to obtain a response text and outputting the response text. Wherein a discriminator needs to be introduced in the optimization process of the dialogue model, the dialogue model and the discriminator are alternately optimized until the mixed loss function of the dialogue model andthe loss function value of the discriminator are not reduced or lower than a preset value, and model optimization is completed. Compared with a traditional method, the method is low in calculation consumption and high in convergence speed, and the information processing accuracy is further improved.

Description

technical field [0001] The invention belongs to the field of artificial intelligence and visual dialogue, and in particular relates to a method, system and device for self-evolving intelligent dialogue based on physical environment games. Background technique [0002] In the field of artificial intelligence, the most common method of training a model that can understand data is supervised training. Supervised training maximizes the probability of sample data and corresponding labels, designs the objective function from the perspective of statistical distribution, and updates the model parameters. Supervised training requires a large amount of data, and the only goal is to "explain data from a statistical point of view", which is not the same as the actual learning process of humans. Therefore, there is an obvious defect in supervised learning: when some important parts of the target task lack When using labels and references, the effect of supervised learning will be signif...

AI Technical Summary

Problems solved by technology

[0006] In order to solve the above-mentioned problems in the prior art, that is, in order to solve the problems of large calculation consumption, slow convergence speed and low accuracy of processing information of the intelligent system, the present invention provides a self-evolving intelligent dialogue method based on physical environment game, including:

Images