Information prediction method, model training method and server

Abstract

The invention discloses an information prediction method. The method comprises the steps of obtaining a to-be-predicted image; extracting a to-be-predicted feature set in the to-be-predicted image, wherein the to-be-predicted feature set comprises a first to-be-predicted feature, a second to-be-predicted feature and a third to-be-predicted feature, the first to-be-predicted feature represents theimage feature of the first area, the second to-be-predicted feature represents the image feature of the second area, the third to-be-predicted feature represents the attribute feature related to the interactive operation, and the range of the first area is smaller than that of the second area; and obtaining a first label and / or a second label corresponding to the to-be-predicted feature set through the target joint model, the first label representing a label related to the operation content, and the second label representing a label related to the operation intention. The invention further discloses a model training method and a server. According to the method, only one joint model is used for predicting microoperation and the large situation, the problem of hard switching in a hierarchical model is effectively solved, and the prediction convenience is improved.

Description

Technical field [0001] The present invention relates to the field of artificial intelligence technology, in particular to a method of information prediction, a method of model training, and a server. Background technique [0002] Artificial intelligence (AI) programs have defeated top professional players in chess games with clear rules. In contrast, the operation of multiplayer online battle arena (MOBA) games is more complicated and closer to real-world scenarios. Overcoming AI problems in MOBA games will help us explore and solve complex problems in the real world. [0003] Based on the complexity of the MOBA game operation, the operation of the entire MOBA game can usually be divided into two types: overall view and micro-operation to reduce the complexity of the entire MOBA game. See figure 1 , figure 1 A schematic diagram of the hierarchical model for the existing scheme. As shown in the figure, it is divided into the overall decision-making process of "jungle", "clearing s...

AI Technical Summary

Problems solved by technology

[0004] However, hierarchical modeling requires separate design and training of the macro model and the micro model. That is to say, the two models are independent of each other. In practical applications, it is necessary to decide which model to use for prediction. Therefore, , there is a problem of hard switching between the two models, which is not conducive to the convenience of prediction

Images