基于深度强化学习的计算机多媒体图像分类技术研究  

Research on Computer Multimedia Image Classification Technology Based on Deep Reinforcement Learning

作　　者：陈伟[1] CHEN Wei(Hebei Vocational and Technical College of Building Materials,Qinhuangdao,Hebei 066000,China)

机构地区：[1]河北建材职业技术学院,河北秦皇岛066000

出　　处：《移动信息》2025年第4期314-316,共3页Mobile Information

摘　　要：文中深入探索了基于深度强化学习的计算机多媒体图像分类技术。首先,阐述了深度强化学习原理,包括智能体、环境、状态、动作和奖励机制,及深度神经网络在逼近价值函数或策略函数的应用。其次,介绍了多媒体图像颜色、纹理、形状特征提取方法和卷积神经网络在其中的作用。同时,详细描述了模型整体架构,含卷积神经网络感知模块和包括策略网络、价值网络的决策模块,以及奖励机制设计。最后,从训练数据集准备、训练参数设置和利用强化学习策略优化训练3个方面阐述了模型训练与优化过程,为多媒体图像分类提供了全面的技术方案。This paper delves into the computer multimedia image classification technology based on deep reinforcement learning.Firstly,the principles of deep reinforcement learning were elucidated,including agents,environment,state,action,and reward mechanisms,as well as the application of deep neural networks in approximating value or strategy functions.Secondly,the methods for extracting color,texture,and shape features from multimedia images and the role of convolutional neural networks were introduced.At the same time,the overall architecture of the model was described in detail,including the convolutional neural network perception module and the decision module including the policy network and value network,as well as the design of the reward mechanism.Finally,the model training and optimization process was elaborated from three aspects:preparation of training dataset,setting of training parameters,and optimization of training using reinforcement learning strategies,providing a comprehensive technical solution for multimedia image classification.

关 键 词：深度强化学习 多媒体图像 图像分类 模型构建 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]