| 黄东瑞*,毛克彪**,郭中华*,徐乐园*,胡泽民*,赵瑞*.几种神经网络经典模型综述[J].高技术通讯(中文),2023,33(8):860~871 |
| 几种神经网络经典模型综述 |
| A review of classical models of neural networks |
| |
| DOI：10. 3772/ j. issn. 1002-0470. 2023. 08. 008 |
| 中文关键词: 深度学习； 卷积神经网络(CNN)； 循环神经网络(RNN)； 生成对抗网络(GAN)； 图神经网络(GNN) |
| 英文关键词: deep learning, convolutional neural network (CNN), recurrent neural network (RNN), generative adversarial network (GAN), graph neural network (GNN) |
| 基金项目: |
| 作者 | 单位 | | 黄东瑞* | (*宁夏大学物理与电子电气工程学院银川 750021)
(**中国农业科学院农业资源与农业区划研究所北京 100081) | | 毛克彪** | | | 郭中华* | | | 徐乐园* | | | 胡泽民* | | | 赵瑞* | |
|
| 摘要点击次数: 933 |
| 全文下载次数: 1356 |
| 中文摘要: |
| 近年来深度学习在众多领域都有突出表现并展现出巨大应用潜力。神经网络模型是深度学习的重要载体，因此有必要对其进行深入分析。然而神经网络模型发展至今，呈现出种类多样化、应用专有化等特点，例如有用于目标检测的YOLO系列模型、机器翻译的Transformer系列模型等。本文试图通过对几种主要神经网络经典模型的剖析，找到一条了解深度学习的高效路径。本文首先对深度学习的发展进行概述；然后分别对卷积神经网络(CNN)、循环神经网络(RNN)、生成对抗网络(GAN)、图神经网络(GNN)从模型介绍、原理分析、网络训练、模型改进方面进行详细阐述；最后对上述神经网络模型进行总结并对深度学习未来发展进行展望。 |
| 英文摘要: |
| In recent years, deep learning has shown outstanding performance and great application potential in many fields. Neural network model is an important carrier of deep learning, so it is necessary to analyze it deeply. However, so far, the development of neural network models has shown the characteristics of diversification and application specialization, such as YOLO series models for object detection and Transformer series models for machine translation. This paper attempts to find an efficient way to understand deep learning by analyzing several classical neural network models. This review first summarizes the development of deep learning. Then the convolutional neural network (CNN), recurrent neural network (RNN), generative adversarial network (GAN) and graph neural network (GNN) are described in detail from the aspects of model introduction, principle analysis, network training and model improvement. Finally, the above neural network models are summarized and the future development of deep learning is prospected. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
