深度学习驱动的可视化

Visualization Driven by Deep Learning

近年来,深度学习技术在充足的计算能力之下得到飞速的发展.在可视化与可视分析流程中,部分需要人为参与的环节和问题,已然能够借用数据驱动的方式来替代和解决.该综述基于经典的可视化与可视分析流程,即可视化的压缩、构建、交互、评估等各个方面,审视了其中能够借助深度学习技术的部分,并对相关研究进行了详细而系统的论述,探讨深度学习技术与可视化结合带来的技术红利,深刻分析深度学习驱动的可视化未来发展的前景.

In recent years,deep learning technology developed rapidly with sufficient computing ability.In the visualization and visual analysis process,some steps that require human participation can be replaced and solved by data-driven methods.This article is based on the division of classic visualization and visual analysis process,including construction,interaction,evaluation,and other aspects of visualization.This article reviewed the parts that can use deep learning technology,carried out a detailed and systematic discussion of related research,discussed the technical benefit brought by the combination of deep learning technology and visualization,and deeply analyzed the prospects of the development of visualization driven by deep learning.