IDLat: An Importance-Driven Latent Generation Method for Scientific Data

IDLat: An Importance-Driven Latent Generation Method for Scientific Data
复制标题

IDLat:一种重要驱动的科学数据潜在生成方法

DOI:
10.1109/tvcg.2022.3209419
复制
发表时间:
2023
影响因子:
5.2
通讯作者:
Shen, Han-Wei
Shen, Han-Wei
中科院分区:
计算机科学1区
文献类型:
--
作者:
Shen, Jingyi;Li, Haoyu;Xu, Jiayi;Biswas, Ayan;Shen, Han-Wei

文献摘要

参考文献

相似文献

基于深度学习的潜在表示已广泛用于许多科学可视化应用,如等面相似性分析、体绘制、流场合成和数据约简等。然而,现有的潜在表示大多以无监督的方式从原始数据中生成,这使得很难结合领域兴趣来控制潜在表示的大小和重构数据的质量。在本文中,我们提出了一种新的重要性驱动的潜在表示,以促进领域兴趣引导的科学数据可视化和分析。我们利用空间重要性图来表示各种科学兴趣,并将其作为特征转换网络的输入来指导潜在生成。我们通过与自编码器一起训练的无损熵编码算法进一步减小了潜在大小,提高了存储和记忆效率。我们用来自多个科学可视化应用的数据定性和定量地评估了我们的方法生成的潜在表示的有效性和效率。
Deep learning based latent representations have been widely used for numerous scientific visualization applications such as isosurface similarity analysis, volume rendering, flow field synthesis, and data reduction, just to name a few. However, existing latent representations are mostly generated from raw data in an unsupervised manner, which makes it difficult to incorporate domain interest to control the size of the latent representations and the quality of the reconstructed data. In this paper, we present a novel importance-driven latent representation to facilitate domain-interest-guided scientific data visualization and analysis. We utilize spatial importance maps to represent various scientific interests and take them as the input to a feature transformation network to guide latent generation. We further reduced the latent size by a lossless entropy encoding algorithm trained together with the autoencoder, improving the storage and memory efficiency. We qualitatively and quantitatively evaluate the effectiveness and efficiency of latent representations generated by our method with data from multiple scientific visualization applications.
DOI: 10.1109/cluster48925.2021.00034
发表时间: 2021-05
期刊: 2021 IEEE International Conference on Cluster Computing (CLUSTER)
影响因子: --
作者:
Jinyang Liu;S. Di;Kai Zhao;Sian Jin;Dingwen Tao;Xin Liang;Zizhong Chen;F. Cappello
通讯作者: Jinyang Liu;S. Di;Kai Zhao;Sian Jin;Dingwen Tao;Xin Liang;Zizhong Chen;F. Cappello
为时变多元数据选择代表性时间步长的深度学习方法
DOI: 10.1109/visual.2019.8933759
发表时间: 2019
期刊: Proceedings of IEEE VIS Conference (Short Papers
影响因子: --
作者:
Porter, William P.;Xing, Yunhao;von Ohlen, Blaise R.;Han, Jun;Wang, Chaoli
通讯作者: Wang, Chaoli
DOI: 10.1186/s40668-016-0017-2
发表时间: 2016
期刊: Computational Astrophysics and Cosmology
影响因子: --
作者:
B. Friesen;A. Almgren;Z. Lukic;G. Weber;D. Morozov;V. Beckner;M. Day
通讯作者: M. Day