arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 1804.00816astro-ph.CO
收藏

用深度神经网络分类宇宙的大尺度结构

Classifying the Large Scale Structure of the Universe with Deep Neural Networks

Miguel A. Aragon-Calvo

摘要:

我们首次将深度神经网络应用于宇宙大尺度结构中宇宙细丝和墙壁的语义分割。我们的研究结果基于深度卷积神经网络(CNN),其U-Net架构使用现有的最先进的手动引导分割方法进行训练。我们使用Voronoi模型和N体仿真成功地训练并测试了一个U-Net。从Voronoi模型预测的分割掩模具有分别用于细丝和掩模的0.95和0.97的Dice系数。来自N体模拟的预测分割掩模对于壁和细丝分别具有0.78和0.72的Dice系数。细丝掩膜中相对较低的Dice系数是由U-Net模型预测但不存在于原始分割掩膜中的细丝的结果。我们的研究结果表明,对于一个定义良好的数据集,如Voronoi模型的U-网具有良好的性能。在N体数据集的情况下,U-Net产生的细丝掩模的质量高于从最先进的方法获得的分割掩模。U-Net的性能比用于训练它的方法更好,甚至能够找到手动引导分割无法识别的纤细细丝。这里介绍的U-Net可以在几分钟内处理一个$512^3$卷,而不需要复杂的预处理。深度CNN作为下一代大容量计算机N体模拟和星系巡天的有效和准确的分析工具具有巨大的潜力。

英文摘要:

We present the first application of deep neural networks to the semantic segmentation of cosmological filaments and walls in the Large Scale Structure of the Universe. Our results are based on a deep Convolutional Neural Network (CNN) with a U-Net architecture trained using an existing state-of-the-art manually-guided segmentation method. We successfully trained an tested an U-Net with a Voronoi model and an N-body simulation. The predicted segmentation masks from the Voronoi model have a Dice coefficient of 0.95 and 0.97 for filaments and mask respectively. The predicted segmentation masks from the N-body simulation have a Dice coefficient of 0.78 and 0.72 for walls and filaments respectively. The relatively lower Dice coefficient in the filament mask is the result of filaments that were predicted by the U-Net model but were not present in the original segmentation mask. Our results show that for a well-defined dataset such as the Voronoi model the U-Net has excellent performance. In the case of the N-body dataset the U-Net produced a filament mask of higher quality than the segmentation mask obtained from a state-of-the art method. The U-Net performs better than the method used to train it, being able to find even the tenuous filaments that the manually-guided segmentation failed to identify. The U-Net presented here can process a $512^3$ volume in a few minutes and without the need of complex pre-processing. Deep CNN have great potential as an efficient and accurate analysis tool for the next generation large-volume computer N-body simulations and galaxy surveys.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑