人工智能培训

搜索

深度学习论文:系外行星大气探测的贝叶斯深度学习(Bayesian Deep Learning for Exoplanet Atmospheric Retriev

[复制链接]
jamiezhao 发表于 5 天前 | 显示全部楼层 |阅读模式
jamiezhao 5 天前 219 0 显示全部楼层
深度学习论文:系外行星大气探测的贝叶斯深度学习(Bayesian Deep Learning for Exoplanet Atmospheric Retrieval)在过去十年中,对系外行星的研究已经从它们的检测转移到了它们的大气特征。大气检索,用于从观测到的光谱确定大气温度和成分的逆向建模技术既耗时又计算密集,需要复杂的算法,将数千万个大气模型与观测数据进行比较,以找出最可能的值和相关的不确定性对于每个模型参数。对于陆上行星来说,回收的大气成分可以清楚地表明保持大气稳定性所必需的气态物质的表面通量,这反过来可以提供对行星上活跃的地质和/或生物过程的洞察。这些大气包含许多分子,其中一些是生物印记,或指示生物活性的分子。传统检索模型的运行时间随着模型参数的数量而变化,因此考虑到更多的分子种类,运行时间会变得过长。机器学习(ML)和计算机视觉方面的最新进展提供了新的方法,在给定足够的数据集训练的情况下,将执行检索的时间减少了几个数量级。在这里,我们提出了一个基于ML的检索框架,称为智能exoplaNetAtmospheric RetrievAl(INARA)它包括贝叶斯深度学习模型以及使用NASA行星频谱发生器(PSG)生成的合成岩石外行星的3,000,000个光谱的数据集。我们的工作代表了岩石,陆地系外行星的第一个ML模型以及在此尺度下产生的光谱的第一个合成数据集。
Over the past decade, the study of exoplanets has shifted from theirdetection to the characterization of their atmospheres.Atmospheric retrieval,the inverse modeling technique used to determine an atmosphere's temperatureand composition from an observed spectrum, is both time-consuming andcompute-intensive, requiring complex algorithms that compare thousands tomillions of atmospheric models to the observational data to find the mostprobable values and associated uncertaintiesfor each model parameter.Forrocky, terrestrial planets, the retrieved atmospheric composition can giveinsight into the surface fluxes of gaseous species necessary to maintain thestability of that atmosphere, which may in turn provide insight into thegeological and/or biological processes active on the planet.These atmospherescontain many molecules, some of which are biosignatures, or moleculesindicative of biological activity.Runtimes of traditional retrieval modelsscale with the number of model parameters, so as more molecular species areconsidered, runtimes can become prohibitively long.Recent advances in machinelearning (ML) and computer vision offer new ways to reduce the time to performa retrieval by orders of magnitude, given a sufficient data set to train with.Here we present an ML-based retrieval framework called Intelligent exoplaNetAtmospheric RetrievAl (INARA)that consists of a Bayesian deep learning modelfor retrieval and a data set of 3,000,000 spectra of synthetic rocky exoplanetsgenerated using the NASA Planetary Spectrum Generator (PSG).Our workrepresents the first ML model for rocky, terrestrial exoplanets and the firstsynthetic data set of spectra generated at this scale.深度学习论文:系外行星大气探测的贝叶斯深度学习(Bayesian Deep Learning for Exoplanet Atmospheric Retrieval) xtYyyPHPytxPXnLN.jpg
URL地址:https://arxiv.org/abs/1811.03390     ----pdf下载地址:https://arxiv.org/pdf/1811.03390    ----深度学习论文:系外行星大气探测的贝叶斯深度学习(Bayesian Deep Learning for Exoplanet Atmospheric Retrieval)
回复

使用道具 举报

您需要登录后才可以回帖 登录 | 立即注册

本版积分规则 返回列表 发新帖

jamiezhao当前离线
新手上路

查看:219 | 回复:0

快速回复 返回顶部 返回列表