The ability to discover new transients via image differencing without direct human intervention is an important task in observational astronomy. For these kind of image classification problems, machine Learning techniques such as Convolutional Neural Networks (CNNs) have shown remarkable success. In this work, we present the results of an automated transient identification on images with CNNs for an extant dataset from the Dark Energy Survey Supernova program (DES-SN), whose main focus was on using Type Ia supernovae for cosmology. By performing an architecture search of CNNs, we identify networks that efficiently select non-artifacts (e.g. supernovae, variable stars, AGN, etc.) from artifacts (image defects, mis-subtractions, etc.), achieving the efficiency of previous work performed with random Forests, without the need to expend any effort in feature identification. The CNNs also help us identify a subset of mislabeled images. Performing a relabeling of the images in this subset, the resulting classification with CNNs is significantly better than previous results.
$\texttt{BTSbot}$: A Multi-input Convolutional Neural Network to Automate and Expedite Bright Transient Identification for the Zwicky Transient Facility
Deep-HiTS: Rotation Invariant Convolutional Neural Network for Transient Detection
Millennium Institute of Astrophysics(千年天体物理研究所);Center for Mathematical Modeling, Universidad de Chile(智利大学数学建模中心);AURA Observatory in Chile(智利AURA天文台);Universidad de Chile(智利大学)
Department of Computer Science and Engineering, Indian Institute of Technology, Gandhinagar, India(印度理工学院计算机科学与工程系);Division of Physics, Mathematics, and Astronomy, California Institute of Technology, Pasadena, CA 91125, USA(加州理工学院物理、数学与天文学系);Center for Data Driven Discovery, California Institute of Technology, Pasadena, CA 91125, USA(数据驱动发现中心);IPAC, California Institute of Technology, 1200 E. California Blvd, Pasadena, CA 91125, USA(IPAC, 加州理工学院);Caltech Optical Observatories, California Institute of Technology, Pasadena, CA 91125, USA(加州理工学院光学观测站)
MaxiMask and MaxiTrack: two new tools for identifying contaminants in astronomical images using convolutional neural networks
在这项工作中,我们提出了两个卷积神经网络分类器检测天文图像中的污染物。经过训练后,我们的分类器能够识别各种污染物,如宇宙射线、热像素和坏像素、持久性效应、卫星或飞机轨迹、残留条纹图案、模糊特征、饱和像素、衍射尖峰和图像中的跟踪错误。它们涵盖了广泛的环境条件,如视宁度、图像采样、探测器类型、光学和恒星密度。第一个分类器MaxiM ask基于每个像素属于给定污染物类的概率,执行语义分割并为每个污染物生成坏像素图。第二个分类器MaxiTrack通过计算焦平面受跟踪误差影响的概率,对整个图像和马赛克进行分类。我们从各种现代电荷耦合器件和近红外相机的真实数据中收集了训练和测试数据,这些数据通过图像模拟进行了增强。我们量化了这两个分类器的性能,并表明M axi M ask在识别宇宙射线命中方面达到了最先进的性能。由于内置的贝叶斯更新机制,这两个分类器都可以进行调整,以满足各种观测环境中的特定科学目标。
New Metrics for Identifying Variables and Transients in Large Astronomical Surveys
International Centre for Radio Astronomy Research - Curtin University(国际射电天文研究中心 - 科廷大学);ARC Centre for Transforming Maintenance through Data Science - Curtin University(ARC 数据科学变革维护中心 - 科廷大学);School of Electrical Engineering, Computing and Mathematical Sciences, Curtin University(电气、计算和数学科学学院,科廷大学);Curtin Biometry and Agricultural Data Analytics, Centre for Crop and Disease Management, Curtin University(科廷生物计量与农业数据分析,作物与疾病管理中心,科廷大学);Astrophysics, Department of Physics, University of Oxford(天体物理学,物理系,牛津大学);Department of Astronomy, University of Cape Town(天文学系,开普敦大学)
Identifying Planetary Transit Candidates in TESS Full-Frame Image Light Curves via Convolutional Neural Networks
NASA Goddard Space Flight Center(美国国家航空航天局戈达德太空飞行中心);Universities Space Research Association(大学空间研究协会);The Catholic University of America(美国天主教大学);Universita di Napoli Federico II(那不勒斯费德里科二世大学);Science Systems and Applications, Inc.(科学系统与应用公司);University of Maryland, Baltimore County(马里兰大学巴尔的摩县分校)
Classifying Image Sequences of Astronomical Transients with Deep Neural Networks
Universidad de los Andes(洛斯安第斯大学);Center for Research and Formation in Artificial Intelligence(人工智能研究与培训中心);Systems and Computing Engineering Department(系统与计算工程系);Departamento de Física(物理系)
Detecting optical transients using artificial neural networks and reference images from different surveys
The University of Texas Rio Grande Valley(德克萨斯大学里奥格兰德河谷分校);University of Warsaw(华沙大学);National Centre for Nuclear Research(国家核研究中心);Texas Tech University(德克萨斯理工大学)
Real-Time Detection of Anomalies in Large-Scale Transient Surveys
University of Cambridge(剑桥大学);Massachusetts Institute of Technology(麻省理工学院);The Alan Turing Institute(艾伦·图灵研究所);University of the Western Cape(西开普大学);South African Radio Astronomy Observatory(南非射电天文台);African Institute for Mathematical Sciences(非洲数学科学研究所);Swinburne University of Technology(斯威本科技大学);University of Illinois at Urbana-Champaign(伊利诺伊大学厄巴纳-香槟分校)
新的时域测量,如Vera C。Rubin Observatory Legacy Survey of Space and Time(LSST)将在每晚观测数百万个瞬变警报,这使得视觉识别新的和有趣的瞬变的标准方法变得不可行。提出了两种实时自动检测异常瞬态光变曲线的新方法。这两种方法都基于一个简单的想法,即如果来自已知瞬态群体的光变曲线可以准确地建模,则任何与模型预测的偏差都可能是异常的。第一种建模方法是使用时间卷积网络(TCN)构建的概率神经网络,第二种是瞬态的可解释贝叶斯参数模型。我们证明了我们的方法的能力,提供异常分数的光变曲线从Zwicky瞬态设施的时间的函数。我们表明,神经网络的灵活性,属性,使他们这样一个强大的工具,许多回归任务,是什么使他们不太适合异常检测相比,我们的参数模型。参数模型能够以高精度和召回分数识别常见超新星类别的异常,对于大多数罕见的类别,如千新星,潮汐破坏事件,中间亮度瞬变和对不稳定超新星,精度-召回曲线(AUCPR)下的面积超过0.79。我们识别异常的能力随着光变曲线的寿命而提高。我们的框架,与瞬态分类器结合使用,将使新的大规模调查的不寻常的瞬变快速和优先的后续行动。