Many astronomers working in the field of AstroInformatics write code as part of their work. Although the programming language of choice is Python, a small number (8%) use R. R has its specific strengths in the domain of statistics, and is often viewed as limited in the size of data it can handle. However, Microsoft R Server is a product that removes these limitations by being able to process much larger amounts of data. I present some highlights of R Server, by illustrating how to fit a convolutional neural network using R. The specific task is to classify galaxies, using only images extracted from the Sloan Digital Skyserver.

This work emphasizes that heterogeneity, diversity, discontinuity, and discreteness in data is to be exploited in classification and regression problems. A global a priori model may not be desirable. For data analytics in cosmology, this is motivated by the variety of cosmological objects such as elliptical, spiral, active, and merging galaxies at a wide range of redshifts. Our aim is matching and similarity-based analytics that takes account of discrete relationships in the data. The information structure of the data is represented by a hierarchy or tree where the branch structure, rather than just the proximity, is important. The representation is related to p-adic number theory. The clustering or binning of the data values, related to the precision of the measurements, has a central role in this methodology. If used for regression, our approach is a method of cluster-wise regression, generalizing nearest neighbour regression. Both to exemplify this analytics approach, and to demonstrate computational benefits, we address the well-known photometric redshift or ‘photo-z’ problem, seeking to match Sloan Digital Sky Survey (SDSS) spectroscopic and photometric redshifts.

Due to limitations in available instrumentation and observation time, a spectroscopic determination of distance is not feasible for all objects in the sky. Therefore statistical methods that estimate redshifts, based on photometric measurement are of tremendous importance to many astrophysical questions. Determining cosmological parameters and understanding evolutionary processes in the universe are just two examples. When perform astrophysical analyses, it is necessary to treat the uncertainties of the estimates correctly. Over-simplification of results and the usage of wrong tools to evaluate the performance of probabilistic redshift estimates were commonly found in the literature. We present proper tools for evaluating uncertain redshift estimates and discuss the necessity of multimodal redshift distributions.

In the modern galaxy surveys photometric redshifts play a central role in a broad range of studies, from gravitational lensing and dark matter distribution to galaxy evolution. Using a dataset of ~ 25,000 galaxies from the second data release of the Kilo Degree Survey (KiDS) we obtain photometric redshifts with five different methods: (i) Random forest, (ii) Multi Layer Perceptron with Quasi Newton Algorithm, (iii) Multi Layer Perceptron with an optimization network based on the Levenberg-Marquardt learning rule, (iv) the Bayesian Photometric Redshift model (or BPZ) and (v) a classical SED template fitting procedure (Le Phare). We show how SED fitting techniques could provide useful information on the galaxy spectral type which can be used to improve the capability of machine learning methods constraining systematic errors and reduce the occurrence of catastrophic outliers. We use such classification to train specialized regression estimators, by demonstrating that such hybrid approach, involving SED fitting and machine learning in a single collaborative framework, is capable to improve the overall prediction accuracy of photometric redshifts.

Imaging atmospheric Cherenkov telescopes (IACTs) are sensitive to rare gamma-ray photons, buried in the background of charged cosmic-ray (CR) particles, the flux of which is several orders of magnitude greater. The ability to separate gamma rays from CR particles is important, as it is directly related to the sensitivity of the instrument. This gamma-ray/CR-particle classification problem in IACT data analysis can be treated with the rapidly-advancing machine learning algorithms, which have the potential to outperform the traditional box-cut methods on image parameters. We present preliminary results of a precise classification of a small set of muon events using a convolutional neural networks model with the raw images as input features. We also show the possibility of using the convolutional neural networks model for regression problems, such as the radius and brightness measurement of muon events, which can be used to calibrate the throughput efficiency of IACTs.

The current archives of LAMOST multi-object spectrograph contain millions of fully reduced spectra, from which the automatic pipelines have produced catalogues of many parameters of individual objects, including their approximate spectral classification. This is, however, mostly based on the global shape of the whole spectrum and on integral properties of spectra in given bandpasses, namely presence and equivalent width of prominent spectral lines, while for identification of some interesting object types (e.g. Be stars or quasars) the detailed shape of only a few lines is crucial. Here the machine learning is bringing a new methodology capable of improving the reliability of classification of such objects even in boundary cases.

We present results of Spark-based semi-supervised machine learning of LAMOST spectra attempting to automatically identify the single and double-peak emission of Hα line typical for Be and B[e] stars. The labelled sample was obtained from archive of 2m Perek telescope at Ondřejov observatory. A simple physical model of spectrograph resolution was used in domain adaptation to LAMOST training domain. The resulting list of candidates contains dozens of Be stars (some are likely yet unknown), but also a bunch of interesting objects resembling spectra of quasars and even blazars, as well as many instrumental artefacts. The verification of a nature of interesting candidates benefited considerably from cross-matching and visualisation in the Virtual Observatory environment.

The Digitised First Byurakan Survey (DFBS) provides low dispersion optical spectra for about 24 million sources. A two-step machine learning algorithm based on similarities to predefined templates is applied to select different classes of rare objects in the dataset automatically, for example late type stars, quasars and white dwarves. Identifying outliers from the groups of common astrophysical objects may lead to discovery of rare objects, such as gamma-ray burst afterglows.

Establishing accurate morphological measurements of galaxies in a reasonable amount of time for future big-data surveys such as EUCLID, the Large Synoptic Survey Telescope or the Wide Field Infrared Survey Telescope is a challenge. Because of its high level of abstraction with little human intervention, deep learning appears to be a promising approach. Deep learning is a rapidly growing discipline that models high-level patterns in data as complex multilayered networks. In this work we test the ability of deep convolutional networks to provide parametric properties of Hubble Space Telescope like galaxies (half-light radii, Sérsic indices, total flux etc..). We simulate a set of galaxies including point spread function and realistic noise from the CANDELS survey and try to recover the main galaxy parameters using deep-learning. We compare the results with the ones obtained with the commonly used profile fitting based software GALFIT. This way showing that with our method we obtain results at least equally good as the ones obtained with GALFIT but, once trained, with a factor 5 hundred time faster.

We present METAPHOR (Machine-learning Estimation Tool for Accurate PHOtometric Redshifts), a method able to provide a reliable PDF for photometric galaxy redshifts estimated through empirical techniques. METAPHOR is a modular workflow, mainly based on the MLPQNA neural network as internal engine to derive photometric galaxy redshifts, but giving the possibility to easily replace MLPQNA with any other method to predict photo-z’s and their PDF. We present here the results about a validation test of the workflow on the galaxies from SDSS-DR9, showing also the universality of the method by replacing MLPQNA with KNN and Random Forest models. The validation test include also a comparison with the PDF’s derived from a traditional SED template fitting method (Le Phare).

We compare the results of using a Random Forest Classifier with the results of using Nonparametric Discriminant Analysis to classify whether a filament channel (in the case of a filament eruption) or an active region (in the case of a flare) is about to produce an event. A large number of descriptors are considered in each case, but it is found that only a small number are needed in order to get most of the improvement in performance over always predicting the majority class. There is little difference in performance between the two classifiers, and neither results in substantial improvements over simply predicting the majority class.

In order to understand how galaxies form and evolve, the measurement of the parameters related to their morphologies and also to the way they interact is one of the most relevant requirements. Due to the huge amount of data that is generated by surveys, the morphological and interaction analysis of galaxies can no longer rely on visual inspection. For dealing with such issue, new approaches based on machine learning techniques have been proposed in the last years with the aim of automating the classification process. We tested Deep Learning using images of galaxies obtained from CANDELS to study the accuracy achieved by this tool considering two different frameworks. In the first, galaxies were classified in terms of their shapes considering five morphological categories, while in the second, the way in which galaxies interact was employed for defining other five categories. The results achieved in both cases are compared and discussed.

The need for accurate photometric redshifts estimation is a topic that has fundamental importance in Astronomy, due to the necessity of efficiently obtaining redshift information without the need of spectroscopic analysis. We propose a method for determining accurate multi-modal photo-z probability density functions (PDFs) using Mixture Density Networks (MDN) and Deep Convolutional Networks (DCN). A comparison with a Random Forest (RF) is performed.

Strong gravitational microlensing (GM) events provide us a possibility to determine both the parameters of microlensed source and microlens. GM can be an important clue to understand the nature of dark matter on comparably small spatial and mass scales (i.e. substructure), especially when speaking about the combination of astrometrical and photometrical data about high amplification microlensing events (HAME). In the same time, fitting of HAME lightcurves of microlensed sources is quite time-consuming process. That is why we test here the possibility to apply the statistical machine learning techniques to determine the source and microlens parameters for the set of HAME lightcurves, using the simulated set of amplification curves of sources microlensed by point masses and clumps of DM with various density profiles.

Machine learning techniques have proven to be increasingly useful in astronomical applications over the last few years, for example in object classification, estimating redshifts and data mining. One example of object classification is classifying galaxy morphology. This is a tedious task to do manually, especially as the datasets become larger with surveys that have a broader and deeper search-space. The Kaggle Galaxy Zoo competition presented the challenge of writing an algorithm to find the probability that a galaxy belongs in a particular class, based on SDSS optical spectroscopy data. The use of convolutional neural networks (convnets), proved to be a popular solution to the problem, as they have also produced unprecedented classification accuracies in other image databases such as the database of handwritten digits (MNIST †) and large database of images (CIFAR ‡). We experiment with the convnets that comprised the winning solution, but using broad classifications. The effect of changing the number of layers is explored, as well as using a different activation function, to help in developing an intuition of how the networks function and to see how they can be applied to radio galaxy images.

For the vast amounts of spectra produced by LAMOST, the pipeline basing on PCAZ method is limited by the bad flux calibration and low S/N data. This work focuses on the study of the efficient recognition methods of galaxy spectra of LAMOST basing on spectral lines information. The new method searches spectral lines and extracts the information of spectral lines (position, height, and width et al.) automatically. Using the spectral lines information which are less influenced by the quality of flux calibration and the S/N ratio, galaxy spectra are recognized with the redshift measured through spectral lines matching method. The experiment verified it is feasible for the LAMOST galaxy spectra: the correct recognition rate > 80% for the data with SNR_g > 5, and > 90% for the data with SNR_r > 5. Compared with the redshift of SDSS, the systematic error of our method is 0, and the standard deviation of the error is 0.0002.

We explore photometric redshift estimation of quasars with the SDSS DR12 quasar sample. Firstly the quasar sample is separated into three parts according to different redshift ranges. Then three classifiers based on Extreme Learning Machine (ELM) are created in the three redshift ranges. Finally k-Nearest Neighbor (kNN) approach is applied on the three samples to predict photometric redshifts of quasars with multiwavelength photometric data. We compare the performance with different input patterns by ELM-KNN with that only by kNN. The experimental results show that ELM-KNN is feasible and superior to kNN (e.g. rms is 0.0751 vs. 0.2626 for SDSS sample), in other words, the ensemble method has the potential to increase regressor performance beyond the level reached by an individual regressor alone and will be a good choice when facing much more complex data.

The time domain is the emerging forefront of astronomical research with new facilities and instruments providing unprecedented amounts of data on the temporal behavior of astrophysical populations. Dealing with the size and complexity of this requires new techniques and methodologies. Quasars are an ideal work set for developing and applying these: they vary in a detectable but not easily quantifiable manner whose physical origins are poorly understood. In this paper, we will review how quasars are identified by their variability and how these techniques can be improved, what physical insights into their variability can be gained from studying extreme examples of variability, and what approaches can be taken to increase the number of quasars known. These will demonstrate how astroinformatics is essential to discovering and understanding this important population.

Large Synoptic Survey Telescope will make great contributions to many scientific fields. One of the modules will be time-domain astronomy and detection of transient events. In this paper, some considerations about transient events and alerts are presented.

The existence of multiple subclasses of type Ia supernovae (SNeIa) has been the subject of great debate in the last decade. In this work, we show how machine learning tools facilitate identification of subtypes of SNe Ia. Using Deep Learning for dimensionality reduction, we were capable of performing such identification in a parameter space of significantly lower dimension than its principal component analysis counterpart. This is evidence that the progenitor system and the explosion mechanism can be described with a small number of initial physical parameters. All tools used here are publicly available in the Python package DRACULA (Dimensionality Reduction And Clustering for Unsupervised Learning in Astronomy) and can be found within COINtoolbox (https://github.com/COINtoolbox/DRACULA).

Despite a flood of discoveries over the last ~ 20 years, our knowledge of the exoplanet population is incomplete owing to a gap between the sensitivities of different detection techniques. However, a census of exoplanets at all separations from their host stars is essential to fully understand planet formation mechanisms. Microlensing offers an effective way to bridge the gap around 1–10 AU and is therefore one of the major science goals of the Wide Field Infrared Survey Telescope (WFIRST) mission. WFIRST’s survey of the Galactic Bulge is expected to discover ~ 20,000 microlensing events, including ~ 3000 planets, which represents a substantial data analysis challenge with the modeling software currently available. This paper highlights areas where further work is needed. The community is encouraged to join new software development efforts aimed at making the modeling of microlensing events both more accessible and rigorous.