Search

In astronomical applications of machine learning, the distribution of objects used for building a model is often different from the distribution of the objects the model is later applied to. This is known as sample selection bias, which is a major challenge for statistical inference as one can no longer assume that the labeled training data are representative. To address this issue, one can re-weight the labeled training patterns to match the distribution of unlabeled data that are available already in the training phase. There are many examples in practice where this strategy yielded good results, but estimating the weights reliably from a finite sample is challenging. We consider an efficient nearest neighbor density ratio estimator that can exploit large samples to increase the accuracy of the weight estimates. To solve the problem of choosing the right neighborhood size, we propose to use cross-validation on a model selection criterion that is unbiased under covariate shift. The resulting algorithm is our method of choice for density ratio estimation when the feature space dimensionality is small and sample sizes are large. The approach is simple and, because of the model selection, robust. We empirically find that it is on a par with established kernel-based methods on relatively small regression benchmark datasets. However, when applied to large-scale photometric redshift estimation, our approach outperforms the state-of-the-art. [ABSTRACT FROM AUTHOR]

AstroStat is an easy-to-use tool for performing statistical analysis on data. It has been designed to be compatible with Virtual Observatory (VO) standards thus enabling it to become an integral part of the currently available collection of VO tools. A user can load data in a variety of formats into AstroStat and perform various statistical tests using a menu driven interface. Behind the scenes, all analyses are done using the public domain statistical software—R and the output returned is presented in a neatly formatted form to the user. The analyses performable include exploratory tests, visualizations, distribution fitting, correlation & causation, hypothesis testing, multivariate analysis and clustering. The tool is available in two versions with identical interface and features—as a web service that can be run using any standard browser and as an offline application. AstroStat will provide an easy-to-use interface which can allow for both fetching data and performing power statistical analysis on them. [ABSTRACT FROM AUTHOR]

Machine learning techniques offer a precious tool box for use within astronomy to solve problems involving so-called big data . They provide a means to make accurate predictions about a particular system without prior knowledge of the underlying physical processes of the data. In this article, and the companion papers of this series, we present the set of Generalized Linear Models (GLMs) as a fast alternative method for tackling general astronomical problems, including the ones related to the machine learning paradigm. To demonstrate the applicability of GLMs to inherently positive and continuous physical observables, we explore their use in estimating the photometric redshifts of galaxies from their multi-wavelength photometry. Using the gamma family with a log link function we predict redshifts from the PHoto-z Accuracy Testing simulated catalogue and a subset of the Sloan Digital Sky Survey from Data Release 10. We obtain fits that result in catastrophic outlier rates as low as ∼1% for simulated and ∼2% for real data. Moreover, we can easily obtain such levels of precision within a matter of seconds on a normal desktop computer and with training sets that contain merely thousands of galaxies. Our software is made publicly available as a user-friendly package developed in Python , R and via an interactive web application. This software allows users to apply a set of GLMs to their own photometric catalogues and generates publication quality plots with minimum effort. By facilitating their ease of use to the astronomical community, this paper series aims to make GLMs widely known and to encourage their implementation in future large-scale projects, such as the Large Synoptic Survey Telescope. [ABSTRACT FROM AUTHOR]

Abstract: In order to study the statistical characters and physical significance of the color indexes and orbital parameters of Centaurs, we have analyzed the color and orbit parameter correlations by using statistical methods according to the known color indexes of 39 Centaurs and their respective orbital parameters. We find that all the correlations of B − V vs. V − R, V − R vs. R − I, B − V vs. R − I and B − R vs. R − I are strong. The Centaurs exhibit obvious redblue double-color property, and the point of demarcation is close to B − R =1.4. The distributions of the various color indexes of Centaurs exhibit the normal distribution. For the correlations between colors and orbital parameters, except for the weak correlations between B − V and the orbital inclination i as well as between V − R and the orbital semi-major axis a, no evident correlation has been found between color indexes and orbital parameters. [Copyright &y& Elsevier]

Abstract: The observed distribution of young open clusters is far from uniform. Statistics shows that, when age, spatial distribution and kinematics are considered simultaneously, they tend to appear in clumps. These young cluster groups or families constitute unambiguously coeval, genetically related complexes associated to the underlying spiral structure. In this paper, we derive detailed physical properties for one of them: the Cassiopeia–Perseus family. With a diameter of about 600pc, it is located 2kpc from the Sun, embedded in the Perseus arm, and probably includes 10–20 members. It began to form 20–40 Myr ago although we find distinctive evidence for at least three generations of star formation organized in two distinct fronts, with the oldest clusters located at lower Galactic longitude than the youngest. The plane roughly defined by the structure is inclined ∼30° to the Galactic disk with most candidate members located below the disk and moving away from it. Our results for this cluster of clusters suggest that, within a coherent cloud complex, the first generation of star formation is triggered by the shock wave induced by a spiral arm. The second and subsequent generations are sustained by ionization fronts and supernova shocks created by the evolution of the first generation of massive stars. In this particular case, the front moves with average velocity of about 70km/s in the direction of increasing Galactic longitude. The Cassiopeia–Perseus family and related objects appear to be a close relative of the cluster complexes found in the spiral galaxy M51 or perhaps a younger analog of the Gould Belt. [Copyright &y& Elsevier]

Abstract: With the availability of multi-wavelength, multi-scale and multi-epoch astronomical catalogues, the number of features to describe astronomical objects has increases. The better features we select to classify objects, the higher the classification accuracy is. In this paper, we have used data sets of stars and quasars from near-infrared band and radio band. Then best-first search method was applied to select features. For the data with selected features, the algorithm of decision table was implemented. The classification accuracy is more than 95.9%. As a result, the feature selection method improves the effectiveness and efficiency of the classification method. Moreover the result shows that decision table is robust and effective for discrimination of celestial objects and used for preselecting quasar candidates for large survey projects. [Copyright &y& Elsevier]

Abstract: The automated classification of objects from large catalogs or survey projects is an important task in many astronomical surveys. Faced with various classification algorithms, astronomers should select the method according to their requirements. Here we describe several kinds of decision trees for finding active objects by multi-wavelength data, such as REPTree, Random Tree, Decision Stump, Random Forest, J48, NBTree, AdTree. All decision tree approaches investigated are in the WEKA package. The classification performance of the methods is presented. In the process of classification by decision tree methods, the classification rules are easily obtained, moreover these rules are clear and easy to understand for astronomers. As a result, astronomers are inclined to prefer and apply them, thus know which attributes are important to discriminate celestial objects. The experimental results show that when various decision trees are applied in discriminating active objects (quasars, BL Lac objects and active galaxies) from non-active objects (stars and galaxies), ADTree is the best only in terms of accuracy, Decision Stump is the best only considering speed, J48 is the optimal choice considering both accuracy and speed. [Copyright &y& Elsevier]

Transient and variable phenomena in astrophysical sources are of particular importance to understand the underlying gamma-ray emission processes. In the very-high energy gamma-ray domain, transient and variable sources are related to charged particle acceleration processes that could for instance help understanding the origin of cosmic-rays. The imaging atmospheric Cherenkov technique used for gamma-ray astronomy above ~ 100 GeV is well suited for detecting such events. However, the standard analysis methods are not optimal for such a goal and more sensitive methods are specifically developed in this publication. The sensitivity improvement could therefore be helpful to detect brief and faint transient sources such as Gamma-Ray Bursts. [ABSTRACT FROM AUTHOR]