O. Lahav

An artificial neural network approach to the classification of galaxy spectra

We present a method for the automated classification of galaxies with low signal-to-noise (S/N) ratio spectra typical of redshift surveys. We develop spectral simulations based on the parameters for the 2-degree-field Galaxy Redshift Survey, and with these simulations we investigate the technique of principal component analysis when applied specifically to spectra of low S/N ratio. We relate the objective principal components to features in the spectra and use a small number of components to successfully reconstruct the underlying signal from the low-quality spectra. Using the principal components as input, we train an artificial neural network to classify the noisy simulated spectra into morphological classes, revealing the success of the classification against the observed b(J) magnitude of the source, which we compare with alternative methods of classification. We find that more than 90 per cent of our sample of normal galaxies are correctly classified into one of the five broad morphological classes for simulations at b(J) = 19.7. By dividing the data into separate sets, we show that a classification on to the Hubble sequence is relevant only for normal galaxies, and that spectra with unusual features should be incorporated into a classification scheme based predominantly on their spectral signatures. We discuss how an artificial neural network can be used to distinguish normal and unusual galaxy spectra, and also discuss the possible application of these results to spectra from galaxy redshift surveys.

Combining cosmological data sets: hyperparameters and Bayesian evidence

A method is presented for performing joint analyses of cosmological data sets, in which the weight assigned to each data set is determined directly by its own statistical properties. The weights are considered in a Bayesian context as a set of hyperparameters, which are then marginalized over in order to recover the posterior distribution as a function only of the cosmological parameters of interest. In the case of a Gaussian likelihood function, this marginalization may be performed analytically. Calculation of the Bayesian evidence for the data, with and without the introduction of hyperparameters, enables a direct determination of whether the data warrant the introduction of weights into the analysis; this generalizes the standard likelihood ratio approach to model comparison. The method is illustrated by application to the classic toy problem of fitting a straight line to a set of data. A cosmological illustration of the technique is also presented, in which the latest measurements of the cosmic microwave background power spectrum are used to infer constraints on cosmological parameters.

The 2dF Galaxy Redshift Survey : higher-order galaxy correlation functions

We measure moments of the galaxy count probability distribution function in the Two-degree Field Galaxy Redshift Survey (2dFGRS). The survey is divided into volume-limited subsamples in order to examine the dependence of the higher-order clustering on galaxy luminosity. We demonstrate the hierarchical scaling of the averaged p-point galaxy correlation functions, ξ p , up to p = 6. The hierarchical amplitudes, Sp = ξ p /ξ p-1 2, are approximately independent of the cell radius used to smooth the galaxy distribution on small to medium scales. On larger scales we find that the higher-order moments can be strongly affected by the presence of rare, massive superstructures in the galaxy distribution. The skewness S 3 has a weak dependence on luminosity, approximated by a linear dependence on log luminosity. We discuss the implications of our results for simple models of linear and non-linear bias that relate the galaxy distribution to the underlying mass.

Automated morphological classification of APM galaxies by supervised artificial neural networks

We train Artificial Neural Networks to classify galaxies based solely on the morphology of the galaxy images as they appear on blue survey plates. The images are reduced and morphological features such as bulge size and the number of arms are extracted, all in a fully automated manner. The galaxy sample was first classified by 6 independent experts. We use several definitions for the mean type of each galaxy, based on those classifications. We then train and test the network on these features. We find that the rms error of the network classifications, as compared with the mean types of the expert classifications, is 1.8 Revised Hubble Types. This is comparable to the overall rms dispersion between the experts. This result is robust and almost completely independent of the network architecture used.

Variance and Skewness in the FIRST Survey

We investigate the large-scale clustering of radio sources in the FIRST 1.4-GHz survey by analysing the distribution function (counts in cells). We select a reliable sample from the the FIRST catalogue, paying particular attention to the problem of how to define single radio sources from the multiple components listed. We also consider the incompleteness of the catalogue. We estimate the angular two-point correlation function

The 2dF Galaxy Redshift Survey: Voids and hierarchical scaling models

w(\theta)

Galaxies, Human Eyes, and Artificial Neural Networks

, the variance

A Comparative study of morphological classifications of APM galaxies

\Psi_2

Neural computation as a tool for galaxy classification : methods and examples

, and skewness

The 2dF Galaxy Redshift Survey: hierarchical galaxy clustering

\Psi_3