skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: DETECTING ACTIVE GALACTIC NUCLEI USING MULTI-FILTER IMAGING DATA. II. INCORPORATING ARTIFICIAL NEURAL NETWORKS

Abstract

This is the second paper of the series Detecting Active Galactic Nuclei Using Multi-filter Imaging Data. In this paper we review shapelets, an image manipulation algorithm, which we employ to adjust the point-spread function (PSF) of galaxy images. This technique is used to ensure the image in each filter has the same and sharpest PSF, which is the preferred condition for detecting AGNs using multi-filter imaging data as we demonstrated in Paper I of this series. We apply shapelets on Canada-France-Hawaii Telescope Legacy Survey Wide Survey ugriz images. Photometric parameters such as effective radii, integrated fluxes within certain radii, and color gradients are measured on the shapelets-reconstructed images. These parameters are used by artificial neural networks (ANNs) which yield: photometric redshift with an rms of 0.026 and a regression R-value of 0.92; galaxy morphological types with an uncertainty less than 2 T types for z ≤ 0.1; and identification of galaxies as AGNs with 70% confidence, star-forming/starburst (SF/SB) galaxies with 90% confidence, and passive galaxies with 70% confidence for z ≤ 0.1. The incorporation of ANNs provides a more reliable technique for identifying AGN or SF/SB candidates, which could be very useful for large-scale multi-filter optical surveys that also includemore » a modest set of spectroscopic data sufficient to train neural networks.« less

Authors:
;  [1]
  1. Physics and Astronomy Department, York University, Toronto, ON M3J 1P3 (Canada)
Publication Date:
OSTI Identifier:
22273350
Resource Type:
Journal Article
Resource Relation:
Journal Name: Astronomical Journal (New York, N.Y. Online); Journal Volume: 146; Journal Issue: 4; Other Information: Country of input: International Atomic Energy Agency (IAEA)
Country of Publication:
United States
Language:
English
Subject:
79 ASTROPHYSICS, COSMOLOGY AND ASTRONOMY; ALGORITHMS; ASTRONOMY; ASTROPHYSICS; COLOR; DATA ANALYSIS; FILTERS; GALAXIES; GALAXY NUCLEI; IMAGE PROCESSING; IMAGES; NEURAL NETWORKS; PHOTOMETRY; RED SHIFT; STARS; TELESCOPES

Citation Formats

Dong, X. Y., and De Robertis, M. M., E-mail: xydong@yorku.ca. DETECTING ACTIVE GALACTIC NUCLEI USING MULTI-FILTER IMAGING DATA. II. INCORPORATING ARTIFICIAL NEURAL NETWORKS. United States: N. p., 2013. Web. doi:10.1088/0004-6256/146/4/87.
Dong, X. Y., & De Robertis, M. M., E-mail: xydong@yorku.ca. DETECTING ACTIVE GALACTIC NUCLEI USING MULTI-FILTER IMAGING DATA. II. INCORPORATING ARTIFICIAL NEURAL NETWORKS. United States. doi:10.1088/0004-6256/146/4/87.
Dong, X. Y., and De Robertis, M. M., E-mail: xydong@yorku.ca. 2013. "DETECTING ACTIVE GALACTIC NUCLEI USING MULTI-FILTER IMAGING DATA. II. INCORPORATING ARTIFICIAL NEURAL NETWORKS". United States. doi:10.1088/0004-6256/146/4/87.
@article{osti_22273350,
title = {DETECTING ACTIVE GALACTIC NUCLEI USING MULTI-FILTER IMAGING DATA. II. INCORPORATING ARTIFICIAL NEURAL NETWORKS},
author = {Dong, X. Y. and De Robertis, M. M., E-mail: xydong@yorku.ca},
abstractNote = {This is the second paper of the series Detecting Active Galactic Nuclei Using Multi-filter Imaging Data. In this paper we review shapelets, an image manipulation algorithm, which we employ to adjust the point-spread function (PSF) of galaxy images. This technique is used to ensure the image in each filter has the same and sharpest PSF, which is the preferred condition for detecting AGNs using multi-filter imaging data as we demonstrated in Paper I of this series. We apply shapelets on Canada-France-Hawaii Telescope Legacy Survey Wide Survey ugriz images. Photometric parameters such as effective radii, integrated fluxes within certain radii, and color gradients are measured on the shapelets-reconstructed images. These parameters are used by artificial neural networks (ANNs) which yield: photometric redshift with an rms of 0.026 and a regression R-value of 0.92; galaxy morphological types with an uncertainty less than 2 T types for z ≤ 0.1; and identification of galaxies as AGNs with 70% confidence, star-forming/starburst (SF/SB) galaxies with 90% confidence, and passive galaxies with 70% confidence for z ≤ 0.1. The incorporation of ANNs provides a more reliable technique for identifying AGN or SF/SB candidates, which could be very useful for large-scale multi-filter optical surveys that also include a modest set of spectroscopic data sufficient to train neural networks.},
doi = {10.1088/0004-6256/146/4/87},
journal = {Astronomical Journal (New York, N.Y. Online)},
number = 4,
volume = 146,
place = {United States},
year = 2013,
month =
}
  • An artificial neural network (ANN)-based diagnostic adviser capable of identifying the operating status of a nuclear power plant is described. A dynamic node architecture scheme is used to optimize the architectures of the two backpropagation ANNs that embody the advisor. The first or root network is used to determine whether or not the plant is in a normal operating condition. If the plant is not in a normal condition, the second or classifier network is used to recognize the particular off-normal condition or transient taking place. These networks are developed using simulated plant behavior during both normal and abnormal conditions.more » The adviser is effective at diagnosing 27 distinct transients based on 43 scenarios simulated at various severities that contain up to 3% noise.« less
  • A study of the application of artificial neural networks (ANNs) to active structural control is presented. A simple effective strategy for the on-line control of single-degree-of-freedom (SDF) structures is proposed. The strategy is to apply the control force at every time step to destroy the buildup of the system response, and the control force needed for the next time step is fully determined from the information available at the current time ; therefore the time delay associated with the control algorithm is eliminated. The control algorithm can be implemented for either a closed or open-closed loop controller. The controller usesmore » a trained ANN to determine the control force such that the velocity induced at the preceding time step is canceled. A feedforward neural network with an adaptive backpropagation training method is used in this study. In the backpropagation training, the learning rate is determined by ensuring the decrease of the error function of the input-output training patterns at each training cycle. Numerical examples of SDF systems under earthquake excitations are given to illustrate the effectiveness of the proposed control strategy. The uncertainties in the time history of the excitation and in the modeling of the system, including the magnitudes of the excitations, the natural frequency and nonlinearity of the systems are examined. Significant reduction of the response is observed. Also, is shown that the proposed control strategy has the ability to reduce the peak that occurs during the first few cycles of the time history, an ability that linear control laws lack.« less
  • A new analysis technique, called multi-level interval estimation method, is developed for locating damage in structures. In this method, the artificial neural networks (ANN) analysis method is combined with the statistics theory to estimate the range of damage location. The ANN is multilayer perceptron trained by back-propagation. Natural frequencies and modal shape at a few selected points are used as input to identify the location and severity of damage. Considering the large-scale structures which have lots of elements, multi-level interval estimation method is developed to reduce the estimation range of damage location step-by-step. Every step, estimation range of damage locationmore » is obtained from the output of ANN by using the method of interval estimation. The next ANN training cases are selected from the estimation range after linear transform, and the output of new ANN estimation range of damage location will gained a reduced estimation range. Two numerical example analyses on 10-bar truss and 100-bar truss are presented to demonstrate the effectiveness of the proposed method.« less
  • We used photometry from the Kepler satellite to characterize the variability of four radio-loud active galactic nuclei (AGNs) on timescales from years to minutes. The Kepler satellite produced nearly continuous high precision data sets which provided better temporal coverage than possible with ground based observations. We have now accumulated 11 quarters of data, eight of which were reported in our previous paper. In addition to constructing power spectral densities (PSDs) and characterizing the variability of the last three quarters, we have linked together the individual quarters using a multiplicative scaling process, providing data sets spanning ∼2.8 yr with >98% coveragemore » at a 30 minute sampling rate. We compute PSDs on these connected data sets that yield power law slopes at low frequencies in the approximate range of –1.5 to –2.0, with white noise seen at higher frequencies. These PSDs are similar to those of both the individual quarters and to those of ground-based optical observations of other AGNs. We also have explored a PSD binning method intended to reduce a bias toward shallow slope fits by evenly distributing the points within the PSDs. This tends to steepen the computed PSD slopes, especially when the low frequencies are relatively poorly fit. We detected flares lasting several days in which the brightness increased by ∼15%-20% in one object, as well a smaller flare in another. Two AGNs showed only small, ∼1%-2%, fluctuations in brightness.« less
  • Neural network modeling is a powerful nonlinear regression analysis method that is extremely useful in identifying behavioral trends. This methodology was applied to the problem of predicting Ferrite Number in arc welds as a function of composition. This paper describes the details of the development of the neural network model, named FNN-1999, including the identification of the optimum network architecture and network parameters. The model was trained on the same data as the WRC-1992 constitution diagram and covers a range of Ferrite Numbers from 0 to 117, with a corresponding wide range in composition. Results of the model are presentedmore » in Part 2. It is shown that the accuracy of the FNN-1999 model in predicting Ferrite Number is superior to the accuracy of other models that are currently available, including the WRC-1992 diagram.« less