Laboratory for Cosmological Data Miningbusiness
License: University of Illinois/NCSA Open Source License
Mailing List:The Laboratory for Cosmological Data Mining was founded in 2002 at the University of Illinois by Professor Robert J. Brunner to develop and apply computational technologies to extract cosmological information from the large astrophysical data sets being generated within our community.
- Bayesian source detection and characterization Astronomical surveys like SDSS and The Dark Energy survey have provided enough image information for astronomers to perform source detection of celestial bodies by applying various data mining techniques. In this project we aim to develop an Image based Bayesian model fitting technique to detect and characterize astronomical sources in an image. Using Bayesian Inference, We are going to find best possible functional representations for the sources within single band images in the sky.
- Creating updated, scientifically-calibrated mosaics for the RC3 Catalogue Using the latest SDSS DR10 data, we provide color composite images and scientifically-calibrated FITS mosaics in all SDSS imaging bands, for all the RC3 galaxies that lie within the survey’s footprint. Due to the positional inaccuracy inherent in the RC3 catalog, the mosaicking program uses a recursive algorithm for positional update first, then conduct the mosaicking procedure.The program is generalized into a pipeline,which can be easily extended to future survey data or other source catalogs.
- Crowdsourcing Web Application Traditional Crowdsourcing application does not have functionalities to tag multiple objects and generally focus only on the identification/classification of single object by the viewer. This project aims to develop a user interface which can enable a human user to easily view, identify, and tag multiple features within an image. The user will be able to select pre-assigned labels, as well as enter free-form text to describe unusual or interesting classes.
- Image Pixel Based Photometric Redshift Estimation Current techniques for photometric redshift estimation rely upon reduced integrated information from images. The information that is wasted on a pixel level can be made use of in order to get a better estimate. The performance of this technique can then be checked on different hardware like GPUs and against the machine learning algorithm used.
- Strong Gravitational Lens Time Delays and Detecting Strong lenses in Images To apply new strong gravitational lensing time delay measurements which can enable constraints on dark energy. Quasar variability can be used to measure the time delay between two or more quasar images in a strongly lensed system. To achieve precision cosmological constraints, the error on this measurement needs to be reduced. The proposed approach will reduce this error for finding time delays. This project will also perform model based identification of strong gravitational lenses in Images.