Spectral classification using convolutional neural networks
This work addresses the need for accurate classification methods in astrophysics, but it is incremental as it applies an existing deep learning method to a new domain-specific dataset.
The authors tackled the problem of autonomous spectral classification in astrophysics by training a convolutional neural network to classify objects (quasar, star, or galaxy) from one-dimensional spectra, achieving a success rate of nearly 95% on a dataset of over 60,000 spectra.
There is a great need for accurate and autonomous spectral classification methods in astrophysics. This thesis is about training a convolutional neural network (ConvNet) to recognize an object class (quasar, star or galaxy) from one-dimension spectra only. Author developed several scripts and C programs for datasets preparation, preprocessing and postprocessing of the data. EBLearn library (developed by Pierre Sermanet and Yann LeCun) was used to create ConvNets. Application on dataset of more than 60000 spectra yielded success rate of nearly 95%. This thesis conclusively proved great potential of convolutional neural networks and deep learning methods in astrophysics.