Gamma-ray bursts (GRB) do not easily subdivide into classes; there is tremendous overlap in their behaviors. However, it is difficult to imagine how one mechanism produces the large range of observed GRB temporal and spectral characteristics. There is strong evidence for two GRB classes and weaker (but statistically meaningful) evidence for the existence of a third class using BATSE data. We are applying pattern recognition algorithms from the artificial intelligence (AI) branch of computer science to GRB classification. In addition to science, our
eventual goal is to produce an online AI program library and a detailed GRB database. This approach has already allowed us to discover that the properties of the third class do not require the existence of a new source population but  can be manufactured from one of the other classes via a combination of instrumental biases and data correlations.
 

Return to the Physics and Astronomy Seminar Home Page