Papers by Author | in ( Plenary | Parallel | Poster | Summary ) Sessions

Hastac, an Algorithm for Developing a Tree of Cuts, and its Relation to Neural Networks 

         D. Bowser-Chao, J. Hughes, J. Linnemann 

	          Michigan State University
              Department of Physics and Astronomy
  • Paper (Postscript)
  • Paper (PDF) 
                    
                         Abstract
    We describe Hastac, an engine for finding criteria for separating signal and background samples by means of a tree of cuts on linear combinations of variables. The method is quite fast, and not only allows near-optimal rejection for given efficiency, but easy identification of the most significant variables from a large set candidate variables. A mapping to feed-forward neural nets is possible, offering an excellent starting point for further refinement. The method has been applied to the problem of top quark identification. Results will also be presented with test distributions where the answers are known. 
    Submitter's Name: James T. Linnemann
Submitter's Institution: Michigan State University
Department of Physics and Astronomy
East Lansing, MI 48824  USA

Submitter's EMAIL address: linnemann@msupa.pa.msu.edu
Submitter's phone number: 517-355-3328
Intended Speaker's Name: James T. Linnemann