Applying Neural Networks to Resonance Search in High Energy Physics
It is the aim of many modern experiments in high energy physics to discover and investigate the properties of rare particles. However, often so few of these particles are produced in an experiment that it is very difficult to discover their decay signature against the background. There are various techniques to suppress the background and thus to enhance the relative strength of the signal, but none of the techniques currently used is optimal. In view of the future operation of even larger accelerators, such as the Superconducting Supercollider, where a signal may occur once in a billion events, and millions of events will have to be analyzed each second, it is desirable to optimize the techniques for resonance search. Recently it has been suggested that neural networks may be used for this task. It is still unclear, however, whether neural networks can perform an unbiased data analysis or whether they may produce artificial signals due to information acquired during the training process.
KeywordsCalorimeter Lide Peaked
Unable to display preview. Download preview PDF.
- Peterson, C. Track finding with neural networks. Nucl. Instr. Methods 1989; A279: 537–545Google Scholar
- Alexopoulos, T. Ph.D. thesis, University of Wisconsin — Madison 1991 (unpublished)Google Scholar
- Hertz, J., Krogh, A., and Palmer, R. G. Introduction to the Theory of Neural Computation. Addison-Wesley, Redwood City, California, 1991Google Scholar