Multivariate Identification of Background Contributions for the H ! tt

Within the H ! tt analysis it is very important to understand the background contamination in the signal region coming from events where a jet is misidentified as a hadronic tau (fake events). Currently, the fake rate method is used to estimate the number and distributions of fake events in the signal region. This method relies on the correct identification of different background types. The study presented in this report focuses on the use of boosted decision trees in order to identify different background types. It is shown how the addition of more input variables, leading to a multi-dimensional multi-classification task, improves the overall identification accuracy of the different background types.