Machine learning techniques for background discrimination at the ATLAS experiment

The use of artificial neural network in discrimination between signal and background is investigated for two different processes in the ATLAS detector. For the WH signal against the WZ/Wγ∗ background the neural network shows improvement over the boosted decision trees previously used in Aad et al. [1]. Also for the pair produced stop quark process with two leptons in the final state is archieved good discrimination between the signal and the background of fake/non-prompt leptons.