Optimized Calorimeter Signal Compaction for an Independent Component based ATLAS Electron/Jet Second-level Trigger
Eduardo Simas, Joao SEIXAS, Luiz Caloba · 2009
The ATLAS online trigger system has three filtering levels and relies very much on calorimeter information, which is segmented into seven detection layers.Due to differences both in depth and cell granularity of these layers, trigger algorithms may benefit from performing feature extraction at the layer level.This work addresses electron/jet separation at the second level (LVL2) filtering restricted to calorimeter data.Segmented Independent Component Analysis (SICA) is applied over the calorimeter layers in order to extract relevant features for particle identification.The number of independent components to be extracted from a Region of Interest (RoI) is estimated through different signal compaction strategies, such as Principal Component Analysis, Nonlinear Principal Component Analysis and Principal Components for Discrimination.These compaction techniques are evaluated with respect to dimensionality reduction (and processing speed) and classification efficiency.The hypothesis testing is performed by a Multi-Layer Perceptron classifier fed from the segmented independent components.It is shown that the proposed discriminators outperform the baseline design for ATLAS second-level trigger system, achieving a detection efficiency of 99% for a rejection factor smaller than 2%.