Athena X-IFU event reconstruction: Extreme Learning Machine approach

Abstract

SIRENA is the software aimed at performing the on board event energy reconstruction for the Athena calorimeter X-IFU, in the Digital Readout Electronics unit. Processing will consist in an initial triggering of event pulses followed by an analysis (with SIRENA) to determine the energy content of events. Optimal filtering has been chosen as the baseline algorithm but other techniques are still under study in an effort to get the better results at the lower computing cost. Here we show the performance of the Extreme Learning Machine (ELM) algorithm for single-hidden layer feedforward neural networks (SLFNs).