Realtime identification of Dispersed Radio signals using ML - A Case Study on the Crab Pulsar

Robust realtime identification of dispersed radio astronomical signals that last much less than a second is challenging. Here we explore the utility of machine learning techniques to identify such signals and use data taken on the Crab pulsar using the Effelsberg 100m Radio Telescope. The data corresponds to the frequency range of 1240-1510 MHz, and contains 20 minutes of the pulsar signal. In addition, the DM-time data generated by the realtime pipeline, the associated Tensorflow CNN model is included and the training dataset are included. The data are intended for machine learning tasks focused on single-pulse detection and classification.