The effect of phased recurrent units in the classification of multiple catalogues of astronomical light curves

In the new era of very large telescopes, where data is crucial to expand scientific knowledge, we have witnessed many deep learning applications for automatic classification lightcurves. Recurrent neural networks (RNNs) are one models used these applications, and LSTM unit stands out being an excellent choice representation long time series. general, RNNs assume observations at discrete times, which may not suit irregular sampling A traditional technique address sequences consists adding network's input, but this guaranteed capture irregularities during training. Alternatively, Phased has been created problem by updating its state using times explicitly. work, study effectiveness based architectures astronomical We use seven catalogs containing periodic nonperiodic objects. Our findings show that outperformed PLSTM on 6/7 datasets. However, combination both units enhances results in all