Random sampling method for cryptocurrency market time series forecasting

This paper applies Random Sampling Method (RSM) to classification task for cryptocurrencies time series, which are not-stationary Long Short Term Memory (LSTM) networks have been demonstrated to be particularly useful for learning sequences containing longer term patterns of unknown length, such as at this task. But RSM represents another deep learning algorithm with more flexible architecture, built on the basis of LSTM cells and thus having all the advantages of the traditional algorithm, but more resistant to the class imbalance problem. The main distinguishing feature of RSM is the use of metric learning and special sampling scheme.