An algorithm for the formation of comvolutional ensembles in non-recurrent neural networks for processing time series forecasts in automatic trading systems

Gorelov Mikhail Andreevich

The article deals with the main features of convolutional neural networks, their structure and application features. An algorithm for forming convolutional ensembles for processing time series forecasts based on the averaging of subnetwork predicates is proposed.

Keywords:neural network, time series, recurrent neural network, forecast.