Survey on Real Time Financial Signal Representation and Trading Using Recurrent Neural Network

Abstract

The DRL (deep reinforcement learning) techniques are a combination of DL (deep learning) and RL (reinforcement learning). This field is capable of solve wide range complex decision-making problems that were previously inaccessible to the machine. Thus, now DRL is used in many numbers of new applications in areas such as healthcare, robotics, finance, smart grids, and many more. Here, we study the RDNN (Recurrent Deep Neural Network) structure for the recurrent decision making and environment sensing for the online financial trading management. The proposed techniques are composed of the two parts such as DNN (deep neural network) for learning features and RNN (recurrent neural network) for the RL. For improvement the robustness of market summarization and reducing the uncertainty of the data we are using fuzzy learning technique. In previous study DL has been shown the accurate result in the signal processing problems as recognition of speech and image, DL is designing to real trading applications for the financial signal representation.