Deep Learning based Optimization Model for Digital Currencies and Investment Portfolios

DOI

10.2991/ahis.k.220601.031How to use a DOI?

Keywords

Deep Learning; Digital Currency; Portfolio Model

Abstract

The portfolio problem refers to investors or financial institutions that, in managing investments in all types of bonds, stocks, peripheral financial products, depending on the proportion of value held, constantly recombine to achieve stable and loss-free profits, but also to ensure that risk is minimized. Or control the risk within a certain predictable range and look for the most profitable combination of opportunities. In order to solve the portfolio problem in the digital currency market, this paper attempts model optimization based on a portfolio model based on a deep reinforcement learning framework proposed by previous authors. In this paper, based on the above model, we optimize the experience pool and avoid overfitting the model. Moreover, we perform experience repetition on the experience pool and L1 and L2 regularization on the model to prevent the model from being overfitted and thus improve the fault tolerance of the model. Finally, the Deep Reinforcement Learning algorithm is used to make direct trading decisions to fully explore and learn from the digital currency market and develop reasonable portfolio strategies.

Copyright

© 2022 The Authors. Published by Atlantis Press International B.V.

Open Access

This is an open access article distributed under the CC BY-NC 4.0 license.

Download article (PDF)

TY  - CONF
AU  - Jieyan Dong
PY  - 2022
DA  - 2022/06/02
TI  - Deep Learning based Optimization Model for Digital Currencies and Investment Portfolios
BT  - Proceedings of the 2021 International conference on Smart Technologies and Systems for Internet of Things (STS-IOT 2021)
PB  - Atlantis Press
SP  - 160
EP  - 165
SN  - 2589-4919
UR  - https://doi.org/10.2991/ahis.k.220601.031
DO  - 10.2991/ahis.k.220601.031
ID  - Dong2022
ER  -