Precious Metal Price Prediction Based on Deep Regularization Self-Attention Regression

Junhao Zhou,Xiaoping Yang,Hao Wang,Hong-Ning Dai,Zhanhong He,Wenjuan Lian,Ya Nan Song

doi:10.1109/access.2019.2962202

Abstract

It is non-trivial to predict the prices of precious metals since a number of factors can affect the fluctuations of precious metal prices. Either parametric models or machine learning models cannot accurately forecast the precious metal prices. Though deep learning approaches show their strengths in extracting key features from complicated data, they have the limitations of learning localization and losing some temporal and spatial features. The recent advances in attention mechanisms bring the opportunities to overcome the limitation of deep learning models. In this paper, we originally propose a Regularization Self-Attention Regression Model for precious metal price prediction. In particular, the proposed RSAR model consists of convolutional neural network (CNN) component and Long Short-Term Memory Neural Networks (LSTM) component. Integrating with self-attention mechanism, this model can extract both spatial and temporal features from precious metal price data. Meanwhile, the proper configuration of regularization functions can also lead to the further performance improvement. Extensive experiments on realistic precious metal price dataset show that our proposed approach outperforms other conventional machine learning and deep learning methods.

Highlights

Precious metals typically have higher economic values while they are rare or difficult to be acquired
Motivated by the advances in composite Deep learning (DL) models and attention mechanisms, we propose and develop Regularization Self-Attention Regression Model (RSAR model) to predict daily precious metal prices
RSA mechanism can improve the performance via employing regularization functions (i.e., L1 and L2)

Summary

Introduction

Precious metals typically have higher economic values while they are rare or difficult to be acquired. Precious metals such as Silver and Gold have been used as currency equivalents (or money) while they have been mainly leveraged as financial and industrial commodities recently. It is non-trivial to predict their prices based on historical data since they are often influenced by a number of social-economic factors including production, circulation, industrial demands, sentiment of market. Economists and investors begin to employ machine learning (ML) methods to analyze the massive financial data and predict

Objectives

Results

Conclusion