基于CNN-LSTM-STL多任务模型的陶瓷艺术品价格预测研究

Research on Ceramic Art Price Prediction Based on CNN-LSTM-STL Multi-Task Model

徐永欢 ¹詹棠森¹

扫码查看

作者信息

1. 景德镇陶瓷大学信息工程学院,333403,江西,景德镇
折叠

摘要

随着中国经济的高速发展,我国的陶瓷艺术品市场也迅速发展起来,但在价格上却并没有形成系统的制度,造成了价格上的差异.基于雅昌拍卖网的150组陶瓷数据,先利用卷积神经网络(CNN)进行特征提取,将提取得到的数据输入LSTM模型中,最后将CNN-LSTM模型作为多任务模型中的一个单任务模型,从而得到了改进的CNN-LSTM-STL多任务预测模型.经过实证说明,改进CNN-LSTM-STL预测模型显著提高了序列的模拟和预测的精度,其研究结果为陶瓷艺术品价格的预测提供了依据.

Abstract

With the rapid development of the China's economy,the ceramic art market in China has also rapidly developed.However,a systematic pricing framework has not yet been established,leading to significant price disparities.This paper utilizes 150 sets of ceramic data from Yachang Auction website.First,convolutional neural networks(CNN)are used for feature extraction,and the extracted data is input into the LSTM model.Finally,the CNN-LSTM model is used as a single task model in the multi-task model,resulting in an improved CNN-LSTM-STL multi-task prediction model.Empirical results show that the improved CNN-LSTM-STL prediction model significantly enhances the accuracy of sequence simulation and prediction,and its research results provide a basis for predicting the price of ceramic artworks.

关键词

陶瓷艺术品/卷积神经网络(CNN)/多任务预测/CNN-LSTM-STL预测模型

Key words

ceramic artworks/Convolutional Neural Network(CNN)/multi-task prediction/CNN-LSTM-STL prediction model

引用本文复制引用

出版年

2024

江西科学

江西省科学院

江西科学

影响因子：0.286

ISSN：1001-3679

段落导航