@article{ART003280656},
author={Liu Ming and Seong-Yoon Shin},
title={Adaptive Multi-Modal Deep Learning for Financial Market Prediction: A Multi-Scale Attention Approach},
journal={Journal of The Korea Society of Computer and Information},
issn={1598-849X},
year={2025},
volume={30},
number={12},
pages={297-304}
TY - JOUR
AU - Liu Ming
AU - Seong-Yoon Shin
TI - Adaptive Multi-Modal Deep Learning for Financial Market Prediction: A Multi-Scale Attention Approach
JO - Journal of The Korea Society of Computer and Information
PY - 2025
VL - 30
IS - 12
PB - The Korean Society Of Computer And Information
SP - 297
EP - 304
SN - 1598-849X
AB - Financial market prediction remains challenging due to complex non-linear dependencies and regime shifts.
Existing multi-modal approaches suffer from limited temporal horizons(5-10 days), simplistic features, and static fusion mechanisms. In this paper, we presents an enhanced dual-channel architecture with three innovations: (1) multi-scale temporal convolution capturing 5-40 day patterns; (2) adaptive cross-modal attention dynamically balancing sentiment and technical signals; (3) extended 60-day windows with 16 technical indicators. Experiments demonstrate 81.39% accuracy versus baseline's 58.23%, with ablation studies confirming individual contributions of 7.2%, 5.8%, and 5.6% respectively.It also outperforms state-of-the-art models like Deep Convolutional Transformer and 2CAT in both short-term and long-term forecasting tasks across multiple global stock indices. Moreover, the model’s interpretability is enhanced through attention weight visualization, enabling practitioners to identify key market drivers during different regimes.
KW - Stock prediction;multi-modal learning;temporal convolution;attention mechanism
DO -
UR -
ER -
Liu Ming and Seong-Yoon Shin. (2025). Adaptive Multi-Modal Deep Learning for Financial Market Prediction: A Multi-Scale Attention Approach. Journal of The Korea Society of Computer and Information, 30(12), 297-304.
Liu Ming and Seong-Yoon Shin. 2025, "Adaptive Multi-Modal Deep Learning for Financial Market Prediction: A Multi-Scale Attention Approach", Journal of The Korea Society of Computer and Information, vol.30, no.12 pp.297-304.
Liu Ming, Seong-Yoon Shin "Adaptive Multi-Modal Deep Learning for Financial Market Prediction: A Multi-Scale Attention Approach" Journal of The Korea Society of Computer and Information 30.12 pp.297-304 (2025) : 297.
Liu Ming, Seong-Yoon Shin. Adaptive Multi-Modal Deep Learning for Financial Market Prediction: A Multi-Scale Attention Approach. 2025; 30(12), 297-304.
Liu Ming and Seong-Yoon Shin. "Adaptive Multi-Modal Deep Learning for Financial Market Prediction: A Multi-Scale Attention Approach" Journal of The Korea Society of Computer and Information 30, no.12 (2025) : 297-304.
Liu Ming; Seong-Yoon Shin. Adaptive Multi-Modal Deep Learning for Financial Market Prediction: A Multi-Scale Attention Approach. Journal of The Korea Society of Computer and Information, 30(12), 297-304.
Liu Ming; Seong-Yoon Shin. Adaptive Multi-Modal Deep Learning for Financial Market Prediction: A Multi-Scale Attention Approach. Journal of The Korea Society of Computer and Information. 2025; 30(12) 297-304.
Liu Ming, Seong-Yoon Shin. Adaptive Multi-Modal Deep Learning for Financial Market Prediction: A Multi-Scale Attention Approach. 2025; 30(12), 297-304.
Liu Ming and Seong-Yoon Shin. "Adaptive Multi-Modal Deep Learning for Financial Market Prediction: A Multi-Scale Attention Approach" Journal of The Korea Society of Computer and Information 30, no.12 (2025) : 297-304.
1