GA-Attention-Fuzzy-Stock-Net: An Optimized Neuro-Fuzzy System for Stock Market Price Prediction with Genetic Algorithm and Attention Mechanism

Heliyon, Journal Year: 2025, Volume and Issue: 11(3), P. e42393 - e42393

Published: Feb. 1, 2025

Highlights:•GA-Attention-Fuzzy-Stock-Net outperforms other models in stock price prediction across metrics•Trapezoidal fuzzy membership functions show superior performance neuro-fuzzy systems•Sliding window size impacts accuracy, with 20-day yielding best results•Genetic algorithm effectively optimize hyperparameters for attention mechanism•Comprehensive comparison of architectures advances forecasting methodsAbstractThis study introduces GA-Attention-Fuzzy-Stock-Net, a novel hybrid architecture that integrates genetic algorithms, mechanisms, and systems market prediction. The research investigates the effectiveness different architectural configurations, including variations layer (triangular, trapezoidal, Gaussian) neural network (1D ANN, 2D 1D LSTM, LSTM). model's is evaluated multiple temporal horizons using sliding windows (5-day, 10-day, 20-day) to capture varying dynamics. Genetic algorithms hyperparameters, learning rates architectures, while mechanism enhances ability focus on relevant patterns. utilizes data from major technology stocks. Results demonstrate GA-Attention-Fuzzy-Stock-Net consistently traditional machine approaches baseline evaluation metrics (MSE, MAE, MAPE, R2). findings provide valuable insights practitioners financial markets contribute advancement intelligent time series attributed its unique integration evolutionary optimization, attention-based feature selection, logic's handle uncertainty data.

Language: Английский

---

Hybrid CNN-BiGRU-AM Model with Anomaly Detection for Nonlinear Stock Price Prediction

Electronics, Journal Year: 2025, Volume and Issue: 14(7), P. 1275 - 1275

Published: March 24, 2025

To address challenges in stock price prediction including data nonlinearity and anomalies, we propose a hybrid CNN-BiGRU-AM framework integrated with deep learning-based anomaly detection. First, an detection module identifies irregularities data. The CNN component then extracts local features while filtering anomalous information, followed by nonlinear pattern modeling through BiGRU attention mechanisms. Final predictions undergo secondary screening to ensure reliability. Experimental evaluation on Shanghai Composite (SSE) daily closing prices demonstrates superior performance R2 = 0.9903, RMSE 22.027, MAE 19.043, Sharpe Ratio of 0.65. It is noteworthy that the this model reduced 14.7%, decreased 7.7% compared its ablation model. achieves multi-level feature extraction convolutional operations bidirectional temporal modeling, effectively enhancing generalization via mapping correction. Comparative analysis across models provides practical insights for investment decision-making. This dual-functional system not only improves accuracy but also offers interpretable references market mechanism regulatory policy formulation.

Language: Английский

---

Cross-modal retrieval based on multi-dimensional feature fusion hashing

Frontiers in Physics, Journal Year: 2024, Volume and Issue: 12

Published: June 19, 2024

Along with the continuous breakthrough and popularization of information network technology, multi-modal data, including texts, images, videos, audio, is growing rapidly. We can retrieve different modal data to meet our needs, so cross-modal retrieval has important theoretical significance application value. In addition, because modalities be mutually retrieved by mapping them a unified Hamming space, hash codes have been extensively used in field. However, existing hashing models generate based on single-dimension features, ignoring semantic correlation between features dimensions. Therefore, an innovative method using Multi-Dimensional Feature Fusion Hashing (MDFFH) proposed. To better get image’s multi-dimensional convolutional neural network, Vision Transformer are combined construct image fusion module. Similarly, we apply text module modality obtain text’s features. These two modules effectively integrate dimensions through feature fusion, making generated code more representative semantic. Extensive experiments corresponding analysis results datasets indicate that MDFFH’s performance outdoes other baseline models.

Language: Английский

---

Intelligent Stock Forecasting by Iterative Global-Local Fusion

Lecture notes in computer science, Journal Year: 2024, Volume and Issue: unknown, P. 285 - 295

Published: Jan. 1, 2024

Language: Английский

---