GA-Attention-Fuzzy-Stock-Net: An optimized neuro-fuzzy system for stock market price prediction with genetic algorithm and attention mechanism

Abstract

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

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 31 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.