Financial decision‐making using reinforcement learning with Dirichlet priors and quantum‐inspired genetic optimization

Key Points

• The aim is to develop a reinforcement learning framework for dynamic budget allocation that enhances profitability while adhering to historical spending patterns.
• Hybrid reinforcement learning framework combines Dirichlet priors and quantum-inspired genetic algorithms.
• Trained on Apple Inc.'s financial data from 2009 to 2025, focusing on budget allocation between R&D and SG&A.
• State-evolution models and parameterized qubit-rotation circuits are utilized to refine optimization processes.
• Achieved near-perfect alignment with actual allocations based on cosine similarity and KL divergence metrics.
• Demonstrated effective generalization through improved policy behaviors and convergence visualizations.