همایش بین المللی هوش مصنوعی

صفحه اصلی / دومین همایش بین المللی هوش مصنوعی

WealthAI: An End-to-End Framework for Explainable and Personalized AI-Driven Wealth Management

نویسندگان :

Mohammadali Soltanshahi¹ Amirhossein Kafi² Mojtaba Raeessafari³

1- آزاد 2- آزاد 3- آزاد

کلمات کلیدی :

Artificial Intelligence،Wealth Management،Deep Reinforcement Learning،Explainable AI،Large Language Models،Federated Learning،Portfolio Optimization،Algorithmic Trading،Robo-Advisors،Financial Ethics

چکیده :

The convergence of deep reinforcement learning, large language models, and federated learning is redefining the landscape of wealth management. This paper introduces WealthAI, a novel, end-to-end framework that integrates these technologies to deliver personalized, explainable, and privacy-preserving investment strategies. Our architecture features a multi-agent deep reinforcement learning (DRL) core for dynamic portfolio optimization, an explainable AI (XAI) module that generates human-readable rationales for every decision, and a federated learning protocol that ensures client data privacy. We evaluate WealthAI on a comprehensive dataset spanning 14 years (2010–2024) across equities, cryptocurrencies, and commodities. Results demonstrate a 22.7% annualized return with a Sharpe ratio of 1.68, significantly outperforming traditional benchmarks. Crucially, our XAI module increases user trust by 67% in user studies, while federated learning reduces data leakage risk to near zero. This work provides a practical, ethical, and high-performance blueprint for the future of AI in finance.