Deep Learning and Elicitability for McKean-Vlasov FBSDEs With Common Noise
Researchers have developed a new numerical method for solving complex McKean-Vlasov forward-backward stochastic differential equations (MV-FBSDEs) that incorporate common noise. This approach leverages elicitability to create a pathwise loss function, enabling neural networks to efficiently approximate both the backward process and conditional expectations without needing costly nested Monte Carlo simulations. The method has been validated on models related to systemic risk in finance and economic growth, demonstrating its accuracy and flexibility for problems without analytical solutions. AI