Continuous-time Online Learning via Mean-Field Neural Networks: Regret Analysis in Diffusion Environments
Provides theoretical regret guarantees for online learning with neural networks in continuous-time diffusion environments, addressing a gap in non-stationary online learning theory.
The paper studies continuous-time online learning with diffusion-generated data, using mean-field neural networks. It establishes constant static regret under displacement convexity and linear regret bounds in non-convex settings, with simulations showing outperformance of the online approach.
We study continuous-time online learning where data are generated by a diffusion process with unknown coefficients. The learner employs a two-layer neural network, continuously updating its parameters in a non-anticipative manner. The mean-field limit of the learning dynamics corresponds to a stochastic Wasserstein gradient flow adapted to the data filtration. We establish regret bounds for both the mean-field limit and finite-particle system. Our analysis leverages the logarithmic Sobolev inequality, Polyak-Lojasiewicz condition, Malliavin calculus, and uniform-in-time propagation of chaos. Under displacement convexity, we obtain a constant static regret bound. In the general non-convex setting, we derive explicit linear regret bounds characterizing the effects of data variation, entropic exploration, and quadratic regularization. Finally, our simulations demonstrate the outperformance of the online approach and the impact of network width and regularization parameters.