PathFM: Fast Generative Forecasting of Multivariate Trajectories
Time series foundation models are becoming increasingly popular for general-purpose forecasting. However, most leading models predict marginal quantiles, leaving the dependence structure of future trajectories unspecified. This distinction matters for trajectory-dependent quantities such as cumulative demand, peak load, and event timing, yet point and marginal scores do not rev
Reading in Plain English mode — toggle below to switch
Time series foundation models are becoming increasingly popular for general-purpose forecasting. However, most leading models predict marginal quantiles, leaving the dependence structure of future trajectories unspecified. This distinction matters for trajectory-dependent quantities such as cumulative demand, peak load, and event timing, yet point and marginal scores do not rev
This paper introduces a novel approach to multivariate time series forecasting that addresses a critical limitation of current foundation models: their inability to specify the dependence structure of future trajectories. By generating complete future paths efficiently, PathFM offers a significant advancement for applications requiring trajectory-dependent insights, such as cumulative demand or peak load prediction.