Uncertainty-Gated Generative Modeling

TL;DR

This paper introduces Uncertainty-Gated Generative Modeling (UGGM), a novel approach for financial time-series forecasting that incorporates uncertainty as a control signal to improve risk-sensitive predictions and robustness.

Contribution

The paper presents UGGM, a new framework that integrates uncertainty into generative modeling for time-series, enhancing calibration and risk-awareness over prior methods.

Findings

63.5% MSE reduction on NYISO dataset

Improved robustness under shock intervals

Enhanced calibration and risk sensitivity

Abstract

Financial time-series forecasting is a high-stakes problem where regime shifts and shocks make point-accurate yet overconfident models dangerous. We propose Uncertainty-Gated Generative Modeling (UGGM), which treats uncertainty as an internal control signal that gates (i) representation via gated reparameterization, (ii) propagation via similarity and confidence routing, and (iii) generation via uncertainty-controlled predictive distributions, together with uncertainty-driven regularization and calibration to curb miscalibration. Instantiated on Weak Innovation AutoEncoder (WIAE-GPF), our UG-WIAE-GPF significantly improves risk-sensitive forecasting, delivering a 63.5\% MSE reduction on NYISO (0.3508 $\rightarrow$ 0.1281), with improved robustness under shock intervals (mSE: 0.2739 $\rightarrow$ 0.1748).