Temporal Domain Generalization with Drift-Aware Dynamic Neural Networks

TL;DR

This paper introduces DRAIN, a drift-aware dynamic neural network framework for temporal domain generalization, modeling data and model dynamics to predict future models and guarantee performance under distribution shifts.

Contribution

The paper proposes a Bayesian framework with recurrent graph neural networks to model temporal drift and provides theoretical guarantees for model performance in temporal domain generalization.

Findings

Effective in predicting models under temporal drift

Theoretical analysis confirms generalization bounds

Outperforms existing methods on real-world benchmarks

Abstract

Temporal domain generalization is a promising yet extremely challenging area where the goal is to learn models under temporally changing data distributions and generalize to unseen data distributions following the trends of the change. The advancement of this area is challenged by: 1) characterizing data distribution drift and its impacts on models, 2) expressiveness in tracking the model dynamics, and 3) theoretical guarantee on the performance. To address them, we propose a Temporal Domain Generalization with Drift-Aware Dynamic Neural Network (DRAIN) framework. Specifically, we formulate the problem into a Bayesian framework that jointly models the relation between data and model dynamics. We then build a recurrent graph generation scenario to characterize the dynamic graph-structured neural networks learned across different time points. It captures the temporal drift of model parameters and data distributions and can predict models in the future without the presence of future data. In addition, we explore theoretical guarantees of the model performance under the challenging temporal DG setting and provide theoretical analysis, including uncertainty and generalization error. Finally, extensive experiments on several real-world benchmarks with temporal drift demonstrate the effectiveness and efficiency of the proposed method.