Year: 2022
Type: preprint
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.
Source: arXiv (Cornell University)
Authors: Bai, Guangji, Ling, Chen, Zhao, Liang
Institution: Emory University
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
Awards: CAREER: Spatial Network Deep Generative Modeling, Transformation, and Interpretation, OAC Core: SMALL: DeepJIMU: Model-Parallelism Infrastructure for Large-scale Deep Learning by Gradient-Free Optimization, Phase II I/UCRC [George Mason University]: Center for Spatiotemporal Thinking, Computing and Applications., III: Small: Deep Generative Models for Temporal Graph Generation and Interpretation, III: Small: Graph Generative Deep Learning for Protein Structure Prediction, CRII: III: Interpretable Models for Spatio-Temporal Event Forecasting using Social Sensors
OpenAlex ID: W4302013494