🤖 AI Summary
To address the challenges of modeling spatiotemporal dependencies and generating high-fidelity scenarios in large-scale, multi-stakeholder systems (e.g., power grids, supply chains) suffering from data silos and privacy constraints, this paper proposes SplitVAE—the first decentralized variational autoencoder framework. SplitVAE enables collaborative learning of the global historical distribution without sharing raw data, leveraging federated gradient exchange and cross-node latent space alignment. It thus simultaneously ensures privacy preservation, communication efficiency, and scenario fidelity. Extensive experiments on multiple real-world distributed systems demonstrate that SplitVAE achieves scenario generation quality comparable to state-of-the-art centralized methods, while reducing communication overhead by over 90%. This substantial reduction significantly enhances scalability and robustness under heterogeneous and dynamic network conditions.
📝 Abstract
Stochastic optimization problems in large-scale multi-stakeholder networked systems (e.g., power grids and supply chains) rely on data-driven scenarios to encapsulate complex spatiotemporal interdependencies. However, centralized aggregation of stakeholder data is challenging due to the existence of data silos resulting from computational and logistical bottlenecks. In this paper, we present SplitVAE, a decentralized scenario generation framework that leverages variational autoencoders to generate high-quality scenarios without moving stakeholder data. With the help of experiments on distributed memory systems, we demonstrate the broad applicability of SplitVAEs in a variety of domain areas that are dominated by a large number of stakeholders. Furthermore, these experiments indicate that SplitVAEs can learn spatial and temporal interdependencies in large-scale networks to generate scenarios that match the joint historical distribution of stakeholder data in a decentralized manner. Lastly, the experiments demonstrate that SplitVAEs deliver robust performance compared to centralized, state-of-the-art benchmark methods while significantly reducing data transmission costs, leading to a scalable, privacy-enhancing alternative to scenario generation.