Decision-Aware Conditional GANs for Time Series Data

TL;DR

This paper presents DAT-CGAN, a novel decision-aware generative model for time series that enhances decision support by incorporating decision-related quantities, improving sample efficiency, and demonstrating superior performance on financial data.

Contribution

The paper introduces DAT-CGAN, a decision-aware GAN framework for time series that captures decision-related heterogeneity and improves generative quality for decision-making tasks.

Findings

Better generative quality than GAN baselines.

Improved sample efficiency through overlapped block-sampling.

Effective in financial portfolio decision support.

Abstract

We introduce the decision-aware time-series conditional generative adversarial network (DAT-CGAN) as a method for time-series generation. The framework adopts a multi-Wasserstein loss on structured decision-related quantities, capturing the heterogeneity of decision-related data and providing new effectiveness in supporting the decision processes of end users. We improve sample efficiency through an overlapped block-sampling method, and provide a theoretical characterization of the generalization properties of DAT-CGAN. The framework is demonstrated on financial time series for a multi-time-step portfolio choice problem. We demonstrate better generative quality in regard to underlying data and different decision-related quantities than strong, GAN-based baselines.