Population Aware Diffusion for Time Series Generation

2025-01-01

AI Summary

Existing diffusion models struggle to preserve population-level statistical properties—such as marginal distributions per dimension and cross-dimensional correlation distributions—when generating time series, leading to significant statistical discrepancies between synthetic and real data. To address this, we propose a population-aware diffusion framework featuring a novel statistical-constrained training objective and a dual-channel temporal encoder, explicitly modeling and matching both marginal and cross-correlation distributions of generated sequences. Our method integrates the diffusion process with a statistical consistency loss and a dedicated cross-correlation distribution matching mechanism. Evaluated on multiple benchmark datasets, it reduces average cross-correlation distribution shift by 5.9× compared to prior methods, while achieving state-of-the-art fidelity at the individual sequence level. This work is the first to systematically tackle the lack of population-level statistical consistency in time-series generation, establishing a new paradigm for high-fidelity, trustworthy time-series synthesis.

Technology Category

• Computer Vision: Diffusion Models for Vision
• Natural Language Processing: Generation
• Machine Learning: Time-Series/Data Streams

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

Diffusion models have shown promising ability in generating high-quality time series (TS) data. Despite the initial success, existing works mostly focus on the authenticity of data at the individual level, but pay less attention to preserving the population-level properties on the entire dataset. Such population-level properties include value distributions for each dimension and distributions of certain functional dependencies (e.g., cross-correlation, CC) between different dimensions. For instance, when generating house energy consumption TS data, the value distributions of the outside temperature and the kitchen temperature should be preserved, as well as the distribution of CC between them. Preserving such TS population-level properties is critical in maintaining the statistical insights of the datasets, mitigating model bias, and augmenting downstream tasks like TS prediction. Yet, it is often overlooked by existing models. Hence, data generated by existing models often bear distribution shifts from the original data. We propose Population-aware Diffusion for Time Series (PaD-TS), a new TS generation model that better preserves the population-level properties. The key novelties of PaD-TS include 1) a new training method explicitly incorporating TS population-level property preservation, and 2) a new dual-channel encoder model architecture that better captures the TS data structure. Empirical results in major benchmark datasets show that PaD-TS can improve the average CC distribution shift score between real and synthetic data by 5.9x while maintaining a performance comparable to state-of-the-art models on individual-level authenticity.

Problem

• Diffusion Models
• Time Series Data
• Statistical Characteristics

Innovation

• PaD-TS
• Statistical Feature Preservation
• Dual-Channel Encoding