Title: E4GEN: Event-level Explainable Extreme-Enhanced Time-series Generation

Original: arXiv:2606.01634v1 Announce Type: cross Abstract: Generating realistic time series is essential for scientific research and real-world applications. However, existing methods often emphasize overall distributional fidelity while failing to faithfully capture extreme events. To advance existing research, we propose E4GEN, an explainable diffusion framework for extreme event-aware time-series generation. E4GEN provides systematic insights into when, what, and how to control extreme-event generation through three key components. First, E-Activator learns the dataset-adaptive extreme-control signal activation step during the denoising process without interfering with regular temporal components, including trend and seasonality. Second, E-Predictor determines what control signal to enforce through Self-Driven Semantic Prediction, where each sample derives its own control signal by inferring latent extreme-event information during generation. It also includes a novel Data-Conditioned Training, Noise-Initiated Sampling mechanism to address the issue of unavailable training labels. Third, E-Control specifies how to control extreme-event generation through a trainable Extreme Control Network, which transforms the semantic control signal into layer-wise signals and injects it into the denoising process. We evaluate E4GEN on six datasets with 17 metrics, and extensive experiments show that E4GEN outperforms state-of-the-art models across multiple dimensions, including overall fidelity, extreme-event fidelity, and downstream utility.

Rewritten:

Title: E4GEN: Event-level Explainable Extreme-Enhanced Time-series Generation

Abstract: The creation of authentic time-series data is a critical requirement for both practical implementations and scientific inquiry. Nevertheless, current approaches typically prioritize general distributional accuracy, often neglecting the accurate representation of extreme events. To bridge this gap, we introduce E4GEN, a diffusion-based framework designed for explainable, extreme-event-aware time-series generation. This system offers a structured understanding of the timing, nature, and methodology of controlling extreme events via three distinct modules.

Initially, the E-Activator component identifies the optimal step for activating extreme-control signals during denoising, adapting to specific datasets without disrupting standard temporal features such as seasonality and trends. Secondly, the E-Predictor utilizes Self-Driven Semantic Prediction to decide which control signals to apply. In this process, each individual sample generates its own control signal by deducing latent extreme-event data as generation progresses. To tackle the challenge of missing training labels, this module incorporates a unique Data-Conditioned Training and Noise-Initiated Sampling approach. Finally, the E-Control module dictates the mechanism of control through a trainable Extreme Control Network. This network converts semantic control signals into layer-specific signals, injecting them directly into the denoising workflow.

Our evaluation of E4GEN spans six distinct datasets utilizing 17 different metrics. Comprehensive testing demonstrates that E4GEN surpasses current state-of-the-art models in several areas, notably regarding overall fidelity, the fidelity of extreme events, and utility in downstream tasks.

Source: arXiv Generated at: 2026-06-02 00:00:00 UTC