Scholarlink

<p>With recent technological advances, process logs, which were traditionally deterministic in nature, are being captured from non-deterministic sources, such as uncertain sensors or machine learning models (that predict activities using cameras). In the presence of stochastically known logs, logs that contain probabilistic information, the need for stochastic trace recovery increases, to offer reliable means of understanding the processes that govern such systems. We design a novel deep learning approach for stochastic trace recovery, based on Denoising Diffusion Probabilistic Models (DDPM). In particular, we investigate multiple methods for injecting process knowledge into the DDPM to improve the overall performance of trace recovery. We conduct an empirical evaluation demonstrating state-of-the-art performance with up to a 25% improvement over the previous state-of-the-art method, along with increased robustness under high noise levels. We also demonstrate the pros and cons of the different approaches to process embedding.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Efficient process embedding for diffusion-based trace recovery

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年6月3日

查看全文
  • Maximilian Matyash,
  • Avigdor Gal,
  • Arik Senderovich

摘要

With recent technological advances, process logs, which were traditionally deterministic in nature, are being captured from non-deterministic sources, such as uncertain sensors or machine learning models (that predict activities using cameras). In the presence of stochastically known logs, logs that contain probabilistic information, the need for stochastic trace recovery increases, to offer reliable means of understanding the processes that govern such systems. We design a novel deep learning approach for stochastic trace recovery, based on Denoising Diffusion Probabilistic Models (DDPM). In particular, we investigate multiple methods for injecting process knowledge into the DDPM to improve the overall performance of trace recovery. We conduct an empirical evaluation demonstrating state-of-the-art performance with up to a 25% improvement over the previous state-of-the-art method, along with increased robustness under high noise levels. We also demonstrate the pros and cons of the different approaches to process embedding.