Sequence models outperform LLMs in predictive process monitoring tasks

A new research paper compares the effectiveness of three distinct modeling approaches for predictive process monitoring (PPM). The study evaluates traditional deep sequence models like Long Short-Term Memory (LSTM), foundation models utilizing Large Language Models (LLMs), and tabular foundation models with in-context learning capabilities. The findings indicate that while LLMs are increasingly explored for PPM, sequence models consistently outperform them in next activity prediction tasks. Tabular foundation models show competitiveness in temporal prediction, though LLMs generally lag behind these and sequence models, despite their higher computational cost. AI

IMPACT Sequence models remain competitive for next activity prediction, suggesting LLMs may not be universally superior for all process monitoring tasks.

RANK_REASON The cluster contains a research paper published on arXiv comparing different modeling approaches for predictive process monitoring. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.LG →

• alphaXiv
• arXiv
• CatalyzeX
• Connected Papers
• DagsHub
• Gotit.pub
• Hugging Face
• IArxiv
• Litmaps
• long short-term memory
• ScienceCast
• scite Smart Citations

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →