Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Enabling Efficient and Flexible Interpretability of Data-driven Anomaly Detection in Industrial Processes with AcME-AD

Valentina Zaccaria, Chiara Masiero, David Dandolo, Gian Antonio Susto

TL;DR

The first industrial application of AcME-AD is presented, showcasing its effectiveness through experiments and demonstrating AcME-AD’s potential as a valuable tool for explainable AD and feature-based root cause analysis within industrial environments, paving the way for trustworthy and actionable insights in the age of Industry 5.0.

Abstract

While Machine Learning has become crucial for Industry 4.0, its opaque nature hinders trust and impedes the transformation of valuable insights into actionable decision, a challenge exacerbated in the evolving Industry 5.0 with its human-centric focus. This paper addresses this need by testing the applicability of AcME-AD in industrial settings. This recently developed framework facilitates fast and user-friendly explanations for anomaly detection. AcME-AD is model-agnostic, offering flexibility, and prioritizes real-time efficiency. Thus, it seems suitable for seamless integration with industrial Decision Support Systems. We present the first industrial application of AcME-AD, showcasing its effectiveness through experiments. These tests demonstrate AcME-AD's potential as a valuable tool for explainable AD and feature-based root cause analysis within industrial environments, paving the way for trustworthy and actionable insights in the age of Industry 5.0.

Enabling Efficient and Flexible Interpretability of Data-driven Anomaly Detection in Industrial Processes with AcME-AD

TL;DR

The first industrial application of AcME-AD is presented, showcasing its effectiveness through experiments and demonstrating AcME-AD’s potential as a valuable tool for explainable AD and feature-based root cause analysis within industrial environments, paving the way for trustworthy and actionable insights in the age of Industry 5.0.

Abstract

While Machine Learning has become crucial for Industry 4.0, its opaque nature hinders trust and impedes the transformation of valuable insights into actionable decision, a challenge exacerbated in the evolving Industry 5.0 with its human-centric focus. This paper addresses this need by testing the applicability of AcME-AD in industrial settings. This recently developed framework facilitates fast and user-friendly explanations for anomaly detection. AcME-AD is model-agnostic, offering flexibility, and prioritizes real-time efficiency. Thus, it seems suitable for seamless integration with industrial Decision Support Systems. We present the first industrial application of AcME-AD, showcasing its effectiveness through experiments. These tests demonstrate AcME-AD's potential as a valuable tool for explainable AD and feature-based root cause analysis within industrial environments, paving the way for trustworthy and actionable insights in the age of Industry 5.0.
Paper Structure (12 sections, 1 equation, 5 figures, 2 tables)

This paper contains 12 sections, 1 equation, 5 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Proposed Approach
  4. Experimental Results
  5. Use Case I: Chemical processes
  6. Overall feature importance
  7. Local Interpretability
  8. Use Case II: Packaging equipment
  9. Model selection with AcME-AD
  10. Local Interpretability
  11. AcME-AD and KernelSHAP: time comparison
  12. Conclusions

Figures (5)

  • Figure 1: AcME-AD overall importance bar chart using IF on TEP.
  • Figure 2: KernelSHAP overall importance bar chart using IF on TEP and 10% sampling of the background dataset.
  • Figure 3: AcME-AD Local explanation of an anomaly identified by IF in TEP. It emerges that this data point would be considered normal by modifying the values of xmeas_11 or xmv_9.
  • Figure 4: AcME-AD Local explanation of a randomly sampled anomaly identified by LODA in PIADE.
  • Figure 5: Overall importance bar chart for LODA on PIADE sequences - Equipment 2.