Scholarlink

In smart manufacturing systems, anomaly detection is critical for maintaining operational efficiency, safety, and reliability, especially as industrial processes evolve from automatic systems in Industry 4.0 toward more human-centric frameworks in Industry 5.0 that incorporate human intelligence. In this chapter, we investigate the integration of artificial intelligence (AI) and humans, particularly focusing on Explainable AI’s framework to explain its role in helping to understand the sophistication of AI-based anomaly detection systems. The discussion highlights how approaches to anomaly detection have changed from manual checking to more advanced technological solutions, such as machine learning and deep learning, while emphasizing the importance of human understanding. This research outlined important gaps, such as the tradeoff of performance and interpretability of the AI, privacy concerns, and the adequacy of the explanations provided, while highlighting the need for more reliable, safe, and cooperative AI systems in future manufacturing settings. Lastly, the chapter includes a practical case study on the application of human-centered XAI design alternatives to enhance transparency and make human validation easier as a method for trust and collaboration with the AI system.

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

Human-Centered Explainable Anomaly Detection in Smart Manufacturing: Bridging AI and Human Decision-Making in Industry 5.0

  • 章节

  • 发表日期：

    2026年

查看全文
  • Dac Hieu Nguyen,
  • Dac Phuong Thao Nguyen,
  • Quang Chieu Ta,
  • Kim Duc Tran,
  • Kim Phuc Tran

摘要

In smart manufacturing systems, anomaly detection is critical for maintaining operational efficiency, safety, and reliability, especially as industrial processes evolve from automatic systems in Industry 4.0 toward more human-centric frameworks in Industry 5.0 that incorporate human intelligence. In this chapter, we investigate the integration of artificial intelligence (AI) and humans, particularly focusing on Explainable AI’s framework to explain its role in helping to understand the sophistication of AI-based anomaly detection systems. The discussion highlights how approaches to anomaly detection have changed from manual checking to more advanced technological solutions, such as machine learning and deep learning, while emphasizing the importance of human understanding. This research outlined important gaps, such as the tradeoff of performance and interpretability of the AI, privacy concerns, and the adequacy of the explanations provided, while highlighting the need for more reliable, safe, and cooperative AI systems in future manufacturing settings. Lastly, the chapter includes a practical case study on the application of human-centered XAI design alternatives to enhance transparency and make human validation easier as a method for trust and collaboration with the AI system.