Scholarlink

<p>This paper delves into advanced methods specifically designed to improve<!--Query ID="Q1" Text="Please check and confirm if the authors and their respective affiliations have been correctly identified. Amend if necessary." Resolved="yes"--> the efficiency of digital twin technologies by leveraging machine learning and deep learning techniques. While digital twins have become integral in areas like predictive maintenance, system optimization, and real-time monitoring, their widespread adoption is often hindered by the substantial computational demands they impose. To address this, we explore key methods aimed at enhancing processing speed and reducing resource consumption, which are crucial for enabling digital twins to operate efficiently in real-world applications. The paper focuses on techniques such as split computing, which distributes computational tasks across multiple devices to reduce latency; multi-exit architectures, allowing early prediction and termination in low-complexity scenarios; and compression techniques like pruning and quantization, which reduce the model size and computational load without significantly impacting accuracy. These methods are particularly important in resource-constrained environments, where devices with limited processing power and battery life are prevalent. By integrating these methods, digital twins can achieve real-time responsiveness and operate efficiently even under challenging conditions. In addition, we clarify the distinct contributions of machine learning and deep learning in the context of digital twins, highlighting their role in improving predictive accuracy, decision-making, and overall system adaptability. This comprehensive overview of efficiency-enhancing techniques paves the way for future research and broader application<!--Query ID="Q2" Text="Author’s biography and photograph are desired. Please provide. Otherwise, please confirm if unnecessary." Resolved="yes"--> of digital twins across diverse industries.</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

Enhancing computational efficiency in digital twins: a survey of techniques and challenges for fast inference

  • 期刊论文

  • 发表日期：

    2025年10月20日

查看全文
  • Marrone Dantas,
  • Miguel LPC Silva,
  • Assis T. de Oliveira Filho,
  • Diego de Freitas Bezerra,
  • Eduardo Freitas,
  • Pedro R. X. do Carmo,
  • Djamel Sadok,
  • Judith Kelner,
  • Ricardo Souza

摘要

This paper delves into advanced methods specifically designed to improve the efficiency of digital twin technologies by leveraging machine learning and deep learning techniques. While digital twins have become integral in areas like predictive maintenance, system optimization, and real-time monitoring, their widespread adoption is often hindered by the substantial computational demands they impose. To address this, we explore key methods aimed at enhancing processing speed and reducing resource consumption, which are crucial for enabling digital twins to operate efficiently in real-world applications. The paper focuses on techniques such as split computing, which distributes computational tasks across multiple devices to reduce latency; multi-exit architectures, allowing early prediction and termination in low-complexity scenarios; and compression techniques like pruning and quantization, which reduce the model size and computational load without significantly impacting accuracy. These methods are particularly important in resource-constrained environments, where devices with limited processing power and battery life are prevalent. By integrating these methods, digital twins can achieve real-time responsiveness and operate efficiently even under challenging conditions. In addition, we clarify the distinct contributions of machine learning and deep learning in the context of digital twins, highlighting their role in improving predictive accuracy, decision-making, and overall system adaptability. This comprehensive overview of efficiency-enhancing techniques paves the way for future research and broader application of digital twins across diverse industries.