Guest Editors
Prof. Lianbo Ma
Email: malb@swc.neu.edu.cn
Affiliation: College of Software, Northeastern University, Shenyang, 110819, China
Homepage:
Research Interests: Neural architecture search, machine learning, edge computing
Dr. Shangce Gao
Email: gaosc@eng.u-toyama.ac.jp
Affiliation: Faculty of Engineering, University of Toyama, Toyama, 930-8555, Japan
Homepage:
Research Interests: Machine learning, intelligent optimization, edge computing
Dr. Guo Yu
Email: guo.yu@njtech.edu.cn
Affiliation: Institute of Intelligent Manufacturing, Nanjing Tech University, Nanjing, 211816, China
Homepage:
Research Interests: Neural networks, network communication
Dr. Shi Cheng
Email: cheng@snnu.edu.cn
Affiliation: School of Computer Science, Shannxi Normal University, Xi'an, 710119, China
Homepage:
Research Interests: Large language model, network communication
Summary
The convergence of Artificial Intelligence (AI) and the Industrial Internet of Things (IoT) has brought forward new opportunities for enhancing efficiency, safety, and automation across a wide range of industrial applications. With the growing prevalence of edge computing in industrial scenarios, such as smart factories, power grids, and logistics systems. There is an increasing demand for intelligent models that can operate efficiently under strict resource constraints. Intelligent computation (e.g., Evolutioanry Computation, Neural Architecture Search (NAS), Neural Network, and Renforcement Learning) and Large Machine Learning Models (LMLMs) have emerged as promising technologies to support scalable, adaptive, and intelligent decision-making at the edge.
However, deploying high-performing AI models in Industrial Internet environments presents several critical challenges. Industrial edge devices, such as programmable logic controllers (PLCs) and embedded sensors, typically have limited computational power, memory, and energy budgets. Moreover, industrial IoT systems often operate under dynamic and harsh conditions that require real-time processing, high reliability, and domain-specific adaptability. This involves complex tasks, such as task scheduling and resource allocation, which requires the suppot of edge intelligence. Traditional manually-designed AI models struggle to meet these multifaceted requirements. Furthermore, industrial datasets are often small, noisy, or unlabeled, making it difficult to train robust models without access to large-scale annotated corporation.
To address these issues, this special session aims to explore how the integration of intelligent computation and LMLMs can facilitate the development of edge-intelligent solutions in Industrial IoT contexts. Intelligent computation (e.g., NAS) enables automated model customization and resource-aware optimization, generating efficient architectures tailored to the capabilities of specific edge hardware. LMLMs provide broad generalization power and knowledge-rich representations that can be compressed or distilled into smaller, task-specific models. When combined, they can jointly support intelligent, adaptive, and low-latency decision-making on edge devices, enabling smart industrial systems to continuously evolve with changing operational demands.
Main Topics:
· Intelligent computation techniques for industrial Internet of Things
· Resource allocation for edge computing and cloud computing
· Task scheduling for cloud-edge-end cooperation
· Enhanced AI for edge intelligence
· Joint intelligent search and large model frameworks for industrial Internet of Things
· Model compression and knowledge distillation for edge AI
· Evolutionary computation for cloud-edge networks
· Real-time inference with optimized intelligent computation models
· Federated learning and NAS methods
· Benchmarking NAS and large models in industrial tasks
· Multimodal large models adaptation for information fusion
· Case studies in smart factories and logistics systems
Keywords
Intelligent computation, Neural architecture search, Large machine learning models, Edge Intelligence, Industrial Internet of Things, Resource allocation
Published Papers
Login
Register
Submit a Paper
Propose a Special lssue