Summary of the Special Session
Neural architecture search (NAS) has demonstrated superior performance across many tasks compared to manually designed counterparts, thereby gaining significant traction in deep learning. Various optimizers, including reinforcement learning, differentiable methods, and evolutionary computation (EC), have been employed to discover high-performance network architectures. Among these, EC-based NAS (also known as evolutionary neural architecture search (ENAS)) has become a mainstream method, largely due to its powerful capabilities in searching for global optima and handling non-convex and non-differentiable problems. Currently, EC has advanced in-depth research across several dimensions, including decision space to objective space, single-objective to many-objective, and low-dimensional to high-dimensional problems. However, these emerging theories have not been deeply integrated into ENAS, leading to challenges such as low search efficiency and limited application scenarios. Therefore, the theme of this special session is to bring together researchers exploring ENAS from a deep optimization perspective. Specifically, this session will focus on problem formulations, search mechanisms, evolutionary operators, and adaptation techniques for ENAS. Authors are encouraged to submit their original research to this special session.
Scope and Topics
Novel problem formulations for ENAS
Operation/parameter adaptive learning mechanism in ENAS
Search space (decision variable) analysis techniques
Surrogate models and EC interaction methods for ENAS
Distributed EC optimization for ENAS
Large-scale optimization algorithms for ENAS
New evolutionary operators for ENAS
Fast fitness evaluation algorithms in ENAS
Multi-fidelity evaluation methodology for ENAS
Representation strategies for candidate individuals
Real-world applications of ENAS, e.g. image sequences, image analysis, face recognition, natural language processing, named entity recognition, text mining, network security, engineering problems, financial and business data analysis, etc.
Important Dates
15 January 2025:Paper Submission Deadline 15 March 2025: Paper Acceptance Notification 1 May 2025: Final Paper Submission & Early Registration Deadline 8-12 June 2025: Conference Date
Submissions
Organizers
Dr. Yi Jiang, Hanyang University
Dr. Nan Li, Northeastern University Dr. Jian-Yu Li, Nankai University Professor Zhi-Hui Zhan，Nankai University
