Discrete Graph Diffusion Evolution with Tabu Search for Neural Architecture Generation
International Journal of Neural Systems · 16.11.2026
DOI: 10.1142/s0129065727500341 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Neural Systems, online erschienen am 16.11.2026. Verfasst von Pingyuan Cai und 2 weiteren Autor:innen. Beteiligte Einrichtung: School of Computer Science and School of Software, Nanjing University of Information Science and Technology, Nanjing 210044, P. R. China. Die Arbeit nennt 63 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Jiangsu Distinguished Professor Program, and the Postgraduate Research & Practice Innovation Program of Jiangsu Province.
Abstract (Originalsprache)
Neural Architecture Generation (NAG) seeks to generate high-quality neural architectures more directly than conventional Neural Architecture Search (NAS), but existing diffusion-based approaches commonly rely on pretrained denoisers and do not naturally accommodate constrained discrete architecture spaces. This study proposes Discrete Graph Diffusion Evolution with Tabu Search (DGDETS), a framework with a training-free reverse-generation mechanism that represents architecture decisions as heterogeneous categorical variables and constructs reverse diffusion distributions directly from the current population. Fitness-weighted population members are combined with categorical diffusion compatibility to estimate clean-state distributions, from which discrete reverse transitions generate new architectures without a learned score network or graph denoiser. Constraint-aware projection repairs invalid samples while preserving posterior-supported structural decisions, and late-stage multi-start Tabu Search intensifies promising graph neighborhoods using short-term memory and aspiration. DGDETS is evaluated on NAS-Bench-101, NAS-Bench-201, NAS-Bench-301, and TransNAS-Bench-101, covering heterogeneous architecture representations and downstream tasks. The results show competitive or best-performing architecture quality across the evaluated benchmarks together with low generation time in a controlled same-hardware comparison with directly related diffusion-based NAG baselines. Ablation and sensitivity analyses further demonstrate the importance of fitness-conditioned reverse diffusion, feasibility projection, diversity preservation, and late-stage Tabu refinement. The implementation is available at https://github.com/DK-cpy/DGDETS .
Schlagwörter
Stichwörter der Autor:innen
Autor:innen und Institutionen
- School of Computer Science and School of Software, Nanjing University of Information Science and Technology, Nanjing 210044, P. R. China.
- Department of Mathematics and Information Technology, The Education University of Hong Kong, 10 Lo Ping Road, Tai Po, Hong Kong.
ORCID-Profile: Pingyuan Cai, Yu Xue, Ferrante Neri
Förderung
- National Natural Science Foundation of China
- Jiangsu Distinguished Professor Program, and the Postgraduate Research & Practice Innovation Program of Jiangsu Province
Zitieren
APA
Cai, P., Xue, Y., & Neri, F. (2026). Discrete Graph Diffusion Evolution with Tabu Search for Neural Architecture Generation. International Journal of Neural Systems. https://doi.org/10.1142/s0129065727500341
Vancouver
Cai P, Xue Y, Neri F. Discrete Graph Diffusion Evolution with Tabu Search for Neural Architecture Generation. International Journal of Neural Systems. 2026. doi:10.1142/s0129065727500341
BibTeX
@article{Cai2026,
author = {Cai, Pingyuan and Xue, Yu and Neri, Ferrante},
title = {Discrete Graph Diffusion Evolution with Tabu Search for Neural Architecture Generation},
journal = {International Journal of Neural Systems},
year = {2026},
doi = {10.1142/s0129065727500341}
}
EUVIMED