---
title: "Info-Driven Zero-Cost Proxy: Rethinking Vision Transformer Architecture Evaluation via Information Quantification"
authors:
  - "Yue Yang"
  - "Jiacheng Wang"
  - "Zhenkai Yang"
  - "Menglan Hu"
  - "Gaoyang Liu"
  - "Bo Xu"
  - "Tianyue Zheng"
  - "Kai Peng"
doi: "10.24963/ijcai.2026/581"
url: "https://ijcai.org/proceedings/2026/581"
pdf: "https://ijcai.org/proceedings/2026/0581.pdf"
venue: "Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence"
year: 2026
pages: "5218-5226"
track: "Main Track"
keywords:
  - "Machine Learning: Automated machine learning"
---

# Info-Driven Zero-Cost Proxy: Rethinking Vision Transformer Architecture Evaluation via Information Quantification

Yue Yang, Jiacheng Wang, Zhenkai Yang, Menglan Hu, Gaoyang Liu, Bo Xu, Tianyue Zheng, Kai Peng

*Proceedings of the Thirty-Fifth International Joint Conference on Artificial Intelligence*, pages 5218-5226. DOI: [10.24963/ijcai.2026/581](https://doi.org/10.24963/ijcai.2026/581)

## Abstract

Neural Architecture Search (NAS) automates the design of Vision Transformer (ViT) architectures. However, the high computational cost of training-based methods has made training-free, zero-cost proxies a key research direction. While existing proxies can estimate model potential, they fail to capture critical information transmission characteristics due to the black-box nature of neural networks. Consequently, they suffer from high computational latency and weak correlation with ground-truth performance. Although information transmission determines performance, it has not been effectively quantified or utilized. This limitation remains the core bottleneck for current zero-cost proxies. In this paper, we propose Info-NAS, a zero-cost proxy based on architectural information. This method achieves the first structured quantification of information transmission in ViT. Specifically, it evaluates architectures using three core components: global information volume, local information gradient, and global consistency. Info-NAS derives proxy scores solely from architectural parameters, requiring no forward or backward propagation. Consequently, the calculation and search processes are highly efficient. Moreover, we construct the ViT-Info-Bench dataset to facilitate correlation analysis and algorithm evaluation. Experimental results demonstrate that Info-NAS significantly reduces search overhead while achieving superior ranking accuracy. With a single evaluation requiring only 1.8 ms, Info-NAS outperforms existing methods in both efficiency and performance.