Ziyang Wu

Papers

Total Citations

H-Index

About

Ziyang Wu is a researcher whose work critically examines the scaling assumptions in modern computer vision. His most-cited paper, "When Do We Not Need Larger Vision Models?" (2024), challenges the prevailing trend of ever-increasing model sizes by identifying the specific conditions under which smaller, more efficient architectures can match or outperform their larger counterparts. This contribution is particularly impactful for resource-constrained applications and sustainable AI development, earning 15 citations in a short time. Wu’s research focuses on model efficiency, architectural design, and the practical limits of scaling laws, offering a counterpoint to the "bigger is better" paradigm. By systematically analyzing when larger models are unnecessary, he provides actionable insights for deploying vision systems in real-world scenarios. His work is essential reading for students and researchers interested in efficient deep learning, model compression, and the future of computationally accessible AI.