Xianmin Wang

About

Xianmin Wang is a researcher focused on advancing the efficiency and scalability of deep learning models, particularly through model compression and optimization techniques. Their most-cited work, "Re-training and parameter sharing with the Hash trick for compressing convolutional neural networks" (2020), introduces a novel approach that combines re-training strategies with parameter sharing via the Hash trick to significantly reduce the memory footprint of convolutional neural networks without sacrificing accuracy. This contribution addresses a critical bottleneck in deploying deep learning models on resource-constrained devices, such as mobile and edge platforms. With 5 citations, this paper has already garnered attention from peers working on model efficiency, highlighting its practical relevance. Wang’s research sits at the intersection of machine learning systems and applied AI, offering tangible solutions for real-world deployment challenges. Their work is particularly valuable for students and researchers exploring compression techniques, as it demonstrates how algorithmic innovations—like the Hash trick—can bridge the gap between theoretical model performance and practical hardware limitations.

Research Focus

Key Achievements

Top Papers

1. 1
   Re-training and parameter sharing with the Hash trick for compressing convolutional neural networks

   5 citations · 2020

Key Collaborators

Contact & Links