Mulin Xin

About

Mulin Xin is a researcher specializing in efficient deep learning, with a focus on model compression and optimization for convolutional neural networks (CNNs). Their most notable contribution is the development of a novel approach that combines re-training with parameter sharing via the Hash trick, a technique that significantly reduces model size while preserving accuracy. This work, published in 2020 and garnering 5 citations, addresses a critical challenge in deploying deep neural networks on resource-constrained devices, such as mobile phones and embedded systems. By leveraging the Hash trick to map multiple weights to a single parameter, Xin’s method enables efficient memory usage without sacrificing performance, offering a practical solution for real-world AI applications. This contribution has implications for advancing edge computing and on-device intelligence, making deep learning more accessible and scalable. Xin’s research bridges the gap between theoretical model compression and practical deployment, underscoring their role in pushing the boundaries of efficient AI systems. Their work continues to inspire further exploration into lightweight neural architectures and parameter-sharing strategies.

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