Baichen Liu

About

Baichen Liu is a researcher at the forefront of efficient deep learning, specializing in model compression and neural network optimization for resource-constrained environments. His most notable contribution, "Towards Super Compressed Neural Networks for Object Identification," introduces a pioneering framework that combines quantized low-rank tensor decomposition with self-attention mechanisms. This work directly tackles the critical challenge of deploying deep convolutional neural networks on devices with limited storage and computational power, such as mobile phones. By dramatically reducing both parameter counts and floating-point operations, Liu’s approach enables high-performance object identification without sacrificing accuracy. Though early in its trajectory, the paper has already garnered attention with 2 citations, signaling its potential impact on edge AI and mobile vision applications. Liu’s research bridges the gap between state-of-the-art neural network performance and practical deployment, offering a scalable solution for real-world AI systems. His work is particularly relevant for students and engineers seeking to understand how advanced compression techniques can democratize deep learning across hardware-limited platforms.

Research Focus

Key Achievements

Top Papers

1. 1
   Towards Super Compressed Neural Networks for Object Identification: Quantized Low-Rank Tensor Decomposition with Self-Attention

   2 citations · 2024

Key Collaborators

Contact & Links