Shikun Liu

Title: Innovations of Shikun Liu in Neural Network Architecture

Introduction

Shikun Liu is an accomplished inventor based in London, GB. He has made significant contributions to the field of neural networks, holding a total of 3 patents. His work focuses on enhancing the efficiency and accuracy of neural network architectures.

Latest Patents

One of Shikun Liu's latest patents is titled "Generating and utilizing pruned neural networks." This patent describes a system that automatically and progressively prunes neural networks, reducing their size without compromising accuracy. The system can jointly train portions of a neural network while progressively eliminating redundant subsets at each training iteration. This innovative approach not only increases the accuracy of the neural network but also enhances its efficiency by removing excess portions.

Another notable patent is "Shaping a neural network architecture utilizing learnable sampling layers." This invention relates to shaping the architecture of a neural network by providing a mechanism for at least one sampling layer. The system includes a learnable scaling factor that adjusts the sampling rate based on a dataset while simultaneously learning the network weights. This allows for a more optimized architecture by modifying the sampling rate of the sampling layer.

Career Highlights

Shikun Liu is currently employed at Adobe, Inc., where he continues to push the boundaries of neural network technology. His work has garnered attention for its innovative approaches to improving neural network performance.

Collaborations

Shikun Liu collaborates with talented individuals such as Zhe Lin and Yilin Wang, who contribute to the advancement of their projects.

Conclusion

Shikun Liu's contributions to neural network architecture through his patents demonstrate his commitment to innovation in technology. His work not only enhances the efficiency of neural networks but also sets a foundation for future advancements in the field.