Lingfang Zeng

Title: Innovations of Lingfang Zeng in Neural Network Technology

Introduction

Lingfang Zeng is a prominent inventor based in Hangzhou, China. He has made significant contributions to the field of neural networks, particularly in the execution of computational graphs. With a total of two patents to his name, Zeng's work is paving the way for advancements in deep learning technologies.

Latest Patents

Zeng's latest patents include a "Method for execution of computational graph in neural network model and apparatus thereof." This patent discloses a method for executing a computational graph in a neural network model. It involves creating task execution bodies on a native machine based on a physical computational graph compiled by a deep learning framework. The method also includes a solution for allocating idle memory blocks to each task execution body, enabling the entire computational graph to participate in deep learning training tasks in a pipelining and parallelizing manner.

Another notable patent is the "Intermediate representation method and apparatus for parallel execution of graph computation." This patent outlines a method that compiles a neural network into a computational graph on a computer. It defines branch states and data dependency relationships of tensor variables within the computational graph. The method also transforms control dependencies into data dependencies, allowing for parallel execution of branch threads in the global computational graph, thereby optimizing compilation efficiency.

Career Highlights

Lingfang Zeng is currently associated with Zhejiang Lab, where he continues to innovate in the field of neural networks. His work is instrumental in enhancing the efficiency and effectiveness of deep learning applications.

Collaborations

Zeng collaborates with notable colleagues such as Hongsheng Wang and Guang Chen, contributing to a dynamic research environment that fosters innovation.

Conclusion

Lingfang Zeng's contributions to neural network technology through his patents demonstrate his commitment to advancing the field. His innovative methods for executing computational graphs are set to have a lasting impact on deep learning applications.