Yanming Guo

Title: Yanming Guo: Innovator in Neural Network Training Methods

Introduction

Yanming Guo is a prominent inventor based in Hunan, China. He has made significant contributions to the field of artificial intelligence, particularly in the development of robust neural network training methods. His innovative approach addresses critical challenges in machine learning and enhances the performance of neural networks.

Latest Patents

Yanming Guo holds a patent for a "Training method for robust neural network based on feature matching." This patent outlines a comprehensive training method that includes several key steps. The process begins with the initialization of a first stage model, which consists of a backbone network, a feature matching module, and a fullple loss function. The first stage model is then trained using original training data to develop a second stage model. Following this, the second stage model is subjected to attacks to generate PGD adversarial samples from the original training data. The model is retrained with both the generated adversarial samples and the original training data. Finally, training parameters are adjusted, and the model is trained again, saving the parameters that yield the highest accuracy on an original test set. This innovative method enhances the robustness of neural networks against adversarial attacks.

Career Highlights

Yanming Guo is affiliated with the National University of Defense Technology, where he contributes to research and development in advanced technologies. His work focuses on improving the reliability and efficiency of neural networks, making significant strides in the field of artificial intelligence.

Collaborations

He collaborates with notable colleagues, including Jian Li and Songyang Lao, who share his commitment to advancing technology and innovation in neural networks.

Conclusion

Yanming Guo's contributions to the field of neural networks exemplify the importance of innovation in technology. His patented training method represents a significant advancement in creating more robust artificial intelligence systems.