Jin Yi

Title: Jin Yi - Innovator in Light Backdoor Attack Methods

Introduction

Jin Yi is a notable inventor based in Hangzhou, China. He has made significant contributions to the field of security and machine learning through his innovative patent. His work focuses on developing methods that enhance the covert capabilities of backdoor attacks in deep neural networks (DNNs).

Latest Patents

Jin Yi holds a patent for a "Light backdoor attack method in physical world." This method involves generating a light trigger on a traffic sign using a flashlight. It captures and acquires a backdoor training set through a camera, marking all images in the training set as a target label. The process includes constructing a training set from both the backdoor and clean training sets. Jin then trains a backdoor model using this training set to obtain a trained backdoor model. The effectiveness of the light backdoor attack is evaluated by applying the trained model to a testing set. This innovative approach can poison DNNs, achieving a more covert backdoor attack and providing new ideas for the physical backdoor attack field. Jin Yi has 1 patent to his name.

Career Highlights

Throughout his career, Jin Yi has worked at prominent institutions such as Hangzhou City University and Zhejiang Gongshang University. His experience in these academic environments has allowed him to develop and refine his innovative ideas in the realm of security technology.

Collaborations

Jin Yi has collaborated with notable individuals in his field, including Song Han and Siqi Ren. These partnerships have contributed to the advancement of his research and the successful development of his patented methods.

Conclusion

Jin Yi is a pioneering inventor whose work in light backdoor attack methods has the potential to reshape the landscape of security in machine learning. His innovative approach and dedication to research continue to inspire advancements in this critical field.