Jason Zhi Liang

Title: The Innovative Mind of Jason Zhi Liang

Introduction

Jason Zhi Liang is a prominent inventor based in Fremont, CA, known for his contributions to the field of data augmentation and neural network training. With a total of seven patents to his name, Liang has made significant strides in enhancing machine learning processes.

Latest Patents

One of his latest patents is titled "System and method for evolved data augmentation and selection." This invention describes a process for evolving a data augmentation policy that can be applied to sample data from a dataset for training a neural network. The method involves creating an initial population of candidate data augmentation policy models, which include multiple nodes representing distinct data augmentation operations. These models are evaluated by applying them to sample data and partially training the neural network using the augmented dataset. A fitness score is determined based on the training results, leading to the selection of final policy models or further evolution of the models until an optimal one is chosen.

Another notable patent is "Multiobjective coevolution of deep neural network architectures." This invention introduces an evolutionary AutoML framework called LEAF, which optimizes hyperparameters, network architectures, and network size. LEAF utilizes evolutionary algorithms and distributed computing frameworks to maximize performance while minimizing complexity. The multiobjective evolutionary algorithm calculates the Pareto front for a group of evaluated individuals, balancing multiple objectives effectively.

Career Highlights

Throughout his career, Jason Zhi Liang has worked with Cognizant Technology Solutions U.S. Corporation, where he has applied his innovative ideas to real-world challenges. His experience in the tech industry has allowed him to develop and refine his inventions, contributing to advancements in machine learning and artificial intelligence.

Collaborations

Liang has collaborated with notable professionals in his field, including Risto Miikkulainen and Elliot Meyerson. These partnerships have fostered a creative environment that encourages the exchange of ideas and the development of cutting-edge technologies.

Conclusion

Jason Zhi Liang is a remarkable inventor whose work in data augmentation and neural networks has the potential to revolutionize machine learning. His innovative patents and collaborative efforts highlight his commitment to advancing technology and improving the efficiency of neural network training.