Pengyu He

Title: Pengyu He: Inventor in Machine Learning

Introduction

Pengyu He is an inventor based in Jersey City, New Jersey, whose patent work concerns machine-learning model architecture. His listed patent describes methods for representing training data and adjusting model parameters.

Latest Patents

The available information lists one patent for Pengyu He: “Boosting and Matrix Factorization.” The invention describes generating matrices from a training dataset containing feature and output variables, including sparse and encoded representations of features. It also describes representing each feature with a vector of adjustable parameters and training the model by adjusting those parameters using a loss function and the generated matrices. The listed patent count is 1 patent.

Career Highlights

Pengyu He is associated with Google Inc. His listed invention focuses on combining boosting and matrix factorization techniques in a machine-learning model architecture.

Collaborations

The provided information associates Pengyu He’s work with Google Inc. It does not specify individual co-inventors or other collaborators.

Conclusion

Pengyu He’s listed patent presents a machine-learning approach that uses matrix representations and adjustable parameters to train a model, reflecting his documented contribution to innovation at Google Inc.