Jianwen Xie

Title: Innovations of Jianwen Xie

Introduction

Jianwen Xie is a prominent inventor based in Santa Clara, CA (US). He has made significant contributions to the field of energy-based models, with a focus on image generation and restoration. His innovative work has led to the development of two notable patents.

Latest Patents

Jianwen Xie's latest patents include "Training energy-based models from a single image for internal learning and inference using trained models." This invention differs from prior works by explicitly representing the statistical distribution within a single image using an energy-based generative framework. It employs a pyramid of energy functions, each parameterized by a bottom-up deep neural network, to capture the distributions of patches at various resolutions. The methodology also introduces a coarse-to-fine sequential training and sampling strategy, allowing for efficient model training. Additionally, the model can learn in parallel with a self-supervised task, enhancing its descriptive power without requiring an auxiliary model.

Another significant patent is "Energy-based generative models via coarse-to-fine expanding and sampling." This invention presents a multistage coarse-to-fine expanding and sampling strategy for training energy-based models. The training process begins with a coarse-level EBM from low-resolution images and transitions to a finer-level EBM from higher-resolution images. This approach is computationally efficient and has demonstrated superior performance in image generation, successfully synthesizing high-fidelity images. The framework can also be generalized for unsupervised image-to-image translation.

Career Highlights

Jianwen Xie is currently employed at Baidu USA LLC, where he continues to push the boundaries of innovation in his field. His work has garnered attention for its practical applications in image restoration and out-of-distribution detection.

Collaborations

Jianwen collaborates with notable colleagues, including Ping Li and Zilong Zheng, contributing to a dynamic research environment that fosters innovation.

Conclusion

Jianwen Xie's contributions to energy-based models and image generation highlight his role as a leading inventor in the field. His innovative patents reflect a commitment to advancing technology and improving methodologies in image processing.