Tao Sheng

Title: Tao Sheng - Innovator in Synthetic Data Generation

Introduction

Tao Sheng is a prominent inventor based in Bellevue, WA (US). He has made significant contributions to the field of synthetic data generation, holding a total of 3 patents. His work focuses on developing innovative techniques that enhance machine learning models, particularly in the extraction of key-value pairs from document images.

Latest Patents

Tao Sheng's latest patents include a "Synthetic table generation pipeline for training deep table extraction models." This patent describes techniques for HTML-based image generation, which involves generating hypertext markup language (HTML) code for a table structure consisting of rows and columns. The method also includes generating HTML code for text to populate table cells and creating rendered images of the table. Additionally, it encompasses detecting pixels of different colors and generating bounding boxes around detected text.

Another notable patent is "Generating synthetic training data including document images with key-value pairs." This invention automates the generation of diverse training data for machine learning models. By using a single input document image and associated annotation data, a synthetic data generation system can create numerous synthetic training datapoints. Each datapoint includes a synthetic document image and its corresponding annotation data, which can significantly improve the performance of machine learning models in extracting key-value pairs.

Career Highlights

Tao Sheng is currently employed at Oracle International Corporation, where he continues to innovate and contribute to advancements in technology. His work is instrumental in developing solutions that leverage synthetic data for training machine learning models.

Collaborations

Tao has collaborated with notable coworkers, including Jun Qian and Yazhe Hu, who share his passion for innovation and technology.

Conclusion

Tao Sheng is a distinguished inventor whose work in synthetic data generation is paving the way for advancements in machine learning. His contributions are vital for improving the efficiency and accuracy of data extraction processes.