This inventor holds 1 USPTO granted patent and 1 published patent application. Top assignee: Electronics and Telecommunications Research Institute. Active years: 2026.
Data source: USPTO (United States Patent and Trademark Office) public patent records. Weekly synchronization. How IDiyas builds this profile
Company Filing History:
Years Active: 2026
Title: Yongsik Lee and Dynamic Deep Neural Network Training.
Introduction
Yongsik Lee is an inventor based in Daejeon, South Korea. His patent work includes a method and system for training a dynamic deep neural network.
Latest Patents
Lee is associated with 1 patent, titled “Method and System for Training Dynamic Deep Neural Network.” The invention describes receiving the output of the network’s last layer to produce a first loss, and receiving the routing module’s output based on the input class to produce a second loss. It then calculates a third loss from the first two losses and uses that third loss to update the network’s weights.
Career Highlights
Lee is associated with the Electronics and Telecommunications Research Institute. His patent reflects work on neural-network training methods that combine prediction and routing losses to update model weights.
Collaborations
Lee’s listed coworker is Seungjae Lee.
Conclusion
Yongsik Lee’s patent focuses on a training approach for dynamic deep neural networks, combining multiple loss signals to guide weight updates.
Data source: USPTO (United States Patent and Trademark Office) public patent records. Weekly synchronization. How IDiyas builds this profile