James W Ryland

Title: James W. Ryland: Machine Learning and Neural Network Patents.

Introduction

James W. Ryland is an inventor associated with Dallas, Texas. His patent work focuses on machine learning systems and neural network techniques. He holds 2 patents.

Latest Patents

Ryland’s latest patents address adapting machine learning models and obfuscating neural networks.

“Machine learning model grafting and integration” describes a method for combining machine learning models. It retains the backbone of one model and uses error back-propagation through another model’s backbone to generate an inception basis set. A bridge configured with that set translates features from the retained backbone for use by the second model’s head.

“Neural network obfuscation by discretization” describes a method for obfuscating a neural network by mapping neuron inputs from a continuous domain to a discrete domain. The network applies its activation functions to the discretized inputs to produce discretized outputs.

Career Highlights

The supplied career information lists Raytheon Company and Janssen Biotech, Inc. among the companies where Ryland has worked. His patent portfolio reflects work involving machine learning model integration and neural network obfuscation.

Collaborations

Ryland’s listed coworker is Philip A. Sallee. The supplied information does not specify the projects or patent collaborations they shared.

Conclusion

James W. Ryland’s two patents concern methods for integrating machine learning models and discretizing neural network inputs and outputs, highlighting his contributions to these areas of innovation.