Matthew David Smith

Title: Innovations by Matthew David Smith

Introduction

Matthew David Smith is an accomplished inventor based in Albuquerque, NM (US). He has made significant contributions to the field of digital image segmentation, holding a total of 3 patents. His work focuses on refining image segmentation techniques, particularly under varying domain conditions.

Latest Patents

One of his latest patents is titled "Uncertainty-refined image segmentation under domain shift." This innovative method involves training a neural network for image segmentation using a labeled training dataset from a first domain. During training, a subset of nodes in the neural network is dropped out, allowing the network to receive image data from a second, different domain. For each image element, a vector of N values that sum to 1 is calculated, with each value representing an image segmentation class. A label is assigned to each image element based on the class with the highest value in the vector. The method also includes performing multiple inferences with active dropout layers for each image element, generating an uncertainty value for each. If an image element's uncertainty exceeds a predefined threshold, its label is replaced with a new label corresponding to the class with the next highest value.

Career Highlights

Matthew David Smith works at National Technology & Engineering Solutions of Sandia, LLC, where he applies his expertise in image processing and neural networks. His innovative approaches have garnered attention in the field, contributing to advancements in technology and research.

Collaborations

Some of his notable coworkers include John P Korbin and Carianne Martinez, who collaborate with him on various projects within the organization.

Conclusion

Matthew David Smith's contributions to digital image segmentation and his innovative patents highlight his role as a leading inventor in his field. His work continues to influence advancements in technology and research methodologies.