This inventor holds 1 USPTO granted patent. Top assignee: Google Inc.. Active years: 2025.
Data source: USPTO (United States Patent and Trademark Office) public patent records. Weekly synchronization. How IDiyas builds this profile
Company Filing History:
Years Active: 2025
Title: Felix Stahlberg and Neural Machine Translation
Introduction
Felix Stahlberg is an inventor based in Berlin, Germany, whose patent work addresses ambiguity in neural machine translation. His disclosed invention explores how translation systems can evaluate multiple possible translations of a given text.
Latest Patents
Stahlberg is associated with 1 patent, titled “Modeling Ambiguity in Neural Machine Translation.” The technology uses an encoder to represent an input text exemplar and a decoder to evaluate translation prefixes. A logits module derives bounded conditional probabilities for tokens without normalizing them across the vocabulary at each time step. A loss function combines positive and scaled negative components to determine whether candidate target texts are valid translations of the exemplar.
Career Highlights
Stahlberg works at Google Inc. His patent focuses on representing and assessing ambiguity in neural machine translation, including the evaluation of multiple candidate translations.
Collaborations
Stahlberg’s listed coworker is Shankar Kumar. The available information identifies their professional association but does not specify a particular joint project.
Conclusion
Felix Stahlberg’s patent describes an approach to evaluating alternative translations and addressing ambiguity in neural machine translation.
Data source: USPTO (United States Patent and Trademark Office) public patent records. Weekly synchronization. How IDiyas builds this profile