18243782. UNSUPERVISED METHOD TO GENERATE ANNOTATIONS FOR NATURAL LANGUAGE UNDERSTANDING TASKS simplified abstract (Microsoft Technology Licensing, LLC)

From WikiPatents
Jump to navigation Jump to search

UNSUPERVISED METHOD TO GENERATE ANNOTATIONS FOR NATURAL LANGUAGE UNDERSTANDING TASKS

Organization Name

Microsoft Technology Licensing, LLC

Inventor(s)

Hany Mohamed Hassan Awadalla of Sammamish WA (US)

Subhabrata Mukherjee of Seattle WA (US)

Ahmed Awadallah of Redmond WA (US)

UNSUPERVISED METHOD TO GENERATE ANNOTATIONS FOR NATURAL LANGUAGE UNDERSTANDING TASKS - A simplified explanation of the abstract

This abstract first appeared for US patent application 18243782 titled 'UNSUPERVISED METHOD TO GENERATE ANNOTATIONS FOR NATURAL LANGUAGE UNDERSTANDING TASKS

Simplified Explanation

The abstract describes a method for training a machine learning model using parallel annotations of source instances while ensuring the security of the source instances. This is achieved by generating a coupled machine learning model from two existing models: one trained on unannotated natural language and the other trained on populated target templates. The coupled model can then transform unannotated natural language into annotated machine-readable text.

  • The method involves training a machine learning model with parallel annotations of source instances.
  • The source instances are kept secure during the training process.
  • A coupled machine learning model is generated by combining two existing models.
  • One model is trained on unannotated natural language.
  • The other model is trained on populated target templates.
  • The coupled model can transform unannotated natural language into annotated machine-readable text.

Potential Applications

  • Natural language processing and understanding
  • Machine translation
  • Text summarization
  • Sentiment analysis
  • Chatbots and virtual assistants

Problems Solved

  • Ensuring the security of source instances during machine learning model training
  • Improving the accuracy and efficiency of natural language processing tasks
  • Enabling the transformation of unannotated natural language into annotated machine-readable text

Benefits

  • Enhanced security and privacy of source instances
  • Improved accuracy and quality of machine learning models
  • Faster and more efficient processing of natural language data
  • Enables the development of advanced natural language processing applications


Original Abstract Submitted

A method for training a machine learning model with parallel annotations of source instances and while facilitating security of the source instances can be performed by a system that generates a coupled machine learning model from (1) a first machine learning model trained on a first set of training data comprising unannotated natural language and (2) a second machine learning model trained on populated target templates which are populated with a plurality of vocabulary words. Once formed, the coupled machine learning model is configured to transform unannotated natural language into annotated machine-readable text.

Retrieved from "http://wikipatents.org/index.php?title=18243782._UNSUPERVISED_METHOD_TO_GENERATE_ANNOTATIONS_FOR_NATURAL_LANGUAGE_UNDERSTANDING_TASKS_simplified_abstract_(Microsoft_Technology_Licensing,_LLC)&oldid=17073"