Patents by Inventor Michael Andrew Terry

Michael Andrew Terry has filed for patents to protect the following inventions. This listing includes patent applications that are pending as well as patents that have already been granted by the United States Patent and Trademark Office (USPTO).

  • Patent number: 12236326
    Abstract: Methods, systems, devices, and tangible non-transitory computer readable media for saliency visualization are provided. The disclosed technology can include receiving a data input including a plurality of features. The data input can be segmented into regions. At least one of the regions can include two or more of the features. Attribution scores can be respectively generated for features of the data input. The attribution scores for each feature can be indicative of a respective saliency of such feature. A respective gain value for each region can be determined over one or more iterations based on the respective attribution scores associated with the features included in the region. Further, at each iteration one or more of the regions with the greatest gain values can be added to a saliency mask. Furthermore, at each iteration a saliency visualization can be produced based on the saliency mask.
    Type: Grant
    Filed: August 1, 2023
    Date of Patent: February 25, 2025
    Assignee: GOOGLE LLC
    Inventors: Andrei Kapishnikov, Fernanda Bertini Viégas, Michael Andrew Terry, Tolga Bolukbasi
  • Publication number: 20250036376
    Abstract: The present disclosure provides to transparent and controllable human-AI interaction via chaining of machine-learned language models. In particular, although existing language models (e.g., so-called “large language models” (LLMs)) have demonstrated impressive potential on simple tasks, their breadth of scope, lack of transparency, and insufficient controllability can make them less effective when assisting humans on more complex tasks. In response, the present disclosure introduces the concept of chaining instantiations of machine-learned language models (e.g., LLMs) together, where the output of one instantiation becomes the input for the next, and so on, thus aggregating the gains per step.
    Type: Application
    Filed: October 14, 2024
    Publication date: January 30, 2025
    Inventors: Carrie Cai, Tongshuang Wu, Michael Andrew Terry
  • Patent number: 12141556
    Abstract: The present disclosure provides to transparent and controllable human-AI interaction via chaining of machine-learned language models. In particular, although existing language models (e.g., so-called “large language models” (LLMs)) have demonstrated impressive potential on simple tasks, their breadth of scope, lack of transparency, and insufficient controllability can make them less effective when assisting humans on more complex tasks. In response, the present disclosure introduces the concept of chaining instantiations of machine-learned language models (e.g., LLMs) together, where the output of one instantiation becomes the input for the next, and so on, thus aggregating the gains per step.
    Type: Grant
    Filed: September 30, 2022
    Date of Patent: November 12, 2024
    Assignee: Google LLC
    Inventors: Carrie Cai, Tongshuang Wu, Michael Andrew Terry
  • Publication number: 20240311652
    Abstract: Systems and methods for prompt generation for generative models can include utilizing a specialized markup language. A markup language transform can be utilized to augment user input data to generate a prompt that includes structure and/or wording that facilitates the generation of a generative output that reflects a user's intent. The systems and methods can leverage the specialized markup language and/or an integrated development environment interface to inform a user of the prompt parts and provide editing options.
    Type: Application
    Filed: March 14, 2023
    Publication date: September 19, 2024
    Inventors: Chinmay Kulkarni, Alexander John Fiannaca, Michael Andrew Terry
  • Publication number: 20240054402
    Abstract: Methods, systems, devices, and tangible non-transitory computer readable media for saliency visualization are provided. The disclosed technology can include receiving a data input including a plurality of features. The data input can be segmented into regions. At least one of the regions can include two or more of the features. Attribution scores can be respectively generated for features of the data input. The attribution scores for each feature can be indicative of a respective saliency of such feature. A respective gain value for each region can be determined over one or more iterations based on the respective attribution scores associated with the features included in the region. Further, at each iteration one or more of the regions with the greatest gain values can be added to a saliency mask. Furthermore, at each iteration a saliency visualization can be produced based on the saliency mask.
    Type: Application
    Filed: August 1, 2023
    Publication date: February 15, 2024
    Inventors: Andrei Kapishnikov, Fernanda Bertini Viégas, Michael Andrew Terry, Tolga Bolukbasi
  • Patent number: 11755948
    Abstract: Methods, systems, devices, and tangible non-transitory computer readable media for saliency visualization are provided. The disclosed technology can include receiving a data input including a plurality of features. The data input can be segmented into regions. At least one of the regions can include two or more of the features. Attribution scores can be respectively generated for features of the data input. The attribution scores for each feature can be indicative of a respective saliency of such feature. A respective gain value for each region can be determined over one or more iterations based on the respective attribution scores associated with the features included in the region. Further, at each iteration one or more of the regions with the greatest gain values can be added to a saliency mask. Furthermore, at each iteration a saliency visualization can be produced based on the saliency mask.
    Type: Grant
    Filed: December 18, 2019
    Date of Patent: September 12, 2023
    Assignee: GOOGLE LLC
    Inventors: Andrei Kapishnikov, Tolga Bolukbasi, Fernanda Bertini Viégas, Michael Andrew Terry
  • Publication number: 20230112921
    Abstract: The present disclosure provides to transparent and controllable human-AI interaction via chaining of machine-learned language models. In particular, although existing language models (e.g., so-called “large language models” (LLMs)) have demonstrated impressive potential on simple tasks, their breadth of scope, lack of transparency, and insufficient controllability can make them less effective when assisting humans on more complex tasks. In response, the present disclosure introduces the concept of chaining instantiations of machine-learned language models (e.g., LLMs) together, where the output of one instantiation becomes the input for the next, and so on, thus aggregating the gains per step.
    Type: Application
    Filed: September 30, 2022
    Publication date: April 13, 2023
    Inventors: Carrie Cai, Tongshuang Wu, Michael Andrew Terry
  • Publication number: 20210192382
    Abstract: Methods, systems, devices, and tangible non-transitory computer readable media for saliency visualization are provided. The disclosed technology can include receiving a data input including a plurality of features. The data input can be segmented into regions. At least one of the regions can include two or more of the features. Attribution scores can be respectively generated for features of the data input. The attribution scores for each feature can be indicative of a respective saliency of such feature. A respective gain value for each region can be determined over one or more iterations based on the respective attribution scores associated with the features included in the region. Further, at each iteration one or more of the regions with the greatest gain values can be added to a saliency mask. Furthermore, at each iteration a saliency visualization can be produced based on the saliency mask.
    Type: Application
    Filed: December 18, 2019
    Publication date: June 24, 2021
    Inventors: Andrei Kapishnikov, Tolga Bolukbasi, Fernanda Bertini Viégas, Michael Andrew Terry