知搜-全球专利检索

  1. 登录
  2. 注册

搜索结果

2946 条记录 (耗时 0.036 秒)

    RELEVANT CONTEXT DETERMINATION
    14.
    发明公开
    RELEVANT CONTEXT DETERMINATION 审中-公开

    公开(公告)号：US20240331686A1

    公开(公告)日：2024-10-03

    申请号：US18739466

    申请日：2024-06-11

    申请人： Amazon Technologies, Inc.

    发明人： Kai Wei ,  Thanh Dac Tran ,  Grant Strimel

    IPC分类号： G10L15/18 ,  G06N3/08 ,  G10L15/06 ,  G10L15/16 ,  G10L15/22 ,  G10L15/28

    CPC分类号： G10L15/1815 ,  G06N3/08 ,  G10L15/063 ,  G10L15/16 ,  G10L15/22 ,  G10L15/28 ,  G10L2015/228

    摘要： Techniques for determining and storing relevant context information for a user input, such as a spoken input, are described. In some embodiments, context information is determined to be relevant on an audio frame basis. Context scores for different types of context data (e.g., prior dialog turn data, user profile data, device information, etc.) are determined for individual audio frames corresponding to a spoken input. Based on the corresponding context scores, the most relevant context is stored in a local context cache. The local context cache is updated as subsequent audio frames, of the user input, are processed. The data stored in the context cache is provided to downstream components to perform tasks such as ASR, NLU and SLU.

    METHOD AND APPARATUS FOR GENERATING DATA TO TRAIN MODELS FOR ENTITY RECOGNITION FROM CONVERSATIONS
    17.
    发明公开
    METHOD AND APPARATUS FOR GENERATING DATA TO TRAIN MODELS FOR ENTITY RECOGNITION FROM CONVERSATIONS 审中-公开

    公开(公告)号：US20240296836A1

    公开(公告)日：2024-09-05

    申请号：US18116302

    申请日：2023-03-02

    申请人： Uniphore Technologies, Inc.

    发明人： Gideon Hollander

    IPC分类号： G10L15/08 ,  G10L15/06

    CPC分类号： G10L15/083 ,  G10L15/063 ,  G10L2015/0631 ,  G10L2015/0638

    摘要： In a method and apparatus for generating training data to train models for entity recognition from conversations, the method includes identifying a first text from a first data element on a first graphical user interface (GUI), on which a first action is performed by a first agent, the first data element corresponding to an entity type, wherein the first action comprises at least one of typing, clicking, highlighting, hovering or reading, matching the first text to a first transcribed text within a transcription of a first conversation between the first agent and a first customer, where the first transcribed text corresponds to a time proximate to the time the first action, and determining at least a portion of the first transcribed text as an automatically generated training data (AGTD) for the entity.

    UNSUPERVISED FEDERATED LEARNING OF MACHINE LEARNING MODEL LAYERS
    18.
    发明公开
    UNSUPERVISED FEDERATED LEARNING OF MACHINE LEARNING MODEL LAYERS 审中-公开

    公开(公告)号：US20240296834A1

    公开(公告)日：2024-09-05

    申请号：US18659940

    申请日：2024-05-09

    申请人： GOOGLE LLC

    发明人： Françoise Beaufays ,  Khe Chai Sim ,  Johan Schalkwyk

    IPC分类号： G10L15/06 ,  G10L15/187 ,  G10L15/22 ,  G10L15/30

    CPC分类号： G10L15/063 ,  G10L15/187 ,  G10L15/22 ,  G10L15/30 ,  G10L2015/0635

    摘要： Implementations disclosed herein are directed to unsupervised federated training of global machine learning (“ML”) model layers that, after the federated training, can be combined with additional layer(s), thereby resulting in a combined ML model. Processor(s) can: detect audio data that captures a spoken utterance of a user of a client device; process, using a local ML model, the audio data to generate predicted output(s); generate, using unsupervised learning locally at the client device, a gradient based on the predicted output(s); transmit the gradient to a remote system; update weight(s) of the global ML model layers based on the gradient; subsequent to updating the weight(s), train, using supervised learning remotely at the remote system, a combined ML model that includes the updated global ML model layers and additional layer(s); transmit the combined ML model to the client device; and use the combined ML model to make prediction(s) at the client device.