Method, System, and Computer Program Product for Training Distributed Machine Learning Models

    公开(公告)号:US20240211814A1

    公开(公告)日:2024-06-27

    申请号:US18435374

    申请日:2024-02-07

    IPC分类号: G06N20/20 G06N3/045 G06N3/084

    CPC分类号: G06N20/20 G06N3/045 G06N3/084

    摘要: Provided is a method for training distributed machine learning models. The method may include initializing a distributed machine learning model on a plurality of computing devices. Training data associated with a plurality of samples may be received. Each sample may be forward propagated through the distributed machine learning model to generate an output. A loss for each sample of the plurality of samples may be determined based on the output. The loss for each sample may be backward propagated to each computing device. The parameter(s) of each computational node may be asynchronously updated based on the loss as it is backward propagated and/or while at least one of the samples is forward propagating. The parameter(s) may be stored and/or communicated to the other computing devices. Each of the other computing devices of the plurality of computing devices may store the parameter(s). A system and computer program product are also disclosed.

    Techniques for multi-tiered data storage in multi-tenant caching systems

    公开(公告)号:US12019551B2

    公开(公告)日:2024-06-25

    申请号:US17764879

    申请日:2019-10-04

    发明人: Yu Gu Hongqin Song

    摘要: Embodiments of the invention are directed to systems and methods for utilizing a multi-tiered caching architecture in a multi-tenant caching system. A portion of the in-memory cache may be allocated as dedicated shares (e.g., dedicated allocations) that are each dedicated to a particular tenant, while another portion of the in-memory cache (e.g., a shared allocation) can be shared by all tenants in the system. When a threshold period of time has elapsed since data stored in a dedicated allocation has last been accessed, the data may be migrated to the shared allocation. If data is accessed from the shared allocation, it may be migrated back to the dedicated allocation Utilizing the techniques for providing a multi-tiered approach to a multi-tenant caching system can increase performance and decrease latency with respect to conventional caching systems.

    Systems and methods for refreshing token data

    公开(公告)号:US12010119B2

    公开(公告)日:2024-06-11

    申请号:US17276462

    申请日:2018-09-20

    摘要: A method is disclosed. A computer may receive a request message from an access device. The computer may search a supplemental database for updated supplemental data corresponding to a real credential. The supplemental database may store the real credential, supplemental data, and the updated supplemental data. The updated supplemental data may be provided to the supplemental database in connection with authorization processing involving the real credential. The computer may receive the updated supplemental data associated with the real credential. The computer may transmit a response message to the access device including the updated supplemental data. The request message may be, as examples, an authorization request message comprising a token or a token request message comprising a token.