FEDERATED LEARNING WITH PARTITIONED AND DYNAMICALLY-SHUFFLED MODEL UPDATES

    公开(公告)号:US20220374763A1

    公开(公告)日:2022-11-24

    申请号:US17323099

    申请日:2021-05-18

    IPC分类号: G06N20/00

    摘要: Techniques for distributed federated learning leverage a multi-layered defense strategy to provide for reduced information leakage. In lieu of aggregating model updates centrally, an aggregation function is decentralized into multiple independent and functionally-equivalent execution entities, each running within its own trusted executed environment (TEE). The TEEs enable confidential and remote-attestable federated aggregation. Preferably, each aggregator entity runs within an encrypted virtual machine that support runtime in-memory encryption. Each party remotely authenticates the TEE before participating in the training. By using multiple decentralized aggregators, parties are enabled to partition their respective model updates at model-parameter granularity, and can map single weights to a specific aggregator entity. Parties also can dynamically shuffle fragmentary model updates at each training iteration to further obfuscate the information dispatched to each aggregator execution entity. This architectural prevents the aggregator from being a single point-of-failure, and serves to protect the model even if all aggregators are compromised.

    Trust/value/risk-based access control policy
    5.
    发明授权
    Trust/value/risk-based access control policy 有权
    信任/价值/基于风险的访问控制策略

    公开(公告)号:US09432375B2

    公开(公告)日:2016-08-30

    申请号:US14281955

    申请日:2014-05-20

    摘要: Generating a resource access control decision is provided. A user trust value associated with a user identifier of a user requesting access to a protected resource is modulated based on an estimated risk value associated with a context of a resource access request. The resource access control decision is generated based on the modulated user trust value associated with the user requesting access to the protected resource.

    摘要翻译: 提供了生成资源访问控制决策。 基于与资源访问请求的上下文相关联的估计风险值来调制与请求访问受保护资源的用户的用户标识符相关联的用户信任值。 基于与请求对受保护资源的访问的用户相关联的调制的用户信任值来生成资源访问控制决定。

    TRUSTED AND DECENTRALIZED AGGREGATION FOR FEDERATED LEARNING

    公开(公告)号:US20220374762A1

    公开(公告)日:2022-11-24

    申请号:US17323006

    申请日:2021-05-18

    IPC分类号: G06N20/00 G06N5/04

    摘要: Techniques for distributed federated learning leverage a multi-layered defense strategy to provide for reduced information leakage. In lieu of aggregating model updates centrally, an aggregation function is decentralized into multiple independent and functionally-equivalent execution entities, each running within its own trusted executed environment (TEE). The TEEs enable confidential and remote-attestable federated aggregation. Preferably, each aggregator entity runs within an encrypted virtual machine that support runtime in-memory encryption. Each party remotely authenticates the TEE before participating in the training. By using multiple decentralized aggregators, parties are enabled to partition their respective model updates at model-parameter granularity, and can map single weights to a specific aggregator entity. Parties also can dynamically shuffle fragmentary model updates at each training iteration to further obfuscate the information dispatched to each aggregator execution entity. This architectural prevents the aggregator from being a single point-of-failure, and serves to protect the model even if all aggregators are compromised.

    TRUST/VALUE/RISK-BASED ACCESS CONTROL POLICY
    9.
    发明申请
    TRUST/VALUE/RISK-BASED ACCESS CONTROL POLICY 有权
    信任/价值/基于风险的访问控制策略

    公开(公告)号:US20150106888A1

    公开(公告)日:2015-04-16

    申请号:US14281955

    申请日:2014-05-20

    IPC分类号: H04L29/06 G06F21/31

    摘要: Generating a resource access control decision is provided. A user trust value associated with a user identifier of a user requesting access to a protected resource is modulated based on an estimated risk value associated with a context of a resource access request. The resource access control decision is generated based on the modulated user trust value associated with the user requesting access to the protected resource.

    摘要翻译: 提供了生成资源访问控制决策。 基于与资源访问请求的上下文相关联的估计风险值来调制与请求访问受保护资源的用户的用户标识符相关联的用户信任值。 基于与请求对受保护资源的访问的用户相关联的调制的用户信任值来生成资源访问控制决定。