| CPC G06N 3/105 (2013.01) [G06N 3/045 (2023.01); G06N 3/08 (2013.01)] | 17 Claims |
|
1. A computer system comprising:
a processor operatively coupled to a memory;
an artificial intelligence (AI) platform, in communication with the processor, having tools configured to build a federated learning framework, comprising:
a training manager configured to train the primary machine learning model (MLM), including to capture contributing model updates to train the primary MLM, the contributing model updates transmitted across at least one communication channel;
the hierarchy manager configured to create a secondary MLM logically positioned in a secondary layer of the hierarchy, the secondary MLM operatively coupled to the primary MLM across the at least one communication channel;
the training manager configured to initialize the created secondary MLM, including clone weights and framework of the primary MLM into the secondary MLM, and populate the created secondary MLM with secondary data, the populated data comprising model updates local to the created secondary MLM;
a MLM manager configured to logically position the secondary MLM local to the secondary layer, and limit access to the secondary MLM within the secondary layer;
the MLM manager configured to store primary MLM data globally, wherein the primary MLM 1s accessible to the secondary MLM;
the MLM manager further configured to manage the federated learning framework, including selectively apply a clustering technique to the MLMs based on one or more attribute of the federated learning framework and to manage membership of one or more of the MLMs within one or more formed clusters of MLMs;
the MLM manager further configured to synchronize the secondary MLM with the primary MLM, including to aggregate weight parameters under a secondary MLM setting, and to update the primary MLM with aggregated local neural network model weights included in the captured contributing data from the secondary MLM;
the training manager is supported by a first API that provides functional support to dynamically optimize, orchestrate, and update the models of the federated learning framework;
the hierarchy manager is supported by a different second API that provides functional support to collect and collate activity data across two or more domains;
the MLM manager is supported by a different third API that provides function support for MLMs corresponding to collecting and collating activity data; and
where each of the first API, the second API, and the third API are operatively coupled together to an API orchestrator to transparently thread together the separate APIs.
|