Method for efficient distributed machine learning hyperparameter search
A method of a hyperparameter server improves hyper-parameter search efficiency for devices in a self-organizing network (SON) includes sending configuration for data feature collection to at least one edge device in the self-organizing network, receiving hyper-parameter performance data from the at least one edge device, and training a shared hyperparameter machine learning model using a global training database including the hyperparameter performance data to identify optimal hyperparameters for use by the at least one edge device. A further method of an edge device improves hyperparameter search efficiency for devices in a SON includes receiving configuration for data feature collection from a hyperparameter server, training an edge machine learning model using local training data and selected hyperparameters, and sending performance data to the hyperparameter server obtained from the training of the edge machine learning model.
1 . A method of a hyperparameter server to improve hyperparameter search efficiency for devices in a self-organizing network associated with a communication network, where hyperparameters have values that are set prior to training a machine learning model and that affect the training of the machine learning model, and where the self-organizing network is a network that can configure, optimize and heal itself through automated processes, the method comprising:
sending configuration for data feature collection to a plurality of edge devices in the self-organizing network, wherein the plurality of edge devices are grouped into a cluster of edge devices comprised of base stations having identified commonalities among the base stations;
receiving hyperparameter performance data from the edge devices that are based training a machine learning model at the edge devices based on the configuration; and
training a shared hyperparameter machine learning model using a global training database including the hyperparameter performance data to identify optimal hyperparameters for use by at least one edge device of the cluster.
2 . The method of claim 1 , further comprising:
adding the hyperparameter performance data to the global training database.
3 . The method of claim 1 , further comprising:
servicing a hyperparameter query from the at least one edge device using the shared hyperparameter machine learning model after updating by the training.
4 . The method of claim 1 , wherein the hyperparameter performance data includes a tuple of probability distribution, hyperparameter settings, and performance metrics.
5 . The method of claim 1 , further comprising:
establishing the cluster with a shared hyperparameter machine learning model and training database for the cluster.
6 . A method of an edge device of a communication network to improve efficiency for a hyperparameter search for devices in a self-organizing network associated with the communication network, where the hyperparameter search is for hyperparameters for a machine learning model for the self-organizing network, where hyperparameters have values that are set prior to training the machine learning model and that affect the training of the machine learning model, and where the self-organizing network is a network that can configure, optimize and heal itself through automated processes using the machine learning model the method comprising:
receiving configuration for data feature collection from a hyperparameter server, wherein the configuration for data feature collection is sent to a plurality of edge devices in the self-organizing network, wherein the plurality of edge devices are grouped into a cluster of edge devices comprised of base stations having identified commonalities among the base stations and wherein the edge device comprises a base station within the cluster;
training an edge machine learning model using local training data and selected hyperparameters from the configuration; and
sending hyperparameter performance data to the hyperparameter server obtained from the training of the edge machine learning model for the hyperparameter server to train a shared hyperparameter machine learning model using a global training database including the hyperparameter performance data to identify optimal hyperparameters for use by the edge device.
7 . The method of claim 6 , further comprising:
collecting data features according to the configuration.
8 . The method of claim 6 , further comprising:
sending a hyperparameter query to the hyperparameter server to obtain optimized hyperparameters from a shared hyperparameter machine learning model.
9 . The method of claim 6 , wherein the hyperparameter performance data includes a tuple of probability distribution, hyperparameter settings, and performance metrics.
10 . A non-transitory machine-readable storage medium that provides instructions that, if executed by a processor, will cause said processor to perform a set of operations for hyperparameter determination for a machine learning model for a self-organizing network associated with a communication network, where hyperparameters have values that are set prior to training the machine learning model and that affect the training of the machine learning model, and where the self-organizing network is a network that can configure, optimize and heal itself through automated processes using the machine learning model, the set of operations comprising:
sending configuration for data feature collection to a plurality of edge devices in the self-organizing network, wherein the plurality of edge devices are grouped into a cluster of edge devices comprised of base stations having identified commonalities among the base stations;
receiving hyperparameter performance data from the edge devices that are based training a machine learning model at the edge devices based on the configuration; and
training a shared hyperparameter machine learning model using a global training database including the hyperparameter performance data to identify optimal hyperparameters for use by at least one edge device of the cluster.
11 . The non-transitory machine-readable storage medium of claim 10 , the set of operations further comprising:
adding the hyperparameter performance data to the global training database.
12 . The non-transitory machine-readable storage medium of claim 10 , the set of operations further comprising:
servicing a hyperparameter query from the at least one edge device using the shared hyperparameter machine learning model after updating by the training.
13 . The non-transitory machine-readable storage medium of claim 10 , wherein the hyperparameter performance data includes a tuple of probability distribution, hyperparameter settings, and performance metrics.
14 . The non-transitory machine-readable storage medium of claim 10 , the set of operations further comprising:
establishing the cluster with a shared hyperparameter machine learning model and training database for the cluster.
15 . A non-transitory machine-readable storage medium that provides instructions that, if executed by a processor associated with an edge device, will cause said processor to perform a set of operations for hyperparameter determination for a machine learning model for a self-organizing network associated with a communication network, where hyperparameters have values that are set prior to training the machine learning model and that affect the training of the machine learning model, and where the self-organizing network is a network that can configure, optimize and heal itself through automated processes using the machine learning model, the set of operations comprising:
receiving configuration for data feature collection from a hyperparameter server, wherein the configuration for data feature collection is sent to a plurality of edge devices in the self-organizing network, wherein the plurality of edge devices are grouped into a cluster of edge devices comprised of base stations having identified commonalities among the base stations and wherein the edge device comprises a base station within the cluster;
training an edge machine learning model using local training data and selected hyperparameters from the configuration; and
sending hyperparameter performance data to the hyperparameter server obtained from the training of the edge machine learning model for the hyperparameter server to train a shared hyperparameter machine learning model using a global training database including the hyperparameter performance data to identify optimal hyperparameters for use by the edge device.
16 . The non-transitory machine-readable storage medium of claim 15 , the set of operations further comprising:
collecting data features according to the configuration.
17 . The non-transitory machine-readable storage medium of claim 15 , the set of operations further comprising:
sending a hyperparameter query to the hyperparameter server to obtain optimized hyperparameters from a shared hyperparameter machine learning model.
18 . The non-transitory machine-readable storage medium of claim 15 , wherein the hyperparameter performance data includes a tuple of probability distribution, hyperparameter settings, and performance metrics.