DISTRIBUTED MODEL EXECUTION
Distributed model execution, including: identifying, for each model of a plurality of models, based on one or more execution constraints for the plurality of models, a corresponding node of a plurality of nodes, wherein the plurality of nodes each comprise one or more computing devices or one or more virtual machines; deploying each model of the plurality of models to the identified corresponding node of the plurality of nodes; and wherein the plurality of models are configured to generate, based on data input to at least one model of the plurality of models, a prediction associated with the data.
1 . A method for distributed model execution, comprising:
identifying, for each model of a plurality of models, based on one or more execution constraints for the plurality of models, a corresponding node of a plurality of nodes, wherein the plurality of nodes each comprise one or more computing devices or one or more virtual machines;
deploying each model of the plurality of models to the corresponding node of the plurality of nodes; and
wherein the plurality of models are configured to generate, based on data input to at least one model of the plurality of models, a prediction associated with the data.
2 . The method of claim 1 , wherein the one or more execution constraints comprise one or more model dependencies, one or more encryption constraints, or one or more authorization constraints.
3 . The method of claim 1 , further comprising generating the prediction based on a distributed execution of the plurality of models.
4 . The method of claim 1 , wherein identifying the corresponding node of the plurality of nodes is based on one or more node characteristics of the plurality of nodes.
5 . The method of claim 4 , wherein the one or more node characteristics comprise one or more of: one or more hardware resources of one or more of the plurality of nodes, or one or more software resources of one or more of the plurality of nodes.
6 . The method of claim 1 , wherein identifying the corresponding node of the plurality of nodes is based on one or more model characteristics of the plurality of models.
7 . The method of claim 6 , wherein the one or more model characteristics comprise one or more of: a data type for input data to one or more of the plurality of models, or a calculation type performed by one or more of the plurality of models.
8 . The method of claim 1 , wherein identifying, for each model of the plurality of models, the corresponding node of the plurality of nodes comprises:
calculating, for each model of the plurality of models, a plurality of fitness scores for the plurality of nodes; and
selecting, for each model, based on the plurality of fitness scores, the corresponding node.
9 . The method of claim 1 , further comprising configuring each node to communicate with at least one other node of the plurality of nodes.
10 . The method of claim 9 , wherein configuring each node to communicate with at least one other node of the plurality of nodes comprises configuring each node to provide output to or receive input from at least one other node.
11 . The method of claim 1 , further comprising redeploying one or more models of the plurality of models.
12 . An apparatus for distributed model execution, the apparatus configured to perform steps comprising:
identifying, for each model of a plurality of models, based on one or more execution constraints for the plurality of models, a corresponding node of a plurality of nodes,
wherein the plurality of nodes each comprise one or more computing devices or one or more virtual machines;
deploying each model of the plurality of models to the corresponding node of the plurality of nodes; and
wherein the plurality of models are configured to generate, based on data input to at least one model of the plurality of models, a prediction associated with the data.
13 . The apparatus of claim 12 , wherein the one or more execution constraints comprise one or more model dependencies, one or more encryption constraints, or one or more authorization constraints.
14 . The apparatus of claim 12 , wherein the steps further comprise generating the prediction based on a distributed execution of the plurality of models.
15 . The apparatus of claim 12 , wherein identifying the corresponding node of the plurality of nodes is based on one or more node characteristics of the plurality of nodes.
16 . The apparatus of claim 15 , wherein the one or more node characteristics comprise one or more of: one or more hardware resources of one or more of the plurality of nodes, or one or more software resources of one or more of the plurality of nodes.
17 . The apparatus of claim 12 , wherein identifying the corresponding node of the plurality of nodes is based on one or more model characteristics of the plurality of models.
18 . The apparatus of claim 17 , wherein the one or more model characteristics comprise one or more of: a data type for input data to one or more of the plurality of models, or a calculation type performed by one or more of the plurality of models.
19 . The apparatus of claim 12 , wherein identifying, for each model of the plurality of models, the corresponding node of the plurality of nodes comprises:
calculating, for each model of the plurality of models, a plurality of fitness scores for the plurality of nodes; and
selecting, for each model, based on the plurality of fitness scores, the corresponding node.
20 . The apparatus of claim 12 , wherein the steps further comprise configuring each node to communicate with at least one other node of the plurality of nodes.
21 . The apparatus of claim 20 , wherein configuring each node to communicate with at least one other node of the plurality of nodes comprises configuring each node to provide output to or receive input from at least one other node.
22 . The apparatus of claim 12 , wherein the steps further comprise redeploying one or more models of the plurality of models.