Feature processing tradeoff management

Abstract

At a machine learning service, a set of candidate variables that can be used to train a model is identified, including at least one processed variable produced by a feature processing transformation. A cost estimate indicative of an effect of implementing the feature processing transformation on a performance metric associated with a prediction goal of the model is determined. Based at least in part on the cost estimate, a feature processing proposal that excludes the feature processing transformation is implemented.

1 .- 21 . (canceled) 
     
     
         22 . A computer-implemented method, comprising:
 obtaining, via one or more programmatic interfaces at a service of a cloud computing environment, a request to tune at least a particular hyper-parameter of a machine learning task;   executing, at the service, one or more tuning operations in which respective values of the particular hyper-parameter are utilized for the machine learning task; and   providing, via the one or more programmatic interfaces, a result of the one or more tuning operations.   
     
     
         23 . The computer-implemented method as recited in  claim 22 , further comprising:
 selecting, by the service, a first value of the respective values of the particular hyper-parameter, without receiving an indication of the first value via the one or more programmatic interfaces.   
     
     
         24 . The computer-implemented method as recited in  claim 22 , further comprising:
 obtaining, at the service via the one or more programmatic interfaces, an indication of a criterion for terminating execution of the one or more tuning operations; and   terminating, by the service, execution of the one or more tuning operations based at least in part on the criterion.   
     
     
         25 . The computer-implemented method as recited in  claim 22 , further comprising:
 determining, by the service, at least one value of the respective values of the particular hyper-parameter based at least in part on a size of an input data set of the machine learning task.   
     
     
         26 . The computer-implemented method as recited in  claim 22 , wherein the result provided via the one or more programmatic interfaces comprises a plurality of candidate values identified for the particular hyper-parameter in the one or more tuning operations. 
     
     
         27 . The computer-implemented method as recited in  claim 22 , wherein the machine learning task comprises one or more of: (a) training a machine learning model or (b) executing a trained machine learning model. 
     
     
         28 . The computer-implemented method as recited in  claim 22 , wherein the particular hyper-parameter comprises a dimensionality of a vector representation of a data object. 
     
     
         29 . A system, comprising:
 one or more computing devices;   wherein the one or more computing devices include instructions that upon execution on or across the one or more computing devices cause the one or more computing devices to:
 obtain, via one or more programmatic interfaces at a service of a cloud computing environment, a request to tune at least a particular hyper-parameter of a machine learning task; 
 execute, at the service, one or more tuning operations in which respective values of the particular hyper-parameter are utilized for the machine learning task; and 
 provide, via the one or more programmatic interfaces, a result of the one or more tuning operations. 
   
     
     
         30 . The system as recited in  claim 29 , wherein the one or more computing devices include further instructions that upon execution on or across the one or more computing devices further cause the one or more computing devices to:
 select, by the service, a first value of the respective values of the particular hyper-parameter, without receiving an indication of the first value via the one or more programmatic interfaces.   
     
     
         31 . The system as recited in  claim 29 , wherein the one or more computing devices include further instructions that upon execution on or across the one or more computing devices further cause the one or more computing devices to:
 obtain, at the service via the one or more programmatic interfaces, an indication of a criterion for terminating execution of the one or more tuning operations; and   terminate, by the service, execution of the one or more tuning operations based at least in part on the criterion.   
     
     
         32 . The system as recited in  claim 29 , wherein the one or more computing devices include further instructions that upon execution on or across the one or more computing devices further cause the one or more computing devices to:
 determine, by the service, at least one value of the respective values of the particular hyper-parameter based at least in part on a size of an input data set of the machine learning task.   
     
     
         33 . The system as recited in  claim 29 , wherein the result provided via the one or more programmatic interfaces comprises a plurality of candidate values identified for the particular hyper-parameter in the one or more tuning operations. 
     
     
         34 . The system as recited in  claim 29 , wherein the machine learning task comprises one or more of: (a) training a machine learning model or (b) executing a trained machine learning model. 
     
     
         35 . The system as recited in  claim 29 , wherein the particular hyper-parameter comprises a dimensionality of a vector representation of a data object. 
     
     
         36 . One or more non-transitory computer-accessible storage media storing program instructions that when executed on or across one or more processors cause the one or more processors to:
 obtain, via one or more programmatic interfaces at a service of a cloud computing environment, a request to tune at least a particular hyper-parameter of a machine learning task;   execute, at the service, one or more tuning operations in which respective values of the particular hyper-parameter are utilized for the machine learning task; and   provide, via the one or more programmatic interfaces, a result of the one or more tuning operations.   
     
     
         37 . The one or more non-transitory computer-accessible storage media as recited in  claim 36 , storing further program instructions that upon execution on or across the one or more processors further cause the one or more processors to:
 select, by the service, a first value of the respective values of the particular hyper-parameter, without receiving an indication of the first value via the one or more programmatic interfaces.   
     
     
         38 . The one or more non-transitory computer-accessible storage media as recited in  claim 36 , storing further program instructions that upon execution on or across the one or more processors further cause the one or more processors to:
 obtain, at the service via the one or more programmatic interfaces, an indication of a criterion for terminating execution of the one or more tuning operations; and   terminate, by the service, execution of the one or more tuning operations based at least in part on the criterion.   
     
     
         39 . The one or more non-transitory computer-accessible storage media as recited in  claim 36 , storing further program instructions that upon execution on or across the one or more processors further cause the one or more processors to:
 determine, by the service, at least one value of the respective values of the particular hyper-parameter based at least in part on a size of an input data set of the machine learning task.   
     
     
         40 . The one or more non-transitory computer-accessible storage media as recited in  claim 36 , wherein the result provided via the one or more programmatic interfaces comprises a plurality of candidate values identified for the particular hyper-parameter in the one or more tuning operations. 
     
     
         41 . The one or more non-transitory computer-accessible storage media as recited in  claim 36 , wherein the machine learning task comprises one or more of: (a) training a machine learning model or (b) executing a trained machine learning model.