Automated machine learning pipeline generation
Techniques regarding one or more automated machine learning processes that analyze time series data are provided. For example, one or more embodiments described herein can comprise a system, which can comprise a memory that can store computer executable components. The system can also comprise a processor, operably coupled to the memory, and that can execute the computer executable components stored in the memory. The computer executable components can comprise a time series analysis component that selects a machine learning pipeline for meta transfer learning on time series data by sequentially allocating subsets of training data from the time series data amongst a plurality of machine learning pipeline candidates.
1. A system, comprising:
a memory that stores computer executable components; and
a processor, operably coupled to the memory, and that executes the computer executable components stored in the memory, wherein the computer executable components comprise:
a time series analysis component that selects a machine learning pipeline for meta transfer learning on time series data by sequentially allocating subsets of training data from the time series data amongst a plurality of machine learning pipeline candidates, wherein the time series analysis component comprises:
a learner component that ranks the machine learning pipeline amongst the plurality of machine learning pipeline candidates based on meta-data derived from historic execution of the machine learning pipeline and a characteristic of the time series data; and
a joint optimization component that allocates the subsets of training data using a data allocation using upper bound algorithm, wherein a subset of the subsets is selected based on a chronological order of the subset with respect to a plurality of subsets of the time series data.
2. The system of claim 1 , wherein the time series analysis component further selects the machine learning pipeline based on a runtime threshold that restricts an amount of time allotted to execution of a machine learning task.
3. The system of claim 1 , further comprising:
a use case component that depopulates the plurality of machine learning pipeline candidates based on a domain of the time series data.
4. The system of claim 1 , further comprising:
a hyperparameter component that configures a hyperparameter employed in the machine learning pipeline.
5. The system of claim 1 , further comprising:
a feature component that engineers a feature of the machine learning pipeline based on a semantic relationship defined within a knowledge database.
6. The system of claim 1 , further comprising:
a data collection component that generates a data request based on a data correlation defined by the machine learning pipeline, wherein the data request regards data relevant to a machine learning task and absent from the time series data.
7. The system of claim 1 , further comprising:
an ensemble component that generates an ensemble of machine learning pipelines selected for the meta transfer learning, wherein the machine learning pipeline is comprised within the ensemble of machine learning pipelines.
8. The system of claim 7 , further comprising:
a task component that executes a machine learning task by employing the ensemble of machine learning pipelines, wherein a weight value delineates an amount of contribution attributed to the machine learning pipeline in executing the machine learning task.
9. A computer-implemented method, comprising:
selecting, by a system operatively coupled to a processor, a machine learning pipeline for meta transfer learning on time series data by sequentially allocating subsets of training data from the time series data amongst a plurality of machine learning pipeline candidates, wherein the selecting comprises:
ranking, by the system, the machine learning pipeline amongst the plurality of machine learning pipeline candidates based on meta-data derived from historic execution of the machine learning pipeline and a characteristic of the time series data; and
allocating, by the system, the subsets of training data using a data allocation using upper bound algorithm, wherein a subset of the subsets is selected based on a chronological order of the subset with respect to a plurality of subsets of the time series data.
10. The computer-implemented method of claim 9 , further comprising:
depopulating, by the system, the plurality of machine learning pipeline candidates based on a domain of the time series data.
11. The computer-implemented method of claim 9 , further comprising:
generating, by the system, a data request based on a data correlation defined by the machine learning pipeline, wherein the data request regards data relevant to a machine learning task and absent from the time series data.
12. The computer-implemented method of claim 9 , further comprising:
generating, by the system, an ensemble of machine learning pipelines selected for the meta transfer learning, wherein the machine learning pipeline is comprised within the ensemble of machine learning pipelines; and
executing, by the system, a machine learning task by employing the ensemble of machine learning pipelines, wherein a weight value delineates an amount of contribution attributed to the machine learning pipeline in executing the machine learning task.
13. The computer-implemented method of claim 9 , wherein the selecting the machine learning pipeline is based on a runtime threshold that restricts an amount of time allotted to execution of a machine learning task.
14. A computer program product for generating an automated machine learning process that analyzes time series data, the computer program product comprising a computer readable storage medium having program instructions embodied therewith, the program instructions executable by a processor to cause the processor to:
select, by the processor, a machine learning pipeline for meta transfer learning on the time series data by sequentially allocating subsets of training data from the time series data amongst a plurality of machine learning pipeline candidates, wherein the selecting comprises:
rank, by the processor, the machine learning pipeline amongst the plurality of machine learning pipeline candidates based on meta-data derived from historic execution of the machine learning pipeline and a characteristic of the time series data; and
allocate, by the processor, the subsets of training data using a data allocation using upper bound algorithm, wherein a subset of the subsets is selected based on a chronological order of the subset with respect to a plurality of subsets of the time series data.
15. The computer program product of claim 14 , wherein the program instructions further cause the processor to:
depopulate, by the processor, the plurality of machine learning pipeline candidates based on a domain of the time series data.
16. The computer program product of claim 14 , wherein the program instructions further cause the processor to:
generate, by the processor, a data request based on a data correlation defined by the machine learning pipeline, wherein the data request regards data relevant to a machine learning task and absent from the time series data.
17. The computer program product of claim 16 , wherein the program instructions further cause the processor to:
generate, by the processor, an ensemble of machine learning pipelines selected for the meta transfer learning, wherein the machine learning pipeline is comprised within the ensemble of machine learning pipelines; and
execute, by the processor, the machine learning task by employing the ensemble of machine learning pipelines, wherein a weight value delineates an amount of contribution attributed to the machine learning pipeline in executing the machine learning task.
18. The computer program product of claim 14 , wherein the selecting the machine learning pipeline is based on a runtime threshold that restricts an amount of time allotted to execution of a machine learning task.
19. The computer program product of claim 14 , wherein the program instructions further cause the processor to:
configure, by the processor, a hyperparameter employed in the machine learning pipeline.
20. The computer program product of claim 14 , wherein the program instructions further cause the processor to:
engineer, by the processor, a feature of the machine learning pipeline based on a semantic relationship defined within a knowledge database.