Modular adaptation for cross-domain few-shot learning
A method, apparatus and system for adapting a pre-trained network for application to a different dataset includes arranging at least two different types of active adaptation modules in a pipeline configuration, wherein an output of a previous active adaptation module produces an input for a next active adaptation module in the pipeline in the form of adapted network data until a last active adaptation module, and wherein each of the at least two different types of adaptation modules can be switched on or off, determining at least one respective hyperparameter for each of the at least two different types of active adaptation modules, and applying the at least one respective determined hyperparameter to each of the at least two different types of active adaptation modules for processing received data from the pretrained network to determine an adapted network.
1. A method for adapting a pre-trained network for application to a different dataset, comprising:
arranging at least two different types of active adaptation modules in a pipeline configuration, wherein an output of a previous active adaptation module produces an input for a next active adaptation module in the pipeline in the form of adapted network data until a last active adaptation module, and wherein each of the at least two different types of adaptation modules can be switched on or off;
determining at least one respective hyperparameter for each of the at least two different types of active adaptation modules; and
applying the at least one respective determined hyperparameter to each of the at least two different types of active adaptation modules for processing received data from the pretrained network to determine an adapted network.
2. The method of claim 1 , further comprising:
applying an adaptive learning process to an output of at least one of the at least two different types of active adaptation modules.
3. The method of claim 1 , wherein the arranging, the determining and the applying comprise an iterative process in which at least one of the at least two different types of active adaptation modules in the pipeline or a respective hyperparameter is changed in each subsequent iteration.
4. The method of claim 1 , wherein at least one of the at least two different types of the active adaptation modules or the respective hyperparameters are selected from a collection of historically well-functioning adaptation modules or hyperparameters stored in a storage device.
5. The method of claim 1 , wherein at least one of the at least two different types of the active adaptation modules or the respective hyperparameters are selected based on at least one a user input or an input from a machine learning process.
6. The method of claim 5 , wherein the machine learning process is trained to determine at least one of an adaptation module for the pipeline or a hyperparameter for an adaptation module based on a target dataset.
7. The method of claim 1 , wherein data from the pre-trained network comprises at least one of classification data or data regarding an embedding space.
8. A non-transitory machine-readable medium having stored thereon at least one program, the at least one program including instructions which, when executed by a processor, cause the processor to perform a method in a processor based system for adapting a pre-trained network for application to a different dataset, comprising:
arranging at least two different types of active adaptation modules in a pipeline configuration, wherein an output of a previous active adaptation module produces an input for a next active adaptation module in the pipeline in the form of adapted network data until a last active adaptation module, and wherein each of the at least two different types of adaptation modules can be switched on or off;
determining at least one respective hyperparameter for each of the at least two different types of active adaptation modules; and
applying the at least one respective determined hyperparameter to each of the at least two different types of active adaptation modules for processing received data from the pretrained network to determine an adapted network.
9. The non-transitory machine-readable medium of claim 8 , further comprising:
applying an adaptive learning process to an output of at least one of the at least two different types of active adaptation modules.
10. The non-transitory machine-readable medium of claim 8 , in which the arranging, the determining and the applying steps comprise an iterative process in which at least one of an order of the at least two different types of active adaptation modules in the pipeline or a respective hyperparameter is changed in each subsequent iteration.
11. The non-transitory machine-readable medium of claim 8 , wherein at least one of the at least two different types of the active adaptation modules or the respective hyperparameters are selected from a collection of historically well-functioning adaptation modules or hyperparameters stored in a storage device.
12. The non-transitory machine-readable medium of claim 8 , wherein at least one of the at least two different types of the active adaptation modules or the respective hyperparameters are selected based on at least one a user input or an input from a machine learning process.
13. The non-transitory machine-readable medium of claim 12 , wherein the machine learning process is trained to determine at least one of an adaptation module for the pipeline or a hyperparameter for an adaptation module based on a target dataset.
14. The non-transitory machine-readable medium of claim 8 , wherein data from the pre-trained network comprises at least one of classification data or data regarding an embedding space.
15. A system for adapting a pre-trained network for application to a different dataset, comprising:
a storage device; and
a computing device comprising a processor and a memory having stored therein at least one program, the at least one program including instructions which, when executed by the processor, cause the computing device to perform a method, comprising:
arranging at least two different types of active adaptation modules in a pipeline configuration, wherein an output of a previous active adaptation module produces an input for a next active adaptation module in the pipeline in the form of adapted network data until a last active adaptation module, and wherein each of the at least two different types of adaptation modules can be switched on or off;
determining at least one respective hyperparameter for each of the at least two different types of active adaptation modules; and
applying the at least one respective determined hyperparameter to each of the at least two different types of active adaptation modules for processing received data from the pretrained network to determine an adapted network.
16. The system of claim 15 , further comprising:
applying an adaptive learning process to an output of at least one of the at least two different types of active adaptation modules.
17. The system of claim 15 , in which the arranging, the determining and the applying steps comprise an iterative process in which at least one of an order of the at least two different types of active adaptation modules in the pipeline or a respective hyperparameter is changed in each subsequent iteration.
18. The system of claim 15 , wherein at least one of the at least two different types of the active adaptation modules or the respective hyperparameters are selected from a collection of historically well-functioning adaptation modules or hyperparameters stored in the storage device.
19. The system of claim 15 , wherein at least one of the at least two different types of the active adaptation modules or the respective hyperparameters are selected based on at least one a user input or an input from a machine learning process.
20. The method of claim 19 , wherein the machine learning process is trained to determine at least one of an adaptation module for the pipeline or a hyperparameter for an adaptation module based on a known target dataset.