Methods and apparatus for machine learning model hyperparameter optimization
This application relates to apparatus and methods for optimizing hyperparameters for machine learning models. In some examples, a computing device configures a machine learning model with a first set of hyperparameters from a pool of hyperparameters. The computing device may execute the machine learning model to generate a validation score, and may update parameters of a probability determination model based on the validation score. Further, the computing device may execute the probability determination model to generate probabilities corresponding to the first set of hyperparameters. The computing device may also determine a second set of hyperparameters from the pool of hyperparameters based on the generated probabilities, and may configure the machine learning model with the second set of hyperparameters. The computing device may repeat the process of generating validation scores, updating the parameters, and generating probabilities to determine sets of hyperparameters until at least one condition is satisfied.
1. A system comprising:
a computing device configured to:
configure a machine learning model with a first set of hyperparameters from a pool of hyperparameters;
generate a first validation score based on execution of the machine learning model configured with the first set of hyperparameters;
update parameters of a probability determination model based on the first validation score;
generate a first plurality of values corresponding to the first set of hyperparameters based on a first execution of the probability determination model;
determine a second set of hyperparameters from the pool of hyperparameters based on the first plurality of values;
configure the machine learning model with the second set of hyperparameters;
receive a search request comprising a search query from a web server;
determine search results based on executing the machine learning model configured with the second set of hyperparameters to operate on the search query; and
transmit the search results to the web server.
2. The system of claim 1 , wherein the computing device is configured to:
generate a second validation score based on execution of the machine learning model configured with the second set of hyperparameters; and
determine whether at least one condition is satisfied based on generating the second validation score.
3. The system of claim 2 , wherein determining whether the at least one condition is satisfied comprises:
determining that the at least on condition is satisfied; and
executing the machine learning model configured with the second set of hyperparameters to generate output data.
4. The system of claim 2 , wherein determining whether the at least one condition is satisfied comprises:
determining that the at least on condition is not satisfied;
updating the parameters of the probability determination model based on the second validation score;
generating a second plurality of values corresponding to the second set of hyperparameters based on a second execution of the probability determination model;
determining a third set of hyperparameters from the pool of hyperparameters based on the second plurality of values; and
configuring the machine learning model with the third set of hyperparameters.
5. The system of claim 1 , wherein the first plurality of values are probabilities.
6. The system of claim 1 , wherein the probability determination model is a neural network.
7. The system of claim 1 , wherein generating the first validation score based on execution of the machine learning model comprises:
training the machine learning model;
executing the trained machine learning model to operate on validation data to generate output results; and
generating the first validation score based on the output results and expected results.
8. The system of claim 7 , wherein generating the first validation score based on the output results and expected results comprises generating an F 1 score.
9. The system of claim 7 , wherein training the machine learning model comprises training the machine learning model with user session data and user transaction data.
10. A method comprising:
configuring a machine learning model with a first set of hyperparameters from a pool of hyperparameters;
generating a first validation score based on execution of the machine learning model configured with the first set of hyperparameters;
updating parameters of a probability determination model based on the first validation score;
generating a first plurality of values corresponding to the first set of hyperparameters based on a first execution of the probability determination model;
determining a second set of hyperparameters from the pool of hyperparameters based on the first plurality of values;
configuring the machine learning model with the second set of hyperparameters;
receiving a search request comprising a search query from a web server;
determining search results based on executing the machine learning model configured with the second set of hyperparameters to operate on the search query; and
transmitting the search results to the web server.
11. The method of claim 10 further comprising:
generating a second validation score based on execution of the machine learning model configured with the second set of hyperparameters; and
determining whether at least one condition is satisfied based on generating the second validation score.
12. The method of claim 11 , wherein determining whether the at least one condition is satisfied comprises:
determining that the at least on condition is satisfied; and
executing the machine learning model configured with the second set of hyperparameters to generate output data.
13. The method of claim 11 , wherein determining whether the at least one condition is satisfied comprises:
determining that the at least on condition is not satisfied;
updating the parameters of the probability determination model based on the second validation score;
generating a second plurality of values corresponding to the second set of hyperparameters based on a second execution of the probability determination model;
determining a third set of hyperparameters from the pool of hyperparameters based on the second plurality of values; and
configuring the machine learning model with the third set of hyperparameters.
14. The method of claim 11 wherein generating the first validation score based on execution of the machine learning model comprises:
training the machine learning model;
executing the trained machine learning model to operate on validation data to generate output results; and
generating the first validation score based on the output results and expected results.
15. The method of claim 14 , wherein training the machine learning model comprises training the machine learning model with user session data and user transaction data.
16. A non-transitory computer readable medium having instructions stored thereon, wherein the instructions, when executed by at least one processor, cause a device to perform operations comprising:
configuring a machine learning model with a first set of hyperparameters from a pool of hyperparameters;
generating a first validation score based on execution of the machine learning model configured with the first set of hyperparameters;
updating parameters of a probability determination model based on the first validation score;
generating a first plurality of values corresponding to the first set of hyperparameters based on a first execution of the probability determination model;
determining a second set of hyperparameters from the pool of hyperparameters based on the first plurality of values;
configuring the machine learning model with the second set of hyperparameters;
receiving a search request comprising a search query from a web server;
determining search results based on executing the machine learning model configured with the second set of hyperparameters to operate on the search query; and
transmitting the search results to the web server.
17. The non-transitory computer readable medium of claim 16 further comprising instructions stored thereon that, when executed by the at least one processor, further cause the device to perform operations comprising:
generating a second validation score based on execution of the machine learning model configured with the second set of hyperparameters; and
determining whether at least one condition is satisfied based on generating the second validation score.
18. The non-transitory computer readable medium of claim 17 further comprising instructions stored thereon that, when executed by the at least one processor, further cause the device to perform operations comprising:
determining that the at least on condition is satisfied; and
executing the machine learning model configured with the second set of hyperparameters to generate output data.
19. The non-transitory computer readable medium of claim 17 further comprising instructions stored thereon that, when executed by the at least one processor, further cause the device to perform operations comprising:
determining that the at least on condition is not satisfied;
updating the parameters of the probability determination model based on the second validation score;
generating a second plurality of values corresponding to the second set of hyperparameters based on a second execution of the probability determination model;
determining a third set of hyperparameters from the pool of hyperparameters based on the second plurality of values; and
configuring the machine learning model with the third set of hyperparameters.
20. The non-transitory computer readable medium of claim 16 further comprising instructions stored thereon that, when executed by the at least one processor, further cause the device to perform operations comprising:
training the machine learning model;
executing the trained machine learning model to operate on validation data to generate output results; and
generating the first validation score based on the output results and expected results.