Method and system for training and neural network models for large number of discrete features for information rertieval

1. A computer-implemented method for generating neural network models, the method comprising:

receiving a first set of features at an input layer of a neural network model, each of the features in the first set being associated with a predetermined category;

applying a bloom filter coupled between the input layer and a first hidden layer of the neural network model to the first set of features to generate a second set of features, each of the features in the second set representing a combination of two or more features in the first set, wherein a number of nodes in the first hidden layer is fewer than a number of nodes in the input layer;

training the neural network model by applying the second set of features to the first hidden layer of nodes of the neural network model to generate an output, the neural network model including a plurality of layers of nodes coupled to each other via a connection;

comparing the output of the neural network model with a target value associated with the predetermined category to determine whether the neural network model satisfies a predetermined condition,

wherein the neural network model together with the bloom filter is used to determine a category of a user based on a search query received from the user for content delivery, including

obtaining a third set of features based on the search query, device information of a user device, and user information of the user;

applying the bloom filter to the third set of features to generate a fourth set of features; and

applying the neural network model to the fourth set of features to generate the category of the user.

2. The method of claim 1 , wherein the first set of features includes a first number of features that is greater than a second number of features included in the second set of features.

3. The method of claim 2 , further comprising:

adjusting one or more parameters of the bloom filter in response to determining that the neural network model does not satisfy the predetermine condition;

applying the bloom filter with adjusted parameters to the first set of features to generate a fifth set of features, wherein the fifth set of features includes a third number of features different than the second number of features in the second set; and

training the neural network model based on the fifth set of features.

4. The method of claim 1 , further comprising:

in response to a second search query, performing in a content database to identify a list of content items based on the second search query;

for each of the content items, determining a score for the content item using the neural network model;

ranking the content items based on scores associated with the content items; and

generating a search result having the content items sorted based on the rankings.

5. The method of claim 4 , wherein determining a score for the content item using the neural network model comprises:

obtaining a fifth set of features associated with the content item;

applying the bloom filter to the fifth set of features to generate a sixth set of features; and

applying the neural network model to the sixth set of features to generate the score.

6. The method of claim 1 , further comprising:

performing a search in a content database based on the search query and the category of the user.

7. The method of claim 1 , wherein information that the user intends to receive is determined based on the category of the user, and a type of content that is associated with the category of the user is searched.

8. A non-transitory machine-readable medium having instructions stored therein, which when executed by a processor, cause the processor to perform operations of training classification models, the operations comprising:

receiving a first set of features at an input layer of a neural network model, each of the features in the first set being associated with a predetermined category;

applying a bloom filter coupled between the input layer and a first hidden layer of the neural network model to the first set of features to generate a second set of features, each of the features in the second set representing a combination of two or more features in the first set, wherein a number of nodes in the first hidden layer is fewer than a number of nodes in the input layer;

training a neural network model by applying the second set of features to the first hidden layer of nodes of the neural network model to generate an output, the neural network model including a plurality of layers of nodes coupled to each other via a connection;

comparing the output of the neural network model with a target value associated with the predetermined category to determine whether the neural network model satisfies a predetermined condition,

wherein the neural network model together with the bloom filter is used to determine a category of a user based on a search query received from the user for content delivery, including

obtaining a third set of features based on the search query, device information of a user device, and user information of the user;

applying the bloom filter to the third set of features to generate a fourth set of features; and

applying the neural network model to the fourth set of features to generate the category of the user.

9. The non-transitory machine-readable medium of claim 8 , wherein the first set of features includes a first number of features that is greater than a second number of features included in the second set of features.

10. The non-transitory machine-readable medium of claim 9 , wherein the operations further comprise:

adjusting one or more parameters of the bloom filter in response to determining that the neural network model does not satisfy the predetermine condition;

applying the bloom filter with adjusted parameters to the first set of features to generate a fifth set of features, wherein the fifth set of features includes a third number of features different than the second number of features in the second set; and

training the neural network model based on the fifth set of features.

11. The non-transitory machine-readable medium of claim 8 , wherein the operations further comprise:

in response to a second search query, performing in a content database to identify a list of content items based on the second search query;

for each of the content items, determining a score for the content item using the neural network model;

ranking the content items based on scores associated with the content items; and

generating a search result having the content items sorted based on the rankings.

12. The non-transitory machine-readable medium of claim 11 , wherein determining a score for the content item using the neural network model comprises:

obtaining a fifth set of features associated with the content item;

applying the bloom filter to the fifth set of features to generate a sixth set of features; and

applying the neural network model to the sixth set of features to generate the score.

13. The non-transitory machine-readable medium of claim 8 , wherein the operations further comprise:

performing a search in a content database based on the search query and the category of the user.

14. The non-transitory machine-readable medium of claim 8 , wherein information that the user intends to receive is determined based on the category of the user, and a type of content that is associated with the category of the user is searched.

15. A data processing system, comprising:

a processor; and

a memory coupled to the processor, the memory storing instructions, which when executed by the processor, cause the processor to perform of training classification models, the operations including

receiving a first set of features at an input layer of a neural network model, each of the features in the first set being associated with a predetermined category,

applying a bloom filter coupled between the input layer and a first hidden layer of the neural network model to the first set of features to generate a second set of features, each of the features in the second set representing a combination of two or more features in the first set, wherein a number of nodes in the first hidden layer is fewer than a number of nodes in the input layer,

training a neural network model by applying the second set of features to the first hidden layer of nodes of the neural network model to generate an output, the neural network model including a plurality of layers of nodes coupled to each other via a connection,

comparing the output of the neural network model with a target value associated with the predetermined category to determine whether the neural network model satisfies a predetermined condition,

wherein the neural network model together with the bloom filter is used to determine a category of a user based on a search query received from the user for content delivery, including

obtaining a third set of features based on the search query, device information of a user device, and user information of a user;

applying the bloom filter to the third set of features to generate a fourth set of features; and

applying the neural network model to the fourth set of features to generate the category of the user.

16. The system of claim 15 , wherein the first set of features includes a first number of features that is greater than a second number of features included in the second set of features.

17. The system of claim 16 , wherein the operations further comprise:

adjusting one or more parameters of the bloom filter in response to determining that the neural network model does not satisfy the predetermine condition;

applying the bloom filter with adjusted parameters to the first set of features to generate a fifth set of features, wherein the fifth set of features includes a third number of features different than the second number of features in the second set; and

training the neural network model based on the fifth set of features.

18. The system of claim 15 , wherein the operations further comprise:

in response to a second search query, performing in a content database to identify a list of content items based on the second search query;

for each of the content items, determining a score for the content item using the neural network model;

ranking the content items based on scores associated with the content items; and

generating a search result having the content items sorted based on the rankings.

19. The system of claim 18 , wherein determining a score for the content item using the neural network model comprises:

obtaining a fifth set of features associated with the content item;

applying the bloom filter to the fifth set of features to generate a sixth set of features; and

applying the neural network model to the sixth set of features to generate the score.

20. The system of claim 15 , wherein the operations further comprise:

performing a search in a content database based on the search query and the category of the user.

21. The system of claim 15 , wherein information that the user intends to receive is determined based on the category of the user, and a type of content that is associated with the category of the user is searched.