Method for training feature quantization model, feature quantization method, data query methods and electronic devices

1 . A method for training a feature quantization model, comprising:

obtaining a plurality of source data domains;

obtaining feature information and labeling information of each of the plurality of source data domains;

decomposing the feature information of each of the plurality of source data domains, so as to obtain common feature information and domain-specific feature information of the plurality of source data domains, the common feature information being feature information common to the plurality of source data domains; and

training a feature quantization model according to the common feature information of all the source data domains, and domain-specific feature information and the labeling information of each of the plurality of source data domains, so as to obtain a common feature quantization model;

wherein training the feature quantization model includes:

adjusting the feature quantization model so that for a k-th source data domain, Ex(L(F 0 (x), y)) takes a minimum value, wherein

x represents feature information of the k-th source data domain, y represents labeling information of the k-th source data domain, F 0 represents the common feature quantization model, F 0 (x) represents a feature quantization code obtained after the feature information x is processed by F 0 , L(F 0 (x), y) represents a loss function between the feature quantization code F 0 (x) and the labeling information y, Ex(L(F 0 (x), y)) represents a mathematical expectation of the loss function L(F 0 (x), y) of the feature information x, k takes a value in a

range of from 1 to K, and K is a number of the plurality of source data domains; and

adjusting the feature quantization model so that for the k-th source data domain, Ex(L(φ(F 0 (x), F k (x)), y)) takes a minimum value, and for the k-th source data domain and a p-th source data domain, Ex(L(φ(F 0 (x), F k (x)), y)) is less than Ex(L(φ(F 0 (x), F p (x)), y)), p being not equal to k, wherein

x represents the feature information of the k-th source data domain, y represents the labeling information of the k-th source data domain, F 0 represents the common feature quantization model, F 0 (x) represents the feature quantization code obtained after the feature information x is processed by F 0 , F k represents a domain-specific feature quantization model of the k-th source data domain, F(x) represents a feature quantization code obtained after the feature information x is processed by F k , F p represents a domain-specific feature quantization model of the p-th source data domain, F p (x) represents a feature quantization code obtained after the feature information x is processed by F p , φ(F 0 (x), F k (x)) represents a fusion of F 0 (x) and F k (x), φ(F 0 (x), F p (x)) represents a fusion of F 0 (x) and F p (x), L(φ(F 0 (x), F k (x)), y) represents a loss function between a fused feature quantization code obtained by φ(F 0 (x), F k (X)) and the labeling information y, L(((F 0 (x), F p (x)), y) represents a loss function between a fused feature quantization code obtained by φ(F 0 (x), F p (x)) and the labeling information y, Ex(L(φ(F 0 (x), F p (x)), y)) represents a mathematical expectation function of L(φ(F 0 (x), F p (x)), y), k takes a value in a range of from 1 to K, p takes a value in a range of from 1 to K, and K is the number of the plurality of source data domains.

2 . The method for training the feature quantization model according to claim 1 , wherein

training the feature quantization model according to the common feature information of all the source data domains, and the domain-specific feature information and the labeling information of each of the plurality of source data domains, so as to obtain the common feature quantization model includes:

training the feature quantization model according to the common feature information of all the source data domains, and the domain-specific feature information and the labeling information of each of the plurality of source data domains, so as to obtain the common feature quantization model and a domain-specific feature quantization model of each of the plurality of source data domains.

3 . The method for training the feature quantization model according to claim 2 , wherein a deep neural network algorithm is used to train the feature quantization model.

4 . The method for training the feature quantization model according to claim 1 , wherein the fusion is performed by using an addition or a linear splicing.

5 . A feature quantification method, comprising:

performing feature quantization on a target data set by using a common feature quantization model to obtain a feature quantization code of the target data set, the common feature quantization model being obtained by using the method for training the feature quantization model according to claim 1 .

6 . A data query method applied to a server, comprising:

receiving a target feature quantization code of target query data sent by a client;

comparing the target feature quantization code with a feature quantization code of a target data set to obtain a query result matching the target feature quantization code, wherein the feature quantization code of the target data set is obtained by using the feature quantization method according to claim 5 ; and

returning the query result to the client.

7 . The data query method according to claim 6 , wherein the feature quantization code of the target data set is obtained by using the common feature quantization model to perform feature quantization on the target data set and stored in advance.

8 . An electronic device comprising a processor, a memory, and a computer program stored on the memory and executable on the processor, wherein the computer program is executed by the processor to implement steps of the data query method according to claim 6 .

9 . A non-transitory computer-readable storage medium storing a computer program thereon, wherein the computer program is executed by the processor to implement steps of the data query method according to claim 6 .

10 . An electronic device comprising a processor, a memory, and a computer program stored on the memory and executable on the processor, wherein the computer program is executed by the processor to implement steps of the feature quantization method according to claim 5 .

11 . A non-transitory computer-readable storage medium storing a computer program thereon, wherein the computer program is executed by the processor to implement steps of the feature quantization method according to claim 5 .

12 . A data query method applied to a client, the method comprising:

obtaining input target query data;

performing feature quantization on the target query data by using a common feature quantization model to obtain a target feature quantization code of the target query data, the common feature quantization model being obtained by using the method for training the feature quantization model according to claim 1 ;

sending the target feature quantization code to a server; and

receiving a query result returned, according to the target feature quantization code, by the server.

13 . An electronic device comprising a processor, a memory, and a computer program stored on the memory and executable on the processor, wherein the computer program is executed by the processor to implement steps of the data query method according to claim 12 .

14 . A non-transitory computer-readable storage medium storing a computer program thereon, wherein the computer program is executed by the processor to implement steps of the data query method according to claim 12 .

15 . An electronic device comprising a processor, a memory, and a computer program stored on the memory and executable on the processor, wherein the computer program is executed by the processor to implement steps of the method for training the feature quantization model according to claim 1 .

16 . A non-transitory computer-readable storage medium storing a computer program thereon, wherein the computer program is executed by the processor to implement steps of the method for training the feature quantization model according to claim 1 .