CN111898738A - Mobile terminal user gender prediction method and system based on full-connection neural network - Google Patents
Mobile terminal user gender prediction method and system based on full-connection neural network Download PDFInfo
- Publication number
- CN111898738A CN111898738A CN202010749181.3A CN202010749181A CN111898738A CN 111898738 A CN111898738 A CN 111898738A CN 202010749181 A CN202010749181 A CN 202010749181A CN 111898738 A CN111898738 A CN 111898738A
- Authority
- CN
- China
- Prior art keywords
- mobile terminal
- feature
- gender
- neural network
- discrete
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 51
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 18
- 238000003062 neural network model Methods 0.000 claims abstract description 18
- 238000012545 processing Methods 0.000 claims abstract description 13
- 238000013507 mapping Methods 0.000 claims abstract description 12
- 238000012549 training Methods 0.000 claims abstract description 12
- 238000009434 installation Methods 0.000 claims description 25
- 210000002569 neuron Anatomy 0.000 claims description 12
- 230000004913 activation Effects 0.000 claims description 8
- 238000004458 analytical method Methods 0.000 claims description 7
- 230000011218 segmentation Effects 0.000 claims description 5
- 230000000007 visual effect Effects 0.000 claims description 5
- 239000011159 matrix material Substances 0.000 claims description 4
- 238000012800 visualization Methods 0.000 claims description 2
- 230000008569 process Effects 0.000 abstract description 9
- 239000013598 vector Substances 0.000 description 22
- 230000006870 function Effects 0.000 description 7
- 238000010801 machine learning Methods 0.000 description 6
- 230000006399 behavior Effects 0.000 description 4
- 238000012360 testing method Methods 0.000 description 4
- 241000234435 Lilium Species 0.000 description 3
- 241000234479 Narcissus Species 0.000 description 3
- 241000220317 Rosa Species 0.000 description 3
- 230000008901 benefit Effects 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- VYZAMTAEIAYCRO-UHFFFAOYSA-N Chromium Chemical compound [Cr] VYZAMTAEIAYCRO-UHFFFAOYSA-N 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000002354 daily effect Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000003203 everyday effect Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000035515 penetration Effects 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000012216 screening Methods 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F40/00—Handling natural language data
- G06F40/20—Natural language analysis
- G06F40/205—Parsing
- G06F40/216—Parsing using statistical methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- Probability & Statistics with Applications (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Audiology, Speech & Language Pathology (AREA)
- Medical Informatics (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a method and a system for predicting the gender of a mobile terminal user based on a fully-connected neural network. The method comprises the steps of obtaining sample data of the mobile terminal, carrying out feature classification to obtain a continuity feature and a discrete feature, respectively processing the continuity feature and the discrete feature, expressing the continuity feature and the discrete feature by adopting one-shot codes to obtain a one-shot sample feature, carrying out Embelling mapping on all the one-shot sample features, constructing and training a fully-connected neural network model based on the sample features after the Embelling mapping, adopting the trained fully-connected neural network model, inputting user features of the mobile terminal, predicting the gender of a user and the like. The invention also discloses a prediction system comprising the full-connection neural network model, which is used for the gender prediction method of the mobile terminal user. The technical scheme of the invention can ensure that the modeling process is more complete and the result is more accurate.
Description
Technical Field
The invention belongs to the technical field of mobile internet, and particularly relates to a method and a system for predicting the gender of a mobile terminal user based on a full-connection neural network.
Background
With the rapid development of mobile networks and smart phones, almost everyone can keep away from the phone. The reports of consulting companies show that the penetration of smartphones in the mobile market has risen from 50% in 2014 to 80% in 2019 in the last 5 years, and is expected to reach 95% by the end of 2020. In daily life, people almost use mobile phones to browse webpages, chat and online shopping every day, and the online data of the mobile phones can intuitively reflect the attribute characteristics and behavior preference of users. Therefore, an operator can collect data such as an APP installation list, an APP usage record, a terminal type and a terminal price of a mobile user terminal through an intelligent network management platform, and can perform accurate portrait work of the mobile user by combining with a machine learning algorithm of an open source on GiHtub, for example, information such as the age and the sex of the user can be predicted, and the information is very important customer label attributes in accurate marketing. The method can help the Internet company to know the behavior characteristics of the user, develop products iteratively, and help enterprises to improve the accuracy of advertisement putting, so that the advertisement investment cost is saved.
Amazon's personalized stores and google's personalized searches are good examples of personalized services. The amazon shopping mall may recommend corresponding goods to the user according to the browsing record and the purchasing record of the user to prompt the user to consume, or recommend goods to the user by analyzing the purchasing behavior of other users who have purchased the same or similar products.
The achievement obtained by the personalized search of the google is also attractive, and the personalized search of the google can return personalized results according to the search history and the search keywords of the user so as to meet the requirements of the user.
With the development of information technology and the popularization of smart phones, the application of smart phones is exponentially increased, and more services are provided based on positions. These location-based web applications permit users to publish their geographic location information, search nearby people, share personal experiences, etc., while at the same time the web applications recommend to the users featured stores at the user's location, people or things that the users are interested in, which need to combine the user's geographic location information, hobbies of interest, and personal basic attribute information. However, the information is generally regarded as user privacy information, and is difficult to obtain by many network application companies. Although some network applications require users to fill in relevant information such as gender, birth year and month, education degree and the like when the users register, the information is sensitive to the users, so that many users cannot fill in the relevant information or fill in wrong information at all, and the unreal information has negative effects on personalized recommendation. The actual situation is that most users do not fill in relevant basic attribute information at the time of registration, and the basic attribute information of the users is unknown to the mobile phone application.
The invention patent ZL201610486432.7 granted by China discloses a mobile user gender prediction method based on an installation package list, which comprises the following steps: acquiring a user equipment number with a gender label; screening users corresponding to the user equipment numbers with the gender labels from an installation list library, and acquiring installation package lists of the users; removing users with the number of the installation package lists smaller than M or larger than N; converting the installation package list information into a feature data set; randomly dividing the characteristic data set into a training set and a testing set according to a set proportion; according to the data of the training set, training by using a GBDT model, and then verifying through a test set to obtain a user gender prediction model; and acquiring the users without the gender labels and the installation list thereof from the installation package list library, performing the same characteristic conversion and predicting by using the trained gender model.
The Chinese patent application CN201611127122.2 proposes a method for predicting the gender of a user based on the internet surfing behavior of a mobile phone, which counts the times of clicking each APP by the user within a period of time; arranging the statistical data into a matrix form; performing dimensionality reduction processing on the matrix; dividing the processed data into a training data set and a testing data set, and training a prediction model by using the training data set; the test data set is used to validate the predictive model and calculate accuracy. The method is simple and easy to implement and has high accuracy. The gender of the user is predicted according to the number of times of the APP used by the user, and support is provided for performing related personalized service recommendation according to the preferences of the users with different genders.
However, although various models and machine learning algorithms for predicting the gender of the user exist in the prior art, the inventor finds that most of the prior art focuses on the models and the algorithms themselves, and does not perform matching processing on the algorithm or sample data used by the model, so that the sample data actually used has a large false positive problem and an imbalance problem, and the accuracy of the modeling and prediction effects itself is reduced.
Disclosure of Invention
In order to solve the technical problem, the invention provides a method and a system for predicting the gender of a mobile terminal user based on a full-connection neural network. The method comprises the steps of obtaining sample data of the mobile terminal, carrying out feature classification to obtain a continuity feature and a discrete feature, respectively processing the continuity feature and the discrete feature, expressing the continuity feature and the discrete feature by adopting one-shot codes to obtain a one-shot sample feature, carrying out Embelling mapping on all the one-shot sample features, constructing and training a fully-connected neural network model based on the sample features after the Embelling mapping, adopting the trained fully-connected neural network model, inputting user features of the mobile terminal, predicting the gender of a user and the like. The invention also discloses a prediction system comprising the full-connection neural network model, which is used for the gender prediction method of the mobile terminal user. The technical scheme of the invention can ensure that the modeling process is more complete and the result is more accurate.
In particular, in a first aspect of the invention, a method for gender prediction of a mobile terminal user is provided, which is implemented based on a fully-connected neural network.
More specifically, the method includes the following steps S1-S6:
s1: obtaining sample data of a mobile terminal;
s2: carrying out feature classification on the sample data to obtain a continuity feature and a discrete feature;
s3: after the continuity characteristic and the discrete characteristic are respectively processed differently, one-shot coding expression is adopted to obtain one-shot sample characteristics;
s4: carrying out Embedding mapping on all the one-hot sample characteristics;
s5: constructing a fully-connected neural network model based on the sample characteristics after Embedding mapping and training;
s6: and inputting the characteristics of the mobile terminal user by adopting a trained fully-connected neural network model, and predicting the gender of the user.
In the invention, the gender and age of the user are predicted mainly through the mobile phone installation package list of the user. The current data includes a list of user mobile phone installation packages, application types corresponding to each installation package, mobile phone brands, models under the mobile phone brands, mobile phone screen length and width, mobile phone systems, mobile phone system versions, gender and age of the user, and the like. The owned data is not limited to the above, but must include a list of user mobile phone installation packages, application categories corresponding to each installation package, and the gender and age of the user.
As one of the inventive technical means of the present invention, in order to solve the problem that sample data itself has a large false positive and an unbalanced problem mentioned in the prior art, the step S3 performs different processes on the continuity feature and the discrete feature respectively, which specifically includes:
carrying out segmentation processing on the continuity characteristics, carrying out visual analysis to obtain continuous data segments and discrete data segments, and identifying a first null bit segment between the adjacent discrete data segments and the continuous data segments and a second null bit segment between the adjacent discrete data segments;
if the number of the first null bit segments is less than a first threshold, populating the first null bit segments with a mode of the discrete data segments;
and if the number of the second null bit segments is larger than a second threshold value, deleting the discrete data segments corresponding to the second null bit segments.
As another innovative advantage of the present invention, any two adjacent layers of the fully-connected neural network model, namely any neuron node of the n-1 layer, are connected to all neuron nodes of the n layer, and when each neuron node of the n layer performs calculation, the input value of the activation function is the weight of all neuron nodes of the n-1 layer.
Therefore, in another aspect of the present invention, a prediction system comprising a fully connected neural network model is provided for the aforementioned method for predicting gender of a mobile terminal user.
In the present invention, the fully-connected neural network is composed of two parts, linear and non-linear respectively. For an input vector x ═ x1, x2, …, xn]TAfter passing through the hidden layer, a linear output vector z ═ z1, z2, …, zn is obtained]TThe value of the output vector is mainly determined by the weight vector w and the bias vector b, which are important parameters that the neural network needs to learn, that is to say
z＝w×x+b (1)
After obtaining the linear output vector, the linear output vector is converted into a nonlinear vector according to the requirement, and the output vector is converted through an activation function to obtain the output vector of the hidden layer
h＝[h1,h2,…,hn]T，
Or output vectors of the output layer, i.e.
y＝[y1,y2,…,yn]T。
In the present invention, the weight and bias are important parameters for the connection between each lower layer neuron and all upper layer neurons in the fully-connected neural network. Meanwhile, the activation function is required to participate together, and the learned data features are associated with the targets.
The method is mainly realized by alternately constructing the fully-connected deep neural network through an affinity layer, an activation function ReLU and a Softmax-with-Loss layer.
Further advantages of the invention will be apparent in the detailed description section in conjunction with the drawings attached hereto.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings needed in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art to obtain other drawings without creative efforts.
FIG. 1 is a flowchart of a method for predicting gender of a mobile terminal user according to an embodiment of the present invention
FIGS. 2-3 are schematic diagrams of the corresponding steps of the method of FIG. 1
FIG. 4 is a schematic diagram of the structure of a fully-connected deep neural network used in the present invention
Detailed Description
The invention is further described with reference to the following drawings and detailed description.
Referring to fig. 1, a main flowchart of a method for predicting gender of a mobile terminal user according to an embodiment of the present invention is shown.
Fig. 1 shows a gender prediction method for a mobile terminal user, which is implemented based on a fully-connected neural network.
The method of FIG. 1 includes the following steps S1-S6:
s1: obtaining sample data of a mobile terminal;
s2: carrying out feature classification on the sample data to obtain a continuity feature and a discrete feature;
s3: after the continuity characteristic and the discrete characteristic are respectively processed differently, one-shot coding expression is adopted to obtain one-shot sample characteristics;
s4: carrying out Embedding mapping on all the one-hot sample characteristics;
s5: constructing a fully-connected neural network model based on the sample characteristics after Embedding mapping and training;
s6: and inputting the characteristics of the mobile terminal user by adopting a trained fully-connected neural network model, and predicting the gender of the user.
First, it should be noted that how to train the neural network model is not the key point of the present invention, and various known model training methods exist in the prior art, and the present invention is not described in detail herein. The key point of the method is to acquire sample data and process the sample for use in the subsequent modeling process, so that the modeling process is optimized and the accuracy of the model is improved.
In particular, on the basis of fig. 1, see fig. 2-3.
The obtaining of the mobile terminal sample data in step S1 specifically includes:
the method comprises the steps of obtaining a mobile terminal installation package list, application types corresponding to each installation package, mobile terminal brands, models under the mobile terminal brands, mobile terminal screen sizes, mobile terminal operating systems, mobile terminal system versions and gender and age information of users marked by the mobile terminals.
In this embodiment, the continuity characteristics include a screen size of the mobile terminal, a system version of the mobile terminal.
The discrete type characteristics comprise a mobile terminal installation package list, application types corresponding to each installation package, mobile terminal brands, models under the mobile terminal brands, mobile terminal operating systems and gender and age information of users marked by the mobile terminals.
In step S3, the processing of the continuity feature and the discrete feature is respectively different, and specifically includes:
carrying out segmentation processing on the continuity characteristics, carrying out visual analysis to obtain continuous data segments and discrete data segments, and identifying a first null bit segment between the adjacent discrete data segments and the continuous data segments and a second null bit segment between the adjacent discrete data segments;
if the number of the first null bit segments is less than a first threshold, populating the first null bit segments with a mode of the discrete data segments;
and if the number of the second null bit segments is larger than a second threshold value, deleting the discrete data segments corresponding to the second null bit segments.
For example, for a continuity feature, for example, a mobile terminal system version, the feature may be a feature such as android 4.1.2 or android 4.1.3; the size of the screen of the mobile terminal can be 4.4 inches, 4.5 inches, 4.7 inches and the like; for such a feature, the feature is regarded as a continuity feature in the present invention, and may be subjected to a pre-processing encoding process, for example, a mobile terminal system version may be defined as 412/413/414 … …, etc.;
taking the version data of the mobile terminal system as an example, the continuity features are processed in a segmented manner, and visual analysis is performed to obtain continuous data segments and discrete data segments, as follows:
assume that the mobile terminal system version data contained in the sample data is as follows:
412, 5;
413: 0;
4, 414: 4;
415, 5;
416: 0;
417: 0;
418: 1;
419: 0.
The continuous data segments are 414 and 415, and the discrete data segments are 412-;
wherein 413 is a first null bit segment between an adjacent discrete data segment and the continuous data segment; and 416-417 is the second null bit segment between adjacent discrete data segments.
Of course, the above-mentioned dividing and encoding method is only exemplary, and those skilled in the art may also adopt other dividing and encoding methods, as long as the continuity characteristic can be segmented, and visual analysis is performed to obtain the continuous data segment and the discrete data segment.
In fig. 2 to fig. 3, the step S3 performs different processes on the continuity feature and the discrete feature, which specifically includes:
and performing one-hot representation on the continuity characteristics after the segmentation processing and the visualization analysis.
For the discrete features, one-hot is used directly.
In the present embodiment, one hot encoding is a process of converting class variables into a form that is easily utilized by a machine learning algorithm.
For example, assume that the characteristics of "flowers" might take on the values daffodil (narcissus), lily (lily), rose (rose). one hot encoding converts it into three features: is _ daffodil, is _ lily, is _ rose, these features are all binary.
One-Hot encoding, also known as One-bit-efficient encoding, mainly uses a bit state register to encode each state, each state being represented by its own independent register bit and having only One bit active at any time.
In the actual task of machine learning, features are sometimes not always continuous values, but may be classified values, such as gender, which can be classified as "male" and "female". In a machine learning task, we usually need to digitize such features, as in the following example:
there are three characteristic attributes:
sex: [ "male", "female" ]
The region: [ "Europe", "US", "Asia" ]
The browser: [ "Firefox", "Chrome", "Safari", "Internet Explorer" ]
For a certain sample, such as [ "male", "US", "Internet Explorer" ], we need to digitize the feature of this classification value, and most directly, we can adopt a serialization mode: [0,1,3]. But such feature processing cannot be directly put into a machine learning algorithm.
For the above problem, gender attribute is two-dimensional, and similarly, region is three-dimensional, and browser is thought, so we can use One-Hot coding method to code the samples "[" male "," US "," Internet Explorer "]", where "male" corresponds to [1, 0], similarly, "US" corresponds to [0,1, 0], and "Internet Explorer" corresponds to [0,0,0,1 ]. The result of the full feature digitization is: [1,0,0,1,0,0,0,0,1]. One consequence of this is that the data becomes very sparse.
However, in the embodiment of the present invention, since the filling and deleting of the first null bit segment and the second null bit segment are performed before this, the problem of data sparseness can be effectively avoided.
The step S4 performs embed mapping on all the one-hot sample features, which specifically includes:
and converting the mobile terminal installation package list into a TF-IDF (word frequency-inverse text frequency) matrix for representation, and performing TF-IDF weighted average on Embedding corresponding to the mobile terminal installation package.
Reference is next made to fig. 4.
Fig. 4 provides a fully-connected neural network used in a prediction system including a fully-connected neural network model, which is used in the aforementioned method for predicting gender of a mobile terminal user.
In fig. 4, the fully-connected neural network is composed of two parts, linear and non-linear, respectively. For an input vector x ═ x1, x2, …, xn]TAfter passing through the hidden layer, a linear output vector z ═ z1, z2, …, zn is obtained]TThe value of the output vector is mainly determined by the weight vector w and the bias vector b, which are important parameters that the neural network needs to learn, that is to say
z＝w×x+b (1)
After obtaining the linear output vector, the linear output vector is converted into a nonlinear vector according to the requirement, and the output vector is converted through an activation function to obtain the output vector of the hidden layer
h＝[h1,h2,…,hn]T，
Or output vectors of the output layer, i.e.
y＝[y1,y2,…,yn]T。
In the present invention, the weight and bias are important parameters for the connection between each lower layer neuron and all upper layer neurons in the fully-connected neural network. Meanwhile, the activation function is required to participate together, and the learned data features are associated with the targets.
The method is mainly realized by alternately constructing the fully-connected deep neural network through an affinity layer, an activation function ReLU and a Softmax-with-Loss layer.
Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that changes, modifications, substitutions and alterations can be made in these embodiments without departing from the principles and spirit of the invention, the scope of which is defined in the appended claims and their equivalents.
Claims (10)
1. A gender prediction method for a mobile terminal user is realized based on a fully-connected neural network, and is characterized by comprising the following steps:
s1: obtaining sample data of a mobile terminal;
s2: carrying out feature classification on the sample data to obtain a continuity feature and a discrete feature;
s3: after the continuity characteristic and the discrete characteristic are respectively processed differently, one-shot coding expression is adopted to obtain one-shot sample characteristics;
s4: carrying out Embedding mapping on all the one-hot sample characteristics;
s5: constructing a fully-connected neural network model based on the sample characteristics after Embedding mapping and training;
s6: and inputting the characteristics of the mobile terminal user by adopting a trained fully-connected neural network model, and predicting the gender of the user.
2. The method for gender prediction for a mobile terminal user as claimed in claim 1 wherein:
the obtaining of the mobile terminal sample data in step S1 specifically includes:
the method comprises the steps of obtaining a mobile terminal installation package list, application types corresponding to each installation package, mobile terminal brands, models under the mobile terminal brands, mobile terminal screen sizes, mobile terminal operating systems, mobile terminal system versions and gender and age information of users marked by the mobile terminals.
3. The method for gender prediction for a mobile terminal user as claimed in claim 1 wherein:
in step S3, the processing of the continuity feature and the discrete feature is respectively different, and specifically includes:
carrying out segmentation processing on the continuity characteristics, carrying out visual analysis to obtain continuous data segments and discrete data segments, and identifying a first null bit segment between the adjacent discrete data segments and the continuous data segments and a second null bit segment between the adjacent discrete data segments;
if the number of the first null bit segments is less than a first threshold, populating the first null bit segments with a mode of the discrete data segments;
and if the number of the second null bit segments is larger than a second threshold value, deleting the discrete data segments corresponding to the second null bit segments.
4. A method for gender prediction for a mobile terminal user as claimed in claim 3 wherein:
in step S3, the processing of the continuity feature and the discrete feature is respectively different, and specifically includes:
and performing one-hot representation on the continuity characteristics after the segmentation processing and the visualization analysis.
5. A method for gender prediction for a mobile terminal user according to claim 1 or 3, characterized by:
the continuity characteristics comprise the screen size of the mobile terminal and the system version of the mobile terminal.
6. The method for gender prediction for a mobile terminal user as claimed in claim 1 wherein:
the discrete type characteristics comprise a mobile terminal installation package list, application types corresponding to each installation package, mobile terminal brands, models under the mobile terminal brands, mobile terminal operating systems and gender and age information of users marked by the mobile terminals.
7. The method of mobile terminal user gender prediction as claimed in claim 6, wherein:
for the discrete features, one-hot is used directly.
8. The method of mobile terminal user gender prediction according to claim 1 or 6, characterized by:
the step S4 performs embed mapping on all the one-hot sample features, which specifically includes:
and converting the mobile terminal installation package list into a TF-IDF (word frequency-inverse text frequency) matrix for representation, and performing TF-IDF weighted average on Embedding corresponding to the mobile terminal installation package.
9. The method of any preceding claim, wherein the mobile terminal user gender prediction method is:
any two adjacent layers of the fully-connected neural network model, namely any neuron node of the n-1 layer, are connected with all neuron nodes of the n layer, and when each neuron node of the n layer is calculated, the input value of the activation function is the weight of all the neuron nodes of the n-1 layer.
10. A prediction system comprising a fully connected neural network model for implementing the mobile terminal user gender prediction method of any of claims 1-9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010749181.3A CN111898738A (en) | 2020-07-30 | 2020-07-30 | Mobile terminal user gender prediction method and system based on full-connection neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010749181.3A CN111898738A (en) | 2020-07-30 | 2020-07-30 | Mobile terminal user gender prediction method and system based on full-connection neural network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN111898738A true CN111898738A (en) | 2020-11-06 |
Family
ID=73182628
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010749181.3A Pending CN111898738A (en) | 2020-07-30 | 2020-07-30 | Mobile terminal user gender prediction method and system based on full-connection neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111898738A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113486947A (en) * | 2021-07-01 | 2021-10-08 | 有米科技股份有限公司 | Garment commodity gender classification method and device based on size data |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106126597A (en) * | 2016-06-20 | 2016-11-16 | 乐视控股（北京）有限公司 | User property Forecasting Methodology and device |
| CN106203473A (en) * | 2016-06-24 | 2016-12-07 | 有米科技股份有限公司 | A kind of mobile subscriber's gender prediction's method based on installation kit list |
| CN110705592A (en) * | 2019-09-03 | 2020-01-17 | 平安科技（深圳）有限公司 | Classification model training method, device, equipment and computer readable storage medium |
| CN111125530A (en) * | 2019-12-24 | 2020-05-08 | 中山大学 | Information flow recommendation method based on multi-type feature deep learning |
| CN111339355A (en) * | 2020-05-21 | 2020-06-26 | 北京搜狐新媒体信息技术有限公司 | Video recommendation method and system |
-
2020
- 2020-07-30 CN CN202010749181.3A patent/CN111898738A/en active Pending
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106126597A (en) * | 2016-06-20 | 2016-11-16 | 乐视控股（北京）有限公司 | User property Forecasting Methodology and device |
| WO2017219548A1 (en) * | 2016-06-20 | 2017-12-28 | 乐视控股（北京）有限公司 | Method and device for predicting user attributes |
| CN106203473A (en) * | 2016-06-24 | 2016-12-07 | 有米科技股份有限公司 | A kind of mobile subscriber's gender prediction's method based on installation kit list |
| CN110705592A (en) * | 2019-09-03 | 2020-01-17 | 平安科技（深圳）有限公司 | Classification model training method, device, equipment and computer readable storage medium |
| CN111125530A (en) * | 2019-12-24 | 2020-05-08 | 中山大学 | Information flow recommendation method based on multi-type feature deep learning |
| CN111339355A (en) * | 2020-05-21 | 2020-06-26 | 北京搜狐新媒体信息技术有限公司 | Video recommendation method and system |
Non-Patent Citations (1)
| Title |
|---|
| 杨波: "森林病虫害数据可视分析方法研究", 《中国博士学位论文全文数据库 农业科技辑》, pages 046 - 26 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113486947A (en) * | 2021-07-01 | 2021-10-08 | 有米科技股份有限公司 | Garment commodity gender classification method and device based on size data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11301761B2 (en) | Behavioral prediction for targeted end users | |
| CN111523976B (en) | Commodity recommendation method and device, electronic equipment and storage medium | |
| CN111339415B (en) | Click rate prediction method and device based on multi-interactive attention network | |
| CN104765873A (en) | Method and device for determining similarity among users | |
| CN111008335B (en) | Information processing method, device, equipment and storage medium | |
| CN113379449B (en) | Multimedia resource recall method and device, electronic equipment and storage medium | |
| CN109189922B (en) | Comment evaluation model training method and device | |
| Abinaya et al. | Enhancing top-N recommendation using stacked autoencoder in context-aware recommender system | |
| CN114861050A (en) | Feature fusion recommendation method and system based on neural network | |
| Liu et al. | Mobile E‐Commerce Information System Based on Industry Cluster under Edge Computing | |
| Chkoniya | Challenges in decoding consumer behavior with data science | |
| CN111898738A (en) | Mobile terminal user gender prediction method and system based on full-connection neural network | |
| CN116777528A (en) | Commodity information recommendation method and device, computer equipment and storage medium | |
| CN114429384B (en) | Intelligent product recommendation method and system based on e-commerce platform | |
| Hao et al. | User Behavior Prediction Based on DCGAN: The Case of Sina Weibo | |
| CN115169583A (en) | Training method and device of user behavior prediction system | |
| CN111400567B (en) | AI-based user data processing method, device and system | |
| CN111914927A (en) | Mobile app user gender identification method and system for optimizing data imbalance state | |
| Meng et al. | Forecasting service quality dynamics based on integrated method with GM (1, 1) model and Markov chain | |
| CN113888216A (en) | Advertisement information pushing method and device, electronic equipment and storage medium | |
| Maheswari | The Transformative Power of AI in Marketing FMCG | |
| CN116911913B (en) | Method and device for predicting interaction result | |
| Chen et al. | Tagtag: a novel framework for service tags recommendation and missing tag prediction | |
| CN115170173A (en) | Demand forecasting method, recommendation method, computer device and storage medium | |
| CN118349738A (en) | Recommendation method and device, electronic equipment and readable storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20201106 |