CN112634026A - Credit fraud identification method based on user page operation behavior - Google Patents
Credit fraud identification method based on user page operation behavior Download PDFInfo
- Publication number
- CN112634026A CN112634026A CN202011600420.5A CN202011600420A CN112634026A CN 112634026 A CN112634026 A CN 112634026A CN 202011600420 A CN202011600420 A CN 202011600420A CN 112634026 A CN112634026 A CN 112634026A
- Authority
- CN
- China
- Prior art keywords
- fraud
- operation behavior
- user
- characteristic
- credit
- 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
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/03—Credit; Loans; Processing thereof
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- 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
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q20/00—Payment architectures, schemes or protocols
- G06Q20/38—Payment protocols; Details thereof
- G06Q20/40—Authorisation, e.g. identification of payer or payee, verification of customer or shop credentials; Review and approval of payers, e.g. check credit lines or negative lists
- G06Q20/401—Transaction verification
- G06Q20/4016—Transaction verification involving fraud or risk level assessment in transaction processing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Accounting & Taxation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Finance (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Biophysics (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Computational Linguistics (AREA)
- Strategic Management (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Technology Law (AREA)
- Marketing (AREA)
- Economics (AREA)
- Development Economics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Security & Cryptography (AREA)
- Storage Device Security (AREA)
Abstract
The invention relates to a credit fraud identification method based on user page operation behaviors, which comprises the following steps: step A, collecting and initializing operation behaviors of a user on a system page; step B, establishing a fraud recognition model, comprising the following steps: an input module: is composed of at least two fully connected network layers; a feature embedding module: the system is composed of at least two GRU network layers and used for feature embedding; an output module: the system is composed of at least two fully-connected layers and outputs final characteristics; and C, collecting the operation behavior sequence of the user on the system page, and finally outputting the credit fraud probability value of the user through a fraud identification model. The method can effectively extract and capture the recent operation behavior sequence of the user from the operation behaviors of the system page of the user, and identifies the cheating behavior through the characteristics of the unstructured data such as the dynamic operation behavior sequence, so that the accuracy of identifying credit cheating is obviously improved compared with the existing static data.
Description
Technical Field
The invention relates to a method for training a machine learning model in a dynamic sample mode and identifying user credit fraud, in particular to a credit fraud identification method based on user page operation behaviors.
Background
With the increasing daily life of people involving banks, finance and other fields, there is an increasing illegal act of using personal credit to cheat the operation subjects in these fields. Particularly, the method is obvious in the aspect of personal loan, and an individual user forges own high credit value through operations on the bank APP to induce the bank to loan the credit value, or network black products, telecom fraud, black intermediaries and the like utilize the identity of the individual or induce the individual user to cheat the bank loan through data counterfeiting, identity counterfeiting and the like.
At present, the identification of the personal credit fraud is mainly realized by modeling static data such as credit investigation data of a user, application materials submitted by the user and the like, and the static data cannot well capture dynamic characteristics shown by the fraudulent user during application, so that the current identification mode has a larger improvement space.
Disclosure of Invention
The invention provides a credit fraud identification method based on user page operation behaviors, which is used for more accurately identifying whether a user has credit fraud behaviors when applying for banking business.
The invention discloses a credit fraud identification method based on user page operation behaviors, which comprises the following steps:
a, collecting operation behaviors of a user on a system page on an operation device in batch in a specified time period through a processor, initializing collected event types and numerical variables of the operation behaviors through the processor, and storing the operation behaviors into a first storage area of a disk memory;
b, constructing a fraud identification model in a second storage area of the disk memory, wherein the structure of the fraud identification model comprises the following steps:
an input module: the system comprises at least two fully-connected network layers, a characteristic embedding module and a data processing module, wherein the fully-connected network layers are used for mapping data stored in a first storage area of a disk memory into corresponding characteristic dimensions, forming an operation behavior characteristic sequence according to time sequence by each characteristic dimension and outputting the operation behavior characteristic sequence to the characteristic embedding module;
a feature embedding module: the characteristic dimension reduction method is characterized by comprising at least two GRU (gated Current Unit network) network layers and is used for characteristic embedding, dimension reduction is carried out on each characteristic dimension in an operation behavior characteristic sequence output by an input module into a characteristic vector with a fixed size, processing and calculation (such as distance calculation) are facilitated, and the dimension reduction mode can be analogized to a full connection layer and data are converted into characteristic representation with a fixed size through weighting matrix calculation of an embedding layer. Extracting corresponding data characteristics from the operation behavior characteristic sequence after dimensionality reduction according to a preset scene, wherein the data characteristics are used for showing a specific mode which is adapted to the preset scene in the operation behavior characteristic sequence, and finally outputting a result to an output module;
an output module: the system comprises at least two full connection layers and a characteristic embedding module, wherein the full connection layers are used for mapping the output of the characteristic embedding module into final characteristics to be output; the output module and the input module are different from the neuron in number and layer number in function;
and step C, collecting an operation behavior sequence of a user on a system page on the operation equipment within a period of time, initializing the collected event type and numerical variable of the operation behavior through a processor, inputting the operation behavior sequence into a fraud recognition model in a second storage area of the disk memory, and finally outputting the credit fraud probability value of the user through the fraud recognition model.
The method can effectively extract and capture the recent operation behavior sequence of the user from the operation behaviors of the system page of the user, and identifies the fraudulent behavior through the characteristics of the unstructured data such as the dynamic operation behavior sequence, so that the identification accuracy is obviously improved compared with the existing static data.
Preferably, in the step B, after the fraud recognition model is constructed, the fraud recognition model is optimized, including:
b1, obtaining a fraud label of the user through the step B, wherein the value of the fraud label is 0 or 1;
b2, randomly initializing parameters of the fraud recognition model obtained in the step B; because a neural network model has a plurality of parameters to be optimized, the optimal values of the parameters are not known before optimization, all the parameters are initialized randomly, and then the parameters are iterated to the optimal values by using subsequent steps;
step B3, calculating a cross-entropy loss function by using the output of the fraud identification model initialized in the step B2 and the fraud label of the user;
step B4. optimizing parameters of the fraud identification model by a gradient descent method to minimize the loss function;
step B5. repeats step B4 until the value of the loss function no longer becomes small, resulting in an optimal fraud identification model.
Among them, the principle of the gradient descent algorithm can be referred to as:
https://baike.***.com/item/％E6％A2％AF％E5％BA％A6％E4％B8％8B％E9％99％8D/4864937fr＝aladdin。
specifically, the initialization processing in step a includes encoding the collected operation behavior as a numerical tensor. This process is similar to the process of encoding text when processing natural language.
Specifically, the initialization processing in step a includes normalizing and normalizing the collected numerical variables. For example, for vector X, normalization refers to transforming vector X by (X-mean (X))/std (X), where mean (X) function is the average of vector X and std (X) function is the standard deviation of vector X; the normalization process includes transforming the vector X by (X-min (X))/(max (X) -min (X)), where max (X) is the function of maximizing the vector X and min (X) is the function of minimizing the vector X.
The method can effectively extract and capture the recent operation behavior sequence of the user from the operation behaviors of the system page of the user, and identifies the cheating behavior through the characteristics of the unstructured data such as the dynamic operation behavior sequence, so that the accuracy of identifying credit cheating is obviously improved compared with the existing static data.
The present invention will be described in further detail with reference to the following examples. This should not be understood as limiting the scope of the above-described subject matter of the present invention to the following examples. Various substitutions and alterations according to the general knowledge and conventional practice in the art are intended to be included within the scope of the present invention without departing from the technical spirit of the present invention as described above.
Drawings
FIG. 1 is a flow chart of a credit fraud identification method based on user page operation behavior according to the present invention.
Detailed Description
As shown in FIG. 1, the credit fraud identification method based on the user page operation behavior of the present invention includes:
a, operating behaviors of a user on a system page on operating equipment in a specified time period are collected in batches through a processor, and the collected event types and numerical variables of the operating behaviors are initialized through the processor, wherein the collected operating behaviors are coded into numerical tensors. This process is similar to the process of encoding text when processing natural language. And normalizing and processing the collected numerical variables. For example, for vector X, normalization refers to transforming vector X by (X-mean (X))/std (X), where mean (X) function is the average of vector X and std (X) function is the standard deviation of vector X; the normalization process includes transforming the vector X by (X-min (X))/(max (X) -min (X)), where max (X) is the function of maximizing the vector X and min (X) is the function of minimizing the vector X.
B, constructing a fraud identification model in a second storage area of the disk memory, wherein the structure of the fraud identification model comprises the following steps:
an input module: the system comprises at least two fully-connected network layers, and is used for mapping data stored in a first storage area of a disk memory into corresponding characteristic dimensions, forming an operation behavior characteristic sequence according to a time sequence by using each characteristic dimension, and outputting the operation behavior characteristic sequence to a characteristic embedding module.
A feature embedding module: the characteristic dimension reduction method is characterized by comprising at least two GRU (gated Current Unit network) network layers and is used for characteristic embedding, dimension reduction is carried out on each characteristic dimension in an operation behavior characteristic sequence output by an input module into a characteristic vector with a fixed size, processing and calculation (such as distance calculation) are facilitated, and the dimension reduction mode can be analogized to a full connection layer and data are converted into characteristic representation with a fixed size through weighting matrix calculation of an embedding layer. And then extracting corresponding data characteristics from the operation behavior characteristic sequence after dimension reduction according to a preset scene, wherein the data characteristics are used for showing a specific mode which is adapted to the preset scene in the operation behavior characteristic sequence, and finally outputting the result to an output module.
An output module: the system is composed of at least two fully-connected layers and used for mapping the output of the feature embedding module into a final feature to be output. The output module and the input module are different in function and the number and the layer number of the neurons.
And after the construction of the fraud recognition model is completed, optimizing the fraud recognition model, wherein the method comprises the following steps:
and B1, obtaining a fraud label y of the user through the step B, wherein the value of the fraud label y is 0 or 1.
And B2, randomly initializing parameters of the fraud recognition model obtained in the step B. Because a neural network model has a plurality of parameters to be optimized, the optimal values of the parameters are not known before optimization, all the parameters are initialized randomly, and then the parameters are iterated to the optimal values by using subsequent steps.
Step B3, calculating a cross-entropy loss function by using the output of the fraud identification model initialized in the step B2 and the fraud label y of the user: loss is crossbar (y, output), where output is the output of the initialized fraud recognition model, i.e., the probability of credit fraud for the user that the fraud recognition model outputs for the first time.
Step B4. optimizes the parameters of the fraud identification model by minimizing the loss function by a gradient descent method, so that the probability represented by output is closer to the fraud label y.
Step B5. repeats step B4 until the value of the loss function no longer becomes small, resulting in an optimal fraud identification model.
The existing optimization mode of gradient descent of the loss function can be adopted, and the general idea of the optimization mode is as follows:
the optimized objective function is a loss function that minimizes cross entropy:
Min Loss＝Crossentropy(y,output)，
calculating the partial derivative of the objective function to the bias term parameter:
where W and b are both bias term parameters in the machine learning model,
and
the partial derivatives of W and b are shown separately for the loss function. A machine learning model usually contains three necessary parts, the first part is the variable X to be predicted, the second part is the algorithm of the model itself, and the third part is the target Z to be predicted, for example, Z ═ WX + b for the bias term parameter.
Parameters W and b are then updated:
W0and b0Are all initial values of the bias term parameters obtained after randomly initializing the model parameters.
Where epsilon is the learning rate, this process is repeated several times until Loss no longer diminishes (converges), i.e., the optimization process is completed. For a more specific description reference may be made to:
https://baike.***.com/item/％E6％A2％AF％E5％BA％A6％E4％B8％8B％E9％99％8D/4864937fr＝aladdin。
and step C, collecting an operation behavior sequence of a user on a system page on the operation equipment within a period of time, initializing the collected event type and numerical variable of the operation behavior through a processor, inputting the operation behavior sequence into a fraud recognition model in a second storage area of the disk memory, and finally outputting the credit fraud probability value of the user through the fraud recognition model.
For example, a user has recently browsed an APP page and performs a series of operations on the page, including login, browsing, password modification, identity authentication, and the like. These operations, in chronological order, constitute a sequence of operational behaviors. The operation behavior sequence is coded by the method (the coded operation behavior sequence can be directly processed by a numerical tensor model, and is similar to the process of coding characters in natural language processing), the coded operation behavior sequence data is input to the fraud recognition model, an initial probability value is finally output by the fraud recognition model, and after the optimization process, the final probability value output by the fraud recognition model is the probability of credit fraud of the user, so that the purpose of recognizing the credit fraud is realized.
Claims (4)
1. The credit fraud identification method based on the user page operation behavior is characterized by comprising the following steps:
a, collecting operation behaviors of a user on a system page on an operation device in batch in a specified time period through a processor, initializing collected event types and numerical variables of the operation behaviors through the processor, and storing the operation behaviors into a first storage area of a disk memory;
b, constructing a fraud identification model in a second storage area of the disk memory, wherein the structure of the fraud identification model comprises the following steps:
an input module: the system comprises at least two fully-connected network layers, a characteristic embedding module and a data processing module, wherein the fully-connected network layers are used for mapping data stored in a first storage area of a disk memory into corresponding characteristic dimensions, forming an operation behavior characteristic sequence according to time sequence by each characteristic dimension and outputting the operation behavior characteristic sequence to the characteristic embedding module;
a feature embedding module: the system comprises at least two GRU network layers, is used for embedding characteristics, reduces dimensions of each characteristic latitude in an operation behavior characteristic sequence output by an input module into a characteristic vector with a fixed size, is convenient to process and calculate, extracts corresponding data characteristics from the operation behavior characteristic sequence after dimension reduction according to a preset scene, is used for representing a specific mode which is adapted to the preset scene in the operation behavior characteristic sequence, and finally outputs a result to an output module;
an output module: the system comprises at least two full connection layers and a characteristic embedding module, wherein the full connection layers are used for mapping the output of the characteristic embedding module into final characteristics to be output;
and step C, collecting an operation behavior sequence of a user on a system page on the operation equipment within a period of time, initializing the collected event type and numerical variable of the operation behavior through a processor, inputting the operation behavior sequence into a fraud recognition model in a second storage area of the disk memory, and finally outputting the credit fraud probability value of the user through the fraud recognition model.
2. The method for identifying credit fraud based on user page operation behavior as claimed in claim 1, characterized by: in the step B, after the construction of the fraud recognition model is completed, the fraud recognition model is optimized, and the method comprises the following steps:
b1, obtaining a fraud label of the user through the step B, wherein the value of the fraud label is 0 or 1;
b2, randomly initializing parameters of the fraud recognition model obtained in the step B;
step B3, calculating a cross-entropy loss function by using the output of the fraud identification model initialized in the step B2 and the fraud label of the user;
step B4. optimizing parameters of the fraud identification model by a gradient descent method to minimize the loss function;
step B5. repeats step B4 until the value of the loss function no longer becomes small, resulting in an optimal fraud identification model.
3. A credit fraud identification method based on user page operation behavior according to claim 1 or 2, characterized by: the initialization process of step a includes encoding the collected operation behavior as a numerical tensor.
4. A credit fraud identification method based on user page operation behavior according to claim 1 or 2, characterized by: the initialization process of step a includes normalizing and normalizing the collected numerical variables.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011600420.5A CN112634026A (en) | 2020-12-30 | 2020-12-30 | Credit fraud identification method based on user page operation behavior |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011600420.5A CN112634026A (en) | 2020-12-30 | 2020-12-30 | Credit fraud identification method based on user page operation behavior |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112634026A true CN112634026A (en) | 2021-04-09 |
Family
ID=75287541
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011600420.5A Pending CN112634026A (en) | 2020-12-30 | 2020-12-30 | Credit fraud identification method based on user page operation behavior |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112634026A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113240422A (en) * | 2021-07-09 | 2021-08-10 | 天聚地合(苏州)数据股份有限公司 | Financial product transfer method based on block chain and related device |
| CN114493826A (en) * | 2021-12-22 | 2022-05-13 | 四川新网银行股份有限公司 | Personal credit assessment scoring method based on neural network |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108428132A (en) * | 2018-03-15 | 2018-08-21 | 阿里巴巴集团控股有限公司 | Fraudulent trading recognition methods, device, server and storage medium |
| CN109345260A (en) * | 2018-10-09 | 2019-02-15 | 北京芯盾时代科技有限公司 | A kind of fraud detection model training method and device and fraud detection method and device |
| CN109410036A (en) * | 2018-10-09 | 2019-03-01 | 北京芯盾时代科技有限公司 | A kind of fraud detection model training method and device and fraud detection method and device |
| CN109544190A (en) * | 2018-11-28 | 2019-03-29 | 北京芯盾时代科技有限公司 | A kind of fraud identification model training method, fraud recognition methods and device |
| CN110969441A (en) * | 2019-12-23 | 2020-04-07 | 集奥聚合(北京)人工智能科技有限公司 | Anti-fraud model processing method and device based on deep learning |
-
2020
- 2020-12-30 CN CN202011600420.5A patent/CN112634026A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108428132A (en) * | 2018-03-15 | 2018-08-21 | 阿里巴巴集团控股有限公司 | Fraudulent trading recognition methods, device, server and storage medium |
| CN109345260A (en) * | 2018-10-09 | 2019-02-15 | 北京芯盾时代科技有限公司 | A kind of fraud detection model training method and device and fraud detection method and device |
| CN109410036A (en) * | 2018-10-09 | 2019-03-01 | 北京芯盾时代科技有限公司 | A kind of fraud detection model training method and device and fraud detection method and device |
| CN109544190A (en) * | 2018-11-28 | 2019-03-29 | 北京芯盾时代科技有限公司 | A kind of fraud identification model training method, fraud recognition methods and device |
| CN110969441A (en) * | 2019-12-23 | 2020-04-07 | 集奥聚合(北京)人工智能科技有限公司 | Anti-fraud model processing method and device based on deep learning |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113240422A (en) * | 2021-07-09 | 2021-08-10 | 天聚地合(苏州)数据股份有限公司 | Financial product transfer method based on block chain and related device |
| CN114493826A (en) * | 2021-12-22 | 2022-05-13 | 四川新网银行股份有限公司 | Personal credit assessment scoring method based on neural network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108509862B (en) | Rapid face recognition method capable of resisting angle and shielding interference | |
| CN109344731B (en) | Lightweight face recognition method based on neural network | |
| Taheri et al. | Animal classification using facial images with score‐level fusion | |
| CN112634026A (en) | Credit fraud identification method based on user page operation behavior | |
| Lai et al. | Learning discriminative feature hierarchies for off-line signature verification | |
| Cecotti | Deep random vector functional link network for handwritten character recognition | |
| Mahmoud et al. | A novel method for traffic sign recognition based on DCGAN and MLP with PILAE algorithm | |
| Manjusha et al. | Integrating scattering feature maps with convolutional neural networks for Malayalam handwritten character recognition | |
| CN114612968A (en) | Convolutional neural network-based lip print identification method | |
| Kumar et al. | Ann trained and WOA optimized feature-level fusion of iris and fingerprint | |
| Farhan et al. | A new model for pattern recognition | |
| CN110246509B (en) | Stack type denoising self-encoder and deep neural network structure for voice lie detection | |
| Zhuang et al. | Iris recognition using convolutional neural network | |
| Yu et al. | A multilayer pyramid network based on learning for vehicle logo recognition | |
| Rahaman et al. | BHCDR: real-time bangla handwritten characters and digits recognition using adopted convolutional neural network | |
| Shayegan et al. | A New Dataset Size Reduction Approach for PCA‐Based Classification in OCR Application | |
| CN114692809A (en) | Data processing method and device based on neural cluster, storage medium and processor | |
| Drott et al. | On-line handwritten signature verification using machine learning techniques with a deep learning approach | |
| JP5284530B2 (en) | Information processing method and information processing apparatus | |
| CN113128387B (en) | Drug addiction attack recognition method for drug addicts based on facial expression feature analysis | |
| Evangelin et al. | Reduced optimal feature based biometric authentication using MALO-MKSVM techniques | |
| Oh et al. | A comparative study of feature extraction methods and their application to p-rbf nns in face recognition problem | |
| Soviany et al. | A biometric system with hierarchical feature-level fusion | |
| Abdullah et al. | Independent Component Analysis and Support Vector Neural Network for Face Recognition | |
| Priya et al. | Multimodal biometric authentication using back propagation artificial neural network |
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: 20210409 |
|
| RJ01 | Rejection of invention patent application after publication |