CN117911034A - Credit card abnormal transaction detection method and device - Google Patents
Credit card abnormal transaction detection method and device Download PDFInfo
- Publication number
- CN117911034A CN117911034A CN202410090783.0A CN202410090783A CN117911034A CN 117911034 A CN117911034 A CN 117911034A CN 202410090783 A CN202410090783 A CN 202410090783A CN 117911034 A CN117911034 A CN 117911034A
- Authority
- CN
- China
- Prior art keywords
- credit card
- data
- transaction
- abnormal
- partition
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/2433—Single-class perspective, e.g. one-against-all classification; Novelty detection; Outlier detection
-
- G—PHYSICS
- G06—COMPUTING OR CALCULATING; 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 OR CALCULATING; 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
- G06Q40/04—Trading; Exchange, e.g. stocks, commodities, derivatives or currency exchange
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Accounting & Taxation (AREA)
- Finance (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Strategic Management (AREA)
- Marketing (AREA)
- Development Economics (AREA)
- Data Mining & Analysis (AREA)
- Technology Law (AREA)
- Economics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Computer Security & Cryptography (AREA)
- Financial Or Insurance-Related Operations Such As Payment And Settlement (AREA)
Abstract
The invention provides a credit card abnormal transaction detection method and device, relates to the technical field of artificial intelligence, and can be used in the financial field or other technical fields. The method comprises the following steps: acquiring credit card transaction data to be detected; performing anomaly detection on the credit card transaction data to be detected based on a preset credit card anomaly transaction detection model to obtain a credit card anomaly transaction detection result; the preset credit card abnormal transaction detection model is obtained by training a local abnormal factor algorithm which performs partition data balancing processing in advance based on credit card abnormal transaction detection sample data. The apparatus performs the above method. According to the credit card abnormal transaction detection method and device, the accuracy and the detection efficiency of the detection result of the multidimensional credit card transaction data can be improved through the local outlier factor algorithm for carrying out the partition data balancing processing.
Description
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to a credit card abnormal transaction detection method and device.
Background
In recent years, the whole credit card issuing amount is in an upward trend, the credit card consumption scene is diversified, meanwhile, the telecommunication phishing cases related to credit card transactions are in a continuous high-rise state, and the anti-fraud means are continuously upgraded from the aspects of intercepting and blocking risk transactions, helping customers to raise risk prevention consciousness and the like, so that the anti-fraud means are improved, and the anti-telecommunication phishing striking force related to credit cards is increased. However, the prior prevention and control means cannot prevent the occurrence of fraud related transactions, the existing detection mechanism is usually a general anti-fraud model established by using massive transaction data, and for short-term small-volume, fragmented and high-dimensional transaction data related to credit cards, the insufficient data sample causes the failure to establish a reliable model, and meanwhile, the general anti-fraud model has limitation on the data characteristic pluralism and has poor performance for detecting fraud related to credit cards.
There are many credit card fraud transaction detection methods in the field of data mining, currently, a wider detection method based on distance and density is applied, the latter is more representative of LOF (Local Outlier Factor) local anomaly factor algorithm, the principle is that the density of adjacent data points is calculated by the data quantity of the adjacent data points and compared with the density of other data points to judge whether the data object is an anomaly value, the algorithm thought is concise and is not influenced by data distribution, the detection effect on low-dimensional data is better, but the detection requirement of high-dimensional data related to credit card transactions cannot be met due to higher time complexity and poorer detection efficiency on high-dimensional data sets.
Disclosure of Invention
Aiming at the problems in the prior art, the embodiment of the invention provides a credit card abnormal transaction detection method and device, which can at least partially solve the problems in the prior art.
In one aspect, the present invention provides a method for detecting abnormal transactions of a credit card, including:
Acquiring credit card transaction data to be detected;
Performing anomaly detection on the credit card transaction data to be detected based on a preset credit card anomaly transaction detection model to obtain a credit card anomaly transaction detection result;
The preset credit card abnormal transaction detection model is obtained by training a local abnormal factor algorithm which performs partition data balancing processing in advance based on credit card abnormal transaction detection sample data.
The abnormal credit card transaction data to be detected is subjected to abnormal detection based on a preset credit card abnormal transaction detection model to obtain a credit card abnormal transaction detection result, and the abnormal credit card transaction detection method comprises the following steps:
if the local abnormality factor output by the preset credit card abnormal transaction detection model is greater than 1, determining that the credit card abnormal transaction detection result is abnormal;
And if the local abnormal factor output by the preset credit card abnormal transaction detection model is less than or equal to 1, determining that the credit card abnormal transaction detection result is normal.
The method for obtaining the default credit card abnormal transaction detection model based on the local abnormal factor algorithm of the credit card abnormal transaction detection sample data training and carrying out the partition data balancing processing in advance comprises the following steps:
acquiring multidimensional credit card transaction data, and performing dimension reduction processing on the multidimensional credit card transaction data to obtain target dimension credit card transaction data;
Carrying out partition data balancing processing on the target dimension credit card transaction data to obtain a first partition data set and a second partition data set;
Calculating according to the first partition data set and the second partition data set to obtain a local abnormal factor predicted value;
and training a local anomaly factor algorithm according to the local anomaly factor predicted value and the local anomaly factor true value to obtain the preset credit card anomaly transaction detection model.
The step of carrying out partition data balancing processing on the target dimension credit card transaction data to obtain a first partition data set and a second partition data set comprises the following steps:
Carrying out partition data quantity equalization processing and partition diameter minimization processing on the target dimension credit card transaction data;
the method comprises the steps of taking a partition data quantity equalization processing result and a partition diameter minimization processing result as targets, and solving to obtain an optimal segmentation bit;
And taking the credit card transaction data with the target dimension smaller than the optimal dividing bit as a first partition data set, and taking the credit card transaction data with the target dimension larger than or equal to the optimal dividing bit as a second partition data set.
The method for minimizing the partition diameter of the credit card transaction data with the target dimension comprises the following steps:
calculating the average value between the maximum value and the minimum value of the credit card transaction data of the target dimension;
And carrying out partition diameter minimization processing on the credit card transaction data of the target dimension according to the average value, the maximum value and the minimum value.
The calculating according to the first partition data set and the second partition data set to obtain a local abnormal factor predicted value includes:
according to the first partition data set and the second partition data set, performing double pruning acceleration data retrieval calculation by using a k nearest neighbor algorithm and a range query algorithm to obtain a k distance of a data object;
and calculating the local anomaly factor predicted value according to the kth distance of the data object.
The calculating the local anomaly factor predicted value according to the kth distance of the data object comprises the following steps:
calculating a kth distance neighborhood of the data object according to the kth distance of the data object;
calculating a kth reachable distance of the data object according to the Euclidean distance and the kth distance of the data object;
Calculating to obtain the local reachable density of the data object according to the data object capacity of the kth distance neighborhood corresponding to the kth distance neighborhood and the kth reachable distance of the data object;
and calculating the local abnormal factor predicted value according to the local reachable density of the data object and the capacity of the kth distance neighborhood data object.
In one aspect, the present invention provides a credit card abnormal transaction detection device, comprising:
The acquisition unit is used for acquiring credit card transaction data to be detected;
The detection unit is used for carrying out anomaly detection on the credit card transaction data to be detected based on a preset credit card anomaly transaction detection model to obtain a credit card anomaly transaction detection result;
The preset credit card abnormal transaction detection model is obtained by training a local abnormal factor algorithm which performs partition data balancing processing in advance based on credit card abnormal transaction detection sample data.
In still another aspect, an embodiment of the present invention provides an electronic device, including: a processor, a memory, and a bus, wherein,
The processor and the memory complete communication with each other through the bus;
The memory stores program instructions executable by the processor, the processor invoking the program instructions capable of performing the method of:
Acquiring credit card transaction data to be detected;
Performing anomaly detection on the credit card transaction data to be detected based on a preset credit card anomaly transaction detection model to obtain a credit card anomaly transaction detection result;
The preset credit card abnormal transaction detection model is obtained by training a local abnormal factor algorithm which performs partition data balancing processing in advance based on credit card abnormal transaction detection sample data.
Embodiments of the present invention provide a non-transitory computer readable storage medium comprising:
the non-transitory computer readable storage medium stores computer instructions that cause the computer to perform the method of:
Acquiring credit card transaction data to be detected;
Performing anomaly detection on the credit card transaction data to be detected based on a preset credit card anomaly transaction detection model to obtain a credit card anomaly transaction detection result;
The preset credit card abnormal transaction detection model is obtained by training a local abnormal factor algorithm which performs partition data balancing processing in advance based on credit card abnormal transaction detection sample data.
The credit card abnormal transaction detection method and device provided by the embodiment of the invention acquire the credit card transaction data to be detected; performing anomaly detection on the credit card transaction data to be detected based on a preset credit card anomaly transaction detection model to obtain a credit card anomaly transaction detection result; the preset credit card abnormal transaction detection model is obtained by training a local abnormal factor algorithm which is subjected to partition data balancing processing in advance based on credit card abnormal transaction detection sample data, and the accuracy and the detection efficiency of the detection result of the multidimensional credit card transaction data can be improved by the local outlier factor algorithm which is subjected to the partition data balancing processing.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art. In the drawings:
fig. 1 is a flowchart of a method for detecting abnormal credit card transactions according to an embodiment of the invention.
Fig. 2 is a flowchart of a method for detecting abnormal credit card transactions according to another embodiment of the present invention.
Fig. 3 is an explanatory diagram of a credit card abnormal transaction detection method according to an embodiment of the present invention.
Fig. 4 is a schematic structural diagram of a credit card abnormal transaction detecting device according to an embodiment of the invention.
Fig. 5 is a schematic diagram of an entity structure of an electronic device according to an embodiment of the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present application more apparent, the embodiments of the present application will be described in further detail with reference to the accompanying drawings. The exemplary embodiments of the present application and their descriptions herein are for the purpose of explaining the present application, but are not to be construed as limiting the application. It should be noted that, without conflict, the embodiments of the present application and features of the embodiments may be arbitrarily combined with each other.
Fig. 1 is a flow chart of a method for detecting abnormal credit card transactions according to an embodiment of the present invention, as shown in fig. 1, the method for detecting abnormal credit card transactions according to the embodiment of the present invention includes:
Step S1: and acquiring credit card transaction data to be detected.
Step S2: performing anomaly detection on the credit card transaction data to be detected based on a preset credit card anomaly transaction detection model to obtain a credit card anomaly transaction detection result;
The preset credit card abnormal transaction detection model is obtained by training a local abnormal factor algorithm which performs partition data balancing processing in advance based on credit card abnormal transaction detection sample data.
In the above step S1, the apparatus acquires credit card transaction data to be detected. The apparatus may be a computer device performing the method. It should be noted that, the data acquisition and analysis according to the embodiments of the present invention are authorized by the user. The credit card transaction data to be detected may be credit card transaction data of one dimension which plays a key role in credit card abnormal transaction detection and is reserved through analysis, and may include one dimension of credit card transaction data of a plurality of dimensions as follows:
Date, location, transaction type, amount, vendor, transaction device identification, and customer's behavioral patterns, etc.
In the step S2, the device performs anomaly detection on the credit card transaction data to be detected based on a preset credit card anomaly transaction detection model to obtain a credit card anomaly transaction detection result; the credit card transaction data to be detected can be input into a preset credit card abnormal transaction detection model, local abnormal factors are output through the preset credit card abnormal transaction detection model, and the credit card abnormal transaction detection result is determined according to the values of the local abnormal factors.
The abnormal credit card transaction data to be detected is subjected to abnormal detection based on a preset credit card abnormal transaction detection model to obtain a credit card abnormal transaction detection result, and the abnormal credit card transaction detection method comprises the following steps:
if the local abnormality factor output by the preset credit card abnormal transaction detection model is greater than 1, determining that the credit card abnormal transaction detection result is abnormal;
And if the local abnormal factor output by the preset credit card abnormal transaction detection model is less than or equal to 1, determining that the credit card abnormal transaction detection result is normal.
The preset credit card abnormal transaction detection model is obtained by training a local abnormal factor algorithm which performs partition data balancing processing in advance based on credit card abnormal transaction detection sample data.
Training a local anomaly factor algorithm which performs partition data balancing processing in advance based on credit card anomaly transaction detection sample data to obtain the preset credit card anomaly transaction detection model, wherein the method comprises the following steps:
Acquiring multidimensional credit card transaction data, and performing dimension reduction processing on the multidimensional credit card transaction data to obtain target dimension credit card transaction data; this step corresponds to "generating a pre-divided new data set" in S101 and S102 in fig. 2.
As shown in fig. 3, after the customer initiates the transaction, the transaction data is transferred to the transaction processing center of the bank for processing, and all the transaction data are stored in the transaction database in batches at the frequency of t+1.
According to a series of data characteristic choosing and rejecting rules defined in advance, all meaningful characteristics (such as date, place, transaction type, amount, supplier, transaction equipment identification, customer behavior mode and the like) are reserved, transaction data related to a credit card are extracted, the data are desensitized and then processed and stored in a model database, and a data set required by model training is obtained.
It should be noted that, the data required for training the original detection model comes from stock transaction data, part of the data is marked as abnormal transaction by the transaction processing center of the bank, and incremental data generated in batch on a daily basis is identified by the initial model and then is combined into a training data set to participate in training of the new model. Training new models after the increment data set reaches a certain number, and gradually enhancing the generalization capability of the detection model.
The step S102 achieves space division of the balanced processed data object. The density-based anomaly detection algorithm modeling process involves neighbor searching of data objects, and credit card transaction data is often multidimensional, so that a spherical tree model widely used for various multidimensional space data retrieval and access scenes is selected to store transaction data used for modeling.
It should be noted that, in the conventional ball tree model, the data clusters are divided by relying on two data objects with the farthest space distance, and in the data set with abnormal data, the data imbalance may be caused by such a data division manner, so that the embodiment of the invention uses a data space based on the data principal component and the data quantity balancing manner to construct the data space of the data set to be tested. The step specifically includes three parts, namely (1) generating a pre-partitioned new data set; (2) partition data balancing; and (3) data dividing processing.
Generating a pre-partitioned new data set for (1) is described as follows:
Principal component analysis (PRINCIPLE COMPONENT ANALYSIS, PCA) is applied to transform the data and extract a target feature vector w 1, expressed as:
Wherein x= { X 1,x2......xn}∈Rd is input n-dimensional credit card transaction data, t= { T 1,t2......tn } is target dimension credit card transaction data obtained by PCA conversion of the data, and w is a feature vector, namely
Wherein,For the j-th eigenvalue of the target dimension credit card transaction data with the i-th dimension as the target dimension, x i is the i-th dimension credit card transaction data, and w (j) is the j-th eigenvalue of the eigenvector.
The reason for retaining only the target feature vector w 1 is that the nature of the PCA determines that this vector retains the most important information in the data object.
Using mapping of the input dataset to its target feature vector w 1, a new dataset T (i.e., a dataset made up of target dimension credit card transaction data) is obtained, which is a scalar dataset, which is transformed in a manner that can be expressed as:
T＝X·w(1)。
Carrying out partition data balancing processing on the target dimension credit card transaction data to obtain a first partition data set and a second partition data set; this step corresponds to the "data balancing" and "data dividing processing" in S102. The step of carrying out partition data balancing processing on the target dimension credit card transaction data to obtain a first partition data set and a second partition data set, comprising the following steps:
Carrying out partition data quantity equalization processing and partition diameter minimization processing on the target dimension credit card transaction data;
the method comprises the steps of taking a partition data quantity equalization processing result and a partition diameter minimization processing result as targets, and solving to obtain an optimal segmentation bit;
And taking the credit card transaction data with the target dimension smaller than the optimal dividing bit as a first partition data set, and taking the credit card transaction data with the target dimension larger than or equal to the optimal dividing bit as a second partition data set.
Performing partition diameter minimization processing on the target dimension credit card transaction data, including:
calculating the average value between the maximum value and the minimum value of the credit card transaction data of the target dimension;
And carrying out partition diameter minimization processing on the credit card transaction data of the target dimension according to the average value, the maximum value and the minimum value.
The partition data balancing in (2) is described as follows:
Partition data quantity equalization: when dividing the data space of the data to be measured, dividing each time to make the number of data objects of two space sub-partitions consistent as much as possible so as to accelerate neighbor search, and converting into a request The optimal solution, wherein N represents the data amount of the target dimension credit card transaction data, namely the amount of data to be divided, N 1 represents the amount of data of one part of the sub-partition after division, and N 2 represents the amount of data of another part of the sub-partition after division.
Partition diameter minimization: in the case where the data points of the sub-partitions are as balanced as possible, the data retrieval of the ball tree structure needs to be accelerated so that the dividing diameter of the data partition is as small as possible. In order to simply and efficiently determine the optimal data space division bits, in the new scalar dataset T, the maximum and minimum values of the target dimension credit card transaction data are set to be T max and T min, respectively. The data space division bits should be located as far as possible in between, i.e. the mean valueExtreme zone diameters are avoided. The split bits may be expressed as: /(I)
The data dividing process of (3) is described as follows:
To be used for For the purpose, the optimal solution is determined as the optimal segmentation bit p tar.
Where a is a weight parameter, that is, when there is no obvious difference in the characteristics of the data objects in the data set to be divided, setting a smaller weight a is biased to find the optimal data space division bit by using the partition data quantity balance. And carrying out data division on each sub-partition according to the dividing principle. After all data partitioning is completed, the left partition dataset of any partitionable data partition of the ball tree structure can be expressed as: x L＝{xi|xi∈X,ti＜ptar }, the right partition dataset can be expressed as: x R＝{xi|xi∈X,ti≥ptar }. The left partition data set corresponds to the first partition data set and the right partition data set corresponds to the second partition data set.
Calculating according to the first partition data set and the second partition data set to obtain a local abnormal factor predicted value; this step corresponds to step 103, and specifically includes four parts, namely (1) calculating the kth distance of the data object; (2) calculating the reachable distance of the data object; (3) calculating a local reachable density of the data object; (4) A local anomaly factor predictor LOF k (p) for the data object is calculated.
The calculating according to the first partition data set and the second partition data set to obtain a local abnormal factor predicted value includes:
according to the first partition data set and the second partition data set, performing double pruning acceleration data retrieval calculation by using a k nearest neighbor algorithm and a range query algorithm to obtain a k distance of a data object;
and calculating the local anomaly factor predicted value according to the kth distance of the data object.
The calculating the local anomaly factor predicted value according to the kth distance of the data object comprises the following steps:
calculating a kth distance neighborhood of the data object according to the kth distance of the data object;
calculating a kth reachable distance of the data object according to the Euclidean distance and the kth distance of the data object;
Calculating to obtain the local reachable density of the data object according to the data object capacity of the kth distance neighborhood corresponding to the kth distance neighborhood and the kth reachable distance of the data object;
and calculating the local abnormal factor predicted value according to the local reachable density of the data object and the capacity of the kth distance neighborhood data object.
The calculation of the kth distance of the data object in (1) is described as follows:
Calculating the Euclidean distance between the data objects: for any credit card transaction data object p (p e X) and q (q e X), the true distances of p and q in the data space are denoted as d (p, q), which according to the Euclidean distance definition can be expressed as:
Where p i and q i represent eigenvalues of data objects p and q in the i dimension, and n represents the dimension of the credit card transaction dataset, so 1.ltoreq.i.ltoreq.n.
Calculating a kth distance of the data object: for a set threshold value k, 1.ltoreq.k.ltoreq.S is satisfied, where S is the capacity of the dataset X. The kth distance of the data object p is denoted as k-dis (p). There are two cases such that k-dis (p) =d (p, q):
① At least k data objects q 'E X\ { p }, satisfying d (p, q') +.d (p, q).
② At most k-1 data objects q 'E X\ { p } satisfy d (p, q') < d (p, q).
Namely: in the constructed ball tree structure, a k-nearest neighbor algorithm and a range query algorithm are used for double pruning acceleration data retrieval, euclidean distances of other k data objects which are closer to p in a data set X (excluding q) are calculated, and q is the kth data object closest to p.
The data structure of the data set of the method of the embodiment of the invention is not a linear data structure at this time, so that the k-nearest neighbor algorithm and the range query algorithm can be used for parallel calculation, and the data retrieval speed is greatly improved.
Computing a kth distance neighborhood of the data object: for data object p, there is a data set of a plurality of other data objects that are not p at its kth distance, which is referred to as the kth distance neighborhood of p, denoted as N k-dis(p), expressed as:
when a plurality of data objects exist in X and the spatial distance of the data objects is the same as that of p, the k-th distance neighborhood data object capacity of p is larger than a threshold value k, namely, the capacity of the N k (p).
Calculating the reachable distance of the data object in the step (2) is described as follows:
The kth reachable distance of data objects p to q, centered on data object q, is denoted as reach-dis k (p, q), which is at least the kth distance of q or the true distance between p, q, expressed as: reach-dis k (p, q) =max { k-dis (q), d (p, q) }.
If data object p is located within the kth distance neighborhood of q, then reach-dis k (p, q) =k-dis (q); if the true distance of the data object from q is greater than the kth distance of q, then reach-dis k (p, q) =d (p, q).
Calculating the local reachable density of the data object according to the step (3), wherein the method is as follows:
for data object p, the inverse of the average reachable distance to p for all data objects in its kth distance neighborhood is defined as the local reachable density of p, denoted lrd k (p), expressed as:
Where b is a very small constant to avoid the occurrence of a denominator of 0. For a credit card transaction data object p, the local reachable density is regarded as the data density in a given threshold area, the higher the data density, the greater the likelihood that the data object is clustered with other data objects in the area, and the greater the likelihood that the transaction is a normal transaction. Conversely, if the data concentration is low, indicating that the data object deviates from the majority population, the greater the probability of being an abnormal transaction.
The local anomaly factor predictive value LOF k (p) of the data object calculated in (4) is described as follows:
The average of the local reachable densities of other data objects within the kth distance neighborhood of data object p to the local reachable density ratio of p is defined as the local anomaly factor predictor LOF k (p) for p, expressed as:
for any transaction data, the LOF value of the local abnormal factor LOF value is the most direct standard for judging whether the transaction is abnormal or not, and when the local reachable densities of the data object p and all the data objects in the kth distance field are exactly consistent, the local abnormal factor LOF value of the p is 1, which indicates that the p has the same characteristic as other data objects in the adjacent field, and the probability of belonging to the same cluster is high, so that the local abnormal factor LOF value can be considered as normal transaction.
When the local reachable density of the data object p is greater than the local reachable density of other data objects in the neighborhood, the local anomaly factor LOF value of p is smaller than 1, which indicates that the data object p may be located in a dense region of the data object, and the possibility of abnormal transactions may be excluded, and the data object p may be considered as normal transactions.
When the local reachable density of the data object p is smaller than that of other data objects in the neighborhood, the local anomaly factor LOF value of p is larger than 1, which indicates that the data object p is the anomaly transaction data, and the greater the value of LOF, the more obvious the anomaly characteristic.
And training a local anomaly factor algorithm according to the local anomaly factor predicted value and the local anomaly factor true value to obtain the preset credit card anomaly transaction detection model.
And obtaining abnormal transaction data detected by the model according to LOF value distribution, sequentially calculating confusion matrix, accuracy and recall rate with real labels of training samples of the model, and finally calculating F1_score of the model. Dynamically adjusting an algorithm model training threshold, selecting the threshold with the highest F1_score as the threshold of the algorithm model, obtaining an optimal recognition model, and deploying the optimal recognition model to an abnormal transaction detection center. As shown in fig. 3, after the user initiates the transaction, the transaction processing center completes the data processing and then flows to the detection center, the detection center issues the detection result, the transaction processing center feeds back the result to the user, and meanwhile, the data is stored in the transaction database.
The invention provides a credit card abnormal transaction rapid detection method based on local abnormal factor detection, which utilizes a data space division criterion based on principal component analysis and a data balance tree principle, and adopts a spherical neighbor search data structure optimized by pruning query strategies of k neighbor search and range search in a mixed manner to accelerate the density calculation speed of each data point in a local outlier factor algorithm, and simultaneously adapts to high-dimensional space data retrieval and storage scenes such as credit card transaction data, thereby accurately and rapidly detecting abnormal transactions related to credit cards.
The invention provides a credit card abnormal transaction rapid detection method based on local abnormal factor detection, which optimizes a traditional local outlier factor algorithm by utilizing a neighbor search data structure suitable for a high-dimensional space data retrieval and storage scene, improves the model construction and data search speed of the algorithm and the multi-dimensional data detection precision, finds a proper model threshold of an initial model by using client stock transaction data through multiple times of training, automatically acquires transaction data to carry out enhancement training, dynamically adjusts the detection threshold, and establishes an optimal credit card fraud transaction detection model.
The invention provides a credit card abnormal transaction rapid detection method based on local abnormal factor detection, which has the following beneficial technical effects:
1. At present, a further excavation space exists in the field of transaction card type subdivision for abnormal transaction detection, and the method can finish the rapid detection of abnormal transaction behaviors for credit card-related transactions by extracting the characteristics of the credit card-related transaction data and utilizing a rapid detection algorithm model.
2. The method uses a classical density-based anomaly detection algorithm model, optimizes the defects of the algorithm in the aspect of processing high-dimensional data, optimizes the data space division mode and the search mode, effectively reduces the time complexity of the anomaly detection algorithm, and improves the speed and efficiency of anomaly transaction detection.
3. The abnormal transaction detection model for credit card transactions trained by the method has good transfer learning capability, and can be applied to other scenes related to abnormal transaction detection in an expanded manner.
The credit card abnormal transaction detection method provided by the embodiment of the invention acquires the credit card transaction data to be detected; performing anomaly detection on the credit card transaction data to be detected based on a preset credit card anomaly transaction detection model to obtain a credit card anomaly transaction detection result; the preset credit card abnormal transaction detection model is obtained by training a local abnormal factor algorithm which is subjected to partition data balancing processing in advance based on credit card abnormal transaction detection sample data, and the accuracy and the detection efficiency of the detection result of the multidimensional credit card transaction data can be improved by the local outlier factor algorithm which is subjected to the partition data balancing processing.
Further, the detecting the abnormality of the credit card transaction data to be detected based on the preset credit card abnormal transaction detection model to obtain a credit card abnormal transaction detection result includes:
If the local abnormality factor output by the preset credit card abnormal transaction detection model is greater than 1, determining that the credit card abnormal transaction detection result is abnormal; the description of the embodiments may be referred to above, and will not be repeated.
And if the local abnormal factor output by the preset credit card abnormal transaction detection model is less than or equal to 1, determining that the credit card abnormal transaction detection result is normal. The description of the embodiments may be referred to above, and will not be repeated.
The credit card abnormal transaction detection method provided by the embodiment of the invention can further improve the accuracy of the detection result of the multidimensional credit card transaction data.
Further, training a local anomaly factor algorithm for carrying out partition data balancing processing in advance based on credit card anomaly transaction detection sample data to obtain the preset credit card anomaly transaction detection model, wherein the method comprises the following steps:
Acquiring multidimensional credit card transaction data, and performing dimension reduction processing on the multidimensional credit card transaction data to obtain target dimension credit card transaction data; the description of the embodiments may be referred to above, and will not be repeated.
Carrying out partition data balancing processing on the target dimension credit card transaction data to obtain a first partition data set and a second partition data set; the description of the embodiments may be referred to above, and will not be repeated.
Calculating according to the first partition data set and the second partition data set to obtain a local abnormal factor predicted value; the description of the embodiments may be referred to above, and will not be repeated.
And training a local anomaly factor algorithm according to the local anomaly factor predicted value and the local anomaly factor true value to obtain the preset credit card anomaly transaction detection model. The description of the embodiments may be referred to above, and will not be repeated.
The credit card abnormal transaction detection method provided by the embodiment of the invention can further improve the detection efficiency of multidimensional credit card transaction data.
Further, the performing a partition data balancing process on the target dimension credit card transaction data to obtain a first partition data set and a second partition data set, including:
carrying out partition data quantity equalization processing and partition diameter minimization processing on the target dimension credit card transaction data; the description of the embodiments may be referred to above, and will not be repeated.
The method comprises the steps of taking a partition data quantity equalization processing result and a partition diameter minimization processing result as targets, and solving to obtain an optimal segmentation bit; the description of the embodiments may be referred to above, and will not be repeated.
And taking the credit card transaction data with the target dimension smaller than the optimal dividing bit as a first partition data set, and taking the credit card transaction data with the target dimension larger than or equal to the optimal dividing bit as a second partition data set. The description of the embodiments may be referred to above, and will not be repeated.
The credit card abnormal transaction detection method provided by the embodiment of the invention can further improve the detection efficiency of multidimensional credit card transaction data by accurately determining the optimal division position and reasonably dividing the data.
Further, performing partition diameter minimization processing on the target dimension credit card transaction data, including:
calculating the average value between the maximum value and the minimum value of the credit card transaction data of the target dimension; the description of the embodiments may be referred to above, and will not be repeated.
And carrying out partition diameter minimization processing on the credit card transaction data of the target dimension according to the average value, the maximum value and the minimum value. The description of the embodiments may be referred to above, and will not be repeated.
The credit card abnormal transaction detection method provided by the embodiment of the invention is beneficial to accurately determining the optimal segmentation position through the minimized processing of the partition diameter so as to reasonably divide data.
Further, the calculating the local anomaly factor predicted value according to the first partition data set and the second partition data set includes:
according to the first partition data set and the second partition data set, performing double pruning acceleration data retrieval calculation by using a k nearest neighbor algorithm and a range query algorithm to obtain a k distance of a data object; the description of the embodiments may be referred to above, and will not be repeated.
And calculating the local anomaly factor predicted value according to the kth distance of the data object. The description of the embodiments may be referred to above, and will not be repeated.
The credit card abnormal transaction detection method provided by the embodiment of the invention utilizes two algorithms to calculate in parallel, so that the detection efficiency of multidimensional credit card transaction data can be further improved.
Further, the calculating the local anomaly factor predicted value according to the kth distance of the data object includes:
calculating a kth distance neighborhood of the data object according to the kth distance of the data object; the description of the embodiments may be referred to above, and will not be repeated.
Calculating a kth reachable distance of the data object according to the Euclidean distance and the kth distance of the data object; the description of the embodiments may be referred to above, and will not be repeated.
Calculating to obtain the local reachable density of the data object according to the data object capacity of the kth distance neighborhood corresponding to the kth distance neighborhood and the kth reachable distance of the data object; the description of the embodiments may be referred to above, and will not be repeated.
And calculating the local abnormal factor predicted value according to the local reachable density of the data object and the capacity of the kth distance neighborhood data object. The description of the embodiments may be referred to above, and will not be repeated.
The credit card abnormal transaction detection method provided by the embodiment of the invention can accurately calculate the local abnormal factor predicted value and ensure the subsequent model training efficiency.
It should be noted that, the method for detecting abnormal credit card transactions provided by the embodiment of the present invention may be used in the financial field, and may also be used in any technical field other than the financial field.
Fig. 4 is a schematic structural diagram of a credit card abnormal transaction detection device according to an embodiment of the present invention, as shown in fig. 4, the credit card abnormal transaction detection device according to an embodiment of the present invention includes an obtaining unit 401 and a detecting unit 402, where:
the acquisition unit 401 is configured to acquire credit card transaction data to be detected; the detecting unit 402 is configured to perform anomaly detection on the credit card transaction data to be detected based on a preset credit card anomaly transaction detection model, so as to obtain a credit card anomaly transaction detection result; the preset credit card abnormal transaction detection model is obtained by training a local abnormal factor algorithm which performs partition data balancing processing in advance based on credit card abnormal transaction detection sample data.
Specifically, the acquiring unit 401 in the device is configured to acquire credit card transaction data to be detected; the detecting unit 402 is configured to perform anomaly detection on the credit card transaction data to be detected based on a preset credit card anomaly transaction detection model, so as to obtain a credit card anomaly transaction detection result; the preset credit card abnormal transaction detection model is obtained by training a local abnormal factor algorithm which performs partition data balancing processing in advance based on credit card abnormal transaction detection sample data.
The credit card abnormal transaction detection device provided by the embodiment of the invention acquires the credit card transaction data to be detected; performing anomaly detection on the credit card transaction data to be detected based on a preset credit card anomaly transaction detection model to obtain a credit card anomaly transaction detection result; the preset credit card abnormal transaction detection model is obtained by training a local abnormal factor algorithm which is subjected to partition data balancing processing in advance based on credit card abnormal transaction detection sample data, and the accuracy and the detection efficiency of the detection result of the multidimensional credit card transaction data can be improved by the local outlier factor algorithm which is subjected to the partition data balancing processing.
Further, the detecting unit 402 is specifically configured to:
if the local abnormality factor output by the preset credit card abnormal transaction detection model is greater than 1, determining that the credit card abnormal transaction detection result is abnormal;
And if the local abnormal factor output by the preset credit card abnormal transaction detection model is less than or equal to 1, determining that the credit card abnormal transaction detection result is normal.
The credit card abnormal transaction detection device provided by the embodiment of the invention can further improve the accuracy of the detection result of the multidimensional credit card transaction data.
Further, the credit card abnormal transaction detection device is further configured to:
acquiring multidimensional credit card transaction data, and performing dimension reduction processing on the multidimensional credit card transaction data to obtain target dimension credit card transaction data;
Carrying out partition data balancing processing on the target dimension credit card transaction data to obtain a first partition data set and a second partition data set;
Calculating according to the first partition data set and the second partition data set to obtain a local abnormal factor predicted value;
and training a local anomaly factor algorithm according to the local anomaly factor predicted value and the local anomaly factor true value to obtain the preset credit card anomaly transaction detection model.
The credit card abnormal transaction detection device provided by the embodiment of the invention can further improve the detection efficiency of multidimensional credit card transaction data.
Further, the credit card abnormal transaction detection device is specifically configured to:
Carrying out partition data quantity equalization processing and partition diameter minimization processing on the target dimension credit card transaction data;
the method comprises the steps of taking a partition data quantity equalization processing result and a partition diameter minimization processing result as targets, and solving to obtain an optimal segmentation bit;
And taking the credit card transaction data with the target dimension smaller than the optimal dividing bit as a first partition data set, and taking the credit card transaction data with the target dimension larger than or equal to the optimal dividing bit as a second partition data set.
The credit card abnormal transaction detection device provided by the embodiment of the invention can further improve the detection efficiency of multidimensional credit card transaction data by accurately determining the optimal division position and reasonably dividing the data.
Further, the credit card abnormal transaction detection device is specifically configured to:
calculating the average value between the maximum value and the minimum value of the credit card transaction data of the target dimension;
And carrying out partition diameter minimization processing on the credit card transaction data of the target dimension according to the average value, the maximum value and the minimum value.
The credit card abnormal transaction detection device provided by the embodiment of the invention is beneficial to accurately determining the optimal dividing position through minimizing the diameter of the partition so as to reasonably divide the data.
Further, the credit card abnormal transaction detection device is specifically configured to:
according to the first partition data set and the second partition data set, performing double pruning acceleration data retrieval calculation by using a k nearest neighbor algorithm and a range query algorithm to obtain a k distance of a data object;
and calculating the local anomaly factor predicted value according to the kth distance of the data object.
The credit card abnormal transaction detection device provided by the embodiment of the invention utilizes two algorithms to calculate in parallel, so that the detection efficiency of multidimensional credit card transaction data can be further improved.
Further, the credit card abnormal transaction detection device is specifically configured to:
calculating a kth distance neighborhood of the data object according to the kth distance of the data object;
calculating a kth reachable distance of the data object according to the Euclidean distance and the kth distance of the data object;
Calculating to obtain the local reachable density of the data object according to the data object capacity of the kth distance neighborhood corresponding to the kth distance neighborhood and the kth reachable distance of the data object;
and calculating the local abnormal factor predicted value according to the local reachable density of the data object and the capacity of the kth distance neighborhood data object.
The credit card abnormal transaction detection device provided by the embodiment of the invention can accurately calculate the local abnormal factor predicted value, and ensure the subsequent model training efficiency.
The embodiment of the present invention provides a credit card abnormal transaction detection device, which can be specifically used to execute the processing flow of each method embodiment, and the functions thereof are not described herein again, and reference may be made to the detailed description of the method embodiments.
Fig. 5 is a schematic diagram of an entity structure of an electronic device according to an embodiment of the present invention, as shown in fig. 5, where the electronic device includes: a processor (processor) 501, a memory (memory) 502, and a bus 503;
wherein, the processor 501 and the memory 502 complete communication with each other through a bus 503;
the processor 501 is configured to invoke the program instructions in the memory 502 to perform the methods provided in the above method embodiments, for example, including:
Acquiring credit card transaction data to be detected;
Performing anomaly detection on the credit card transaction data to be detected based on a preset credit card anomaly transaction detection model to obtain a credit card anomaly transaction detection result;
The preset credit card abnormal transaction detection model is obtained by training a local abnormal factor algorithm which performs partition data balancing processing in advance based on credit card abnormal transaction detection sample data.
The present embodiment discloses a computer program product comprising a computer program stored on a non-transitory computer readable storage medium, the computer program comprising program instructions which, when executed by a computer, are capable of performing the methods provided by the above-described method embodiments, for example comprising:
Acquiring credit card transaction data to be detected;
Performing anomaly detection on the credit card transaction data to be detected based on a preset credit card anomaly transaction detection model to obtain a credit card anomaly transaction detection result;
The preset credit card abnormal transaction detection model is obtained by training a local abnormal factor algorithm which performs partition data balancing processing in advance based on credit card abnormal transaction detection sample data.
The present embodiment provides a computer-readable storage medium storing a computer program that causes the computer to execute the methods provided by the above-described method embodiments, for example, including:
Acquiring credit card transaction data to be detected;
Performing anomaly detection on the credit card transaction data to be detected based on a preset credit card anomaly transaction detection model to obtain a credit card anomaly transaction detection result;
The preset credit card abnormal transaction detection model is obtained by training a local abnormal factor algorithm which performs partition data balancing processing in advance based on credit card abnormal transaction detection sample data.
It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and the like) having computer-usable program code embodied therein.
The present invention is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
In the description of the present specification, reference to the terms "one embodiment," "one particular embodiment," "some embodiments," "for example," "an example," "a particular example," or "some examples," etc., means that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiments or examples. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
The foregoing description of the embodiments has been provided for the purpose of illustrating the general principles of the invention, and is not meant to limit the scope of the invention, but to limit the invention to the particular embodiments, and any modifications, equivalents, improvements, etc. that fall within the spirit and principles of the invention are intended to be included within the scope of the invention.
Claims (10)
1. A method for detecting abnormal transactions on a credit card, comprising:
Acquiring credit card transaction data to be detected;
Performing anomaly detection on the credit card transaction data to be detected based on a preset credit card anomaly transaction detection model to obtain a credit card anomaly transaction detection result;
The preset credit card abnormal transaction detection model is obtained by training a local abnormal factor algorithm which performs partition data balancing processing in advance based on credit card abnormal transaction detection sample data.
2. The method for detecting abnormal credit card transactions according to claim 1, wherein the performing abnormal detection on the credit card transaction data to be detected based on a preset abnormal credit card transaction detection model to obtain a credit card abnormal transaction detection result includes:
if the local abnormality factor output by the preset credit card abnormal transaction detection model is greater than 1, determining that the credit card abnormal transaction detection result is abnormal;
And if the local abnormal factor output by the preset credit card abnormal transaction detection model is less than or equal to 1, determining that the credit card abnormal transaction detection result is normal.
3. The credit card abnormal transaction detection method according to claim 1, wherein training a local abnormality factor algorithm that performs a division data balancing process in advance based on credit card abnormal transaction detection sample data to obtain the preset credit card abnormal transaction detection model includes:
acquiring multidimensional credit card transaction data, and performing dimension reduction processing on the multidimensional credit card transaction data to obtain target dimension credit card transaction data;
Carrying out partition data balancing processing on the target dimension credit card transaction data to obtain a first partition data set and a second partition data set;
Calculating according to the first partition data set and the second partition data set to obtain a local abnormal factor predicted value;
and training a local anomaly factor algorithm according to the local anomaly factor predicted value and the local anomaly factor true value to obtain the preset credit card anomaly transaction detection model.
4. The method for detecting abnormal credit card transactions according to claim 3, wherein said performing a partition data balancing process on said target dimension credit card transaction data to obtain a first partition data set and a second partition data set includes:
Carrying out partition data quantity equalization processing and partition diameter minimization processing on the target dimension credit card transaction data;
the method comprises the steps of taking a partition data quantity equalization processing result and a partition diameter minimization processing result as targets, and solving to obtain an optimal segmentation bit;
And taking the credit card transaction data with the target dimension smaller than the optimal dividing bit as a first partition data set, and taking the credit card transaction data with the target dimension larger than or equal to the optimal dividing bit as a second partition data set.
5. The credit card anomaly transaction detection method of claim 4, wherein performing zone diameter minimization processing on the target dimension credit card transaction data comprises:
calculating the average value between the maximum value and the minimum value of the credit card transaction data of the target dimension;
And carrying out partition diameter minimization processing on the credit card transaction data of the target dimension according to the average value, the maximum value and the minimum value.
6. A credit card abnormal transaction detection method according to claim 3, wherein said calculating a local anomaly factor predictor from said first partition data set and said second partition data set includes:
according to the first partition data set and the second partition data set, performing double pruning acceleration data retrieval calculation by using a k nearest neighbor algorithm and a range query algorithm to obtain a k distance of a data object;
and calculating the local anomaly factor predicted value according to the kth distance of the data object.
7. The method for detecting abnormal credit card transactions according to claim 6, wherein said calculating said local anomaly factor predictive value based on a kth distance of a data object includes:
calculating a kth distance neighborhood of the data object according to the kth distance of the data object;
calculating a kth reachable distance of the data object according to the Euclidean distance and the kth distance of the data object;
Calculating to obtain the local reachable density of the data object according to the data object capacity of the kth distance neighborhood corresponding to the kth distance neighborhood and the kth reachable distance of the data object;
and calculating the local abnormal factor predicted value according to the local reachable density of the data object and the capacity of the kth distance neighborhood data object.
8. A credit card abnormal transaction detection device, comprising:
The acquisition unit is used for acquiring credit card transaction data to be detected;
The detection unit is used for carrying out anomaly detection on the credit card transaction data to be detected based on a preset credit card anomaly transaction detection model to obtain a credit card anomaly transaction detection result;
The preset credit card abnormal transaction detection model is obtained by training a local abnormal factor algorithm which performs partition data balancing processing in advance based on credit card abnormal transaction detection sample data.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the steps of the method of any one of claims 1 to 7 when the computer program is executed by the processor.
10. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the steps of the method according to any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410090783.0A CN117911034A (en) | 2024-01-22 | 2024-01-22 | Credit card abnormal transaction detection method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410090783.0A CN117911034A (en) | 2024-01-22 | 2024-01-22 | Credit card abnormal transaction detection method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117911034A true CN117911034A (en) | 2024-04-19 |
Family
ID=90689308
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410090783.0A Pending CN117911034A (en) | 2024-01-22 | 2024-01-22 | Credit card abnormal transaction detection method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117911034A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TWI883938B (en) * | 2024-04-26 | 2025-05-11 | 台灣大哥大股份有限公司 | Abnormal transaction detection method and abnormal transaction detection device |
-
2024
- 2024-01-22 CN CN202410090783.0A patent/CN117911034A/en active Pending
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| TWI883938B (en) * | 2024-04-26 | 2025-05-11 | 台灣大哥大股份有限公司 | Abnormal transaction detection method and abnormal transaction detection device |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US20220284261A1 (en) | Training-support-based machine learning classification and regression augmentation | |
| CN113723514B (en) | Mixed sampling-based secure access log data balance processing method | |
| CN115344693B (en) | Clustering method based on fusion of traditional algorithm and neural network algorithm | |
| CN112668482A (en) | Face recognition training method and device, computer equipment and storage medium | |
| CN109902192A (en) | Remote sensing image retrieval method, system, equipment and medium based on unsupervised depth regression | |
| CN113221065A (en) | Data density estimation and regression method, corresponding device, electronic device, and medium | |
| CN111797899A (en) | A kmeans clustering method and system for low-voltage station area | |
| CN117975204B (en) | A model training method, defect detection method and related device | |
| CN118468061A (en) | Automatic algorithm matching and parameter optimizing method and system | |
| CN120197141B (en) | A multi-modal, multi-source, heterogeneous data fusion method and system | |
| CN114611976A (en) | Power consumer behavior portrait method, system and device | |
| CN117911034A (en) | Credit card abnormal transaction detection method and device | |
| CN119152242A (en) | Prototype weighted comparison learning image recognition method, system, equipment and medium | |
| Rahman et al. | An efficient approach for selecting initial centroid and outlier detection of data clustering | |
| CN118154936B (en) | Machine learning-based variety identification and classification method and system | |
| CN117609529B (en) | Substitution search method and system for electronic components | |
| CN112861974A (en) | Text classification method and device, electronic equipment and storage medium | |
| CN117689979A (en) | Sample generation method, device and storage medium | |
| CN114154590B (en) | User clustering method, device, terminal equipment and storage medium | |
| Yang et al. | Adaptive density peak clustering for determinging cluster center | |
| CN114332847A (en) | Depth image clustering method, system and device based on active selection constraint | |
| Tchaye-Kondi et al. | A new hashing based nearest neighbors selection technique for big datasets | |
| CN119691507B (en) | Unknown target classification method based on zero-order learning | |
| CN116543302B (en) | Parallel DCNN fast scene matching and recognition method based on Spark and feature clustering | |
| CN117971490B (en) | Computing resource optimization method and system based on channel performance intelligent analysis |
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 |