CN109726735A - A kind of mobile applications recognition methods based on K-means cluster and random forests algorithm - Google Patents
A kind of mobile applications recognition methods based on K-means cluster and random forests algorithm Download PDFInfo
- Publication number
- CN109726735A CN109726735A CN201811422296.0A CN201811422296A CN109726735A CN 109726735 A CN109726735 A CN 109726735A CN 201811422296 A CN201811422296 A CN 201811422296A CN 109726735 A CN109726735 A CN 109726735A
- Authority
- CN
- China
- Prior art keywords
- data
- clustering
- cluster
- packet
- random forests
- 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.)
- Withdrawn
Links
Landscapes
- Data Exchanges In Wide-Area Networks (AREA)
Abstract
A kind of mobile applications recognition methods based on K-means cluster and random forests algorithm, the encrypting traffic of a period is turned into several data flows according to TCP session feature is discrete first, each data flow is indicated using input packet time sequence, output grouping time series and input and output packet time sequence；Three time serieses corresponding to every data stream carry out mathematical statistics again, obtain the statistical nature of data packet；Later by K-means clustering algorithm to the statistical nature clustering of encrypting traffic；And by the marking of the purity of calculation method each clustering cluster obtained to clustering of entropy, the sample in the lower clustering cluster of purity is filtered out；Filtered clustering cluster is modeled as data set finally by algorithm of standing abreast at random, realizes the identification to encryption Liu Suoshu mobile application type.This method combines supervised learning and unsupervised learning, realizes and accurately identifies different mobile application types in the miscellaneous encryption flow of application type.
Description
Technical field
The invention belongs to field of information security technology, and in particular to one kind is based on K-means cluster and random forests algorithm
Mobile applications recognition methods.
Background technique
In recent years, as the hardware performance of Intelligent mobile equipment is substantially improved, software function becomes increasingly abundant, and intelligent mobile is set
Standby usage amount is also in sustainable growth.People carry smart phone at any time, and basic voice communication is completed by mobile phone
And the daily communication activity such as short message communication, and the relevant Email in electric internet, social networks.These portable equipments
Just save largely sensitive information relevant to privacy of user.Nowadays most of mobile applications all use SSL/TLS
Agreement is encrypted.Nonetheless, attacker can also be inferred to the sensitive letter of user indirectly by the analysis to encryption flow
Breath.
At the same time, the application program in information security field in identification intelligent equipment and identification intelligent equipment itself side
There are many correlative studys in face.In the case where encrypting environment, merely with data flow, data packet length and some relevant to packet length
Statistical property can effectively realize Application Type identification in encryption flow.There is document to propose to utilize the branch in supervised learning
Holding vector machine and random forests algorithm respectively realizes the identification of 110 kinds of application programs in Google Play.Since different are answered
The data flow of parallel pattern may be generated with program.So data flow quite similar in these different applications is not enough to
For distinguishing application program, and these have the similar data flow of different labels that can hinder our supervised learnings to a certain extent
The study of algorithm.So those classifiers are determined that the lower sample of prediction probability regards by setting " prediction probability threshold value " by author
For above-mentioned interference sample, this kind of sample is not learnt and predicted.But prediction probability is not high, in addition to presentation class device is to this
Item decision does not have except enough certainty, it is also possible to it is perfect to mean that classifier learns not yet, it is thus possible to interfere
The erroneous judgement problem of sample.
Summary of the invention
It is a kind of based on K-means cluster the technical problem to be solved by the present invention is to overcome the deficiencies of the prior art and provide
With the mobile applications recognition methods of random forests algorithm, the mode that clustering algorithm and comentropy combine is interfered to filter
Sample reasonably avoids and does not cause the problem of judging interference sample by accident since classifier learns perfect.
The present invention provides a kind of mobile applications recognition methods based on K-means cluster and random forests algorithm, packet
Include following steps:
Step S1, the encrypting traffic of a period is turned into several data flows according to TCP session feature is discrete, often
A data flow is indicated using input packet time sequence, output grouping time series and input and output packet time sequence；
Step S2, three time serieses corresponding to every data stream carry out mathematical statistics, and the statistics for obtaining data packet is special
Sign；
Step S3, by K-means clustering algorithm to the statistical nature clustering of encrypting traffic；
Step S4, it is given a mark, and filtered by the purity of the calculation method of entropy each clustering cluster obtained to clustering
Fall the sample in the lower clustering cluster of purity；
Step S5, filtered clustering cluster is modeled by standing abreast algorithm at random as data set, is realized to encryption
The identification of Liu Suoshu mobile application type.
As further technical solution of the present invention, specific step is as follows for discretization in step S1:
Step S11, by continuous refined net flow discretization as unit of burst, happening suddenly, it is specified to be less than for time interval
The grouping of threshold value；
Step S12, multiple encrypting traffics are isolated from each burst, encrypting traffic is by a burst and together
The relevant grouping composition of a pair of of quaternary ancestral；
Step S13, every data stream is indicated with three brother's packet time sequences, by each grouping flowed into data flow
The sequence that is sequentially arranged of packet length, as input packet time sequence；By each grouping for being flowed out in data flow
The sequence that packet length sorts in chronological order, as output grouping time series；By each of being flowed in and out in data flow point
The sequence that group is sequentially arranged, as input and output packet time sequence.
Further, specific step is as follows by step S2:
Step S21, each packet time sequence corresponding to every data stream carries out statistical nature extraction, statistical nature packet
It is absolutely inclined to include data packet length minimum value, data packet length maximum value, data packet length average value, data packet length median
Difference, data packet length standard deviation, data packet length variance, data packet length deflection, data packet length kurtosis, data packet length
Data packet number in percentile (from 10% to 90%) and the packet time sequence totally 18 statistical natures；
Step S22, corresponding by input packet time sequence by the corresponding statistical nature of above-mentioned each packet time sequence
Statistical nature, the corresponding statistical nature of output grouping time series and the corresponding statistical nature of input and output packet time sequence
Sequence be combined into length be 54 encrypting traffic feature vector；
Step S23, every data stream is handled by step 22, until all Data Stream Processings finish.
Further, specific step is as follows by step S3:
Step S31, it is searched for by line style and chooses clustering cluster quantity constant k；
Step S32, it is modeled using constant k as parameter by K-means clustering algorithm；
Step S33, the Dunn index and silhouette coefficient of cluster result are obtained, Clustering Effect is assessed；
Step S34, circulation step S31- step S33 is until Clustering Effect reaches best.
Further, specific step is as follows by step S4:
Step S41, the comentropy of each clustering cluster is calculated by comentropy calculation formula；
Step S42, entropy threshold is set, and filtering is more than the sample of the clustering cluster of the entropy of threshold value；
Step S43, it is modeled by random forests algorithm；
Step S44, circulation appeal step, adjustment entropy threshold are until the classifying quality of random forests algorithm model is best.
Further, specific step is as follows by step S5:
Step S51, will be randomly divided by the data set of step 3 and step 4 data prediction training set, verifying collection and
Three parts of test set；
Step S52, using random forests algorithm using training set as data training classifier；
Step S53, classifier is detected to the effect of mobile application type identification with verifying collection；
Step S54, it adjusts base learner quantity in random forest, choose measurement index of attribute node etc. in base learner
Parameter；
Step S55, the application type recognition effect that circulation step S52 and step S53 collects verifying up to classifier is best,
Finally with the recognition effect of test set detection final mask.
The present invention by machine learning unsupervised learning and supervised learning combine, propose a kind of combining information entropy
The mobile application kind identification method of clustering cluster purity marking thought, and tentatively achieve ideal experimental result.K-
Means clustering algorithm will characterize similar encrypting traffic and be aggregated in the same cluster, carry out to data set effective preliminary
Analysis.Comentropy clustering cluster purity marking thought, realizes the filtering to similar encrypting traffic caused by different application, keeps away
Exempt to learn not perfect due to classifier and cause the problem of judging interference sample by accident, established for the accurate study of subsequent classification algorithm
Basis.
It is compared with existing scheme, since method proposed by the present invention uses K-means clustering algorithm and comentropy cluster
Cluster purity marking thought, reduces the erroneous judgement to interference sample, improves the accuracy rate of mobile application identification.
Detailed description of the invention
Fig. 1 is flow diagram of the invention；
Fig. 2 is the schematic diagram of encryption flow discretization process of the invention.
Fig. 3 is data set sample information entropy distribution map of the invention.
Specific embodiment
Referring to Fig. 1, the present embodiment provides a kind of mobile applications based on K-means cluster and random forests algorithm
Recognition methods, as shown in Figure 1, method includes the following steps:
Step 1: encrypting traffic is indicated with packet time sequence
It is indicated by encrypting traffic discretization and in the form of three packet time sequences, the specific steps are as follows:
1.1, by continuous refined net flow discretization as unit of burst.Burst refers to the small Mr. Yu of adjacent time interval
A series of groupings of a threshold value；
1.2, multiple encrypting traffics are isolated from each burst.It is related to same a pair of of four-tuple in a burst
Grouping form a data stream；
1.3, every data stream is indicated with three packet time sequences.Three time serieses are respectively as follows: (1) by data flow
The sequence that the packet length of each grouping of middle inflow is sequentially arranged；(2) by the packet for each grouping flowed out in data flow
The sequence that length is sequentially arranged；(3) sequence that each grouping flowed in and out in data flow is sequentially arranged.
As shown in Fig. 2, in one section of encryption flow, as long as the time interval of a grouping and previous grouping is less than burst
Threshold value, then the grouping is just divided into the same burst with previous grouping.If the previous packet time of some grouping and it
Interval is greater than burst threshold, then the grouping from the grouping is reclassified as next burst.In a burst, with same a pair
The relevant grouping of four-tuple forms a data flow.Data flow and the difference of TCP session be, TCP session may be across
More multiple bursts, and a data flow is the part TCP session content in some burst, a burst may also include multiple
Data flow.
Step 2: the statistical nature of encrypting traffic is extracted
Statistical nature extraction is carried out for each data flow sample in multiple data sets, the specific steps are as follows:
2.1, each packet time sequence corresponding to every data stream carries out statistical nature extraction.Wherein statistical nature
Including data packet length minimum value, data packet length maximum value, data packet length average value, data packet length median is absolutely inclined
Difference, data packet length standard deviation, data packet length variance, data packet length deflection, data packet length kurtosis, data packet length
Data packet number in percentile (from 10% to 90%) and the packet time sequence totally 18 statistical natures；
2.2, by the corresponding statistical nature of above-mentioned each packet time sequence, by the corresponding statistics of input packet time sequence
Feature, the corresponding statistical nature of output grouping time series and the corresponding statistical nature of input and output packet time sequence it is suitable
Sequence is combined into the encrypting traffic feature vector that length is 54；
2.3, the processing for every step of data stream application 2.1 and 2.2 that data are concentrated, until entire data set has been handled
Finish.
Step 3: K-means cluster is carried out to data set
K-means cluster is carried out to every feature vector by step 2 processing, adjustment clustering parameter imitates cluster
Fruit is best, the specific steps are as follows:
3.1, clustering cluster quantity constant k is chosen by linear search；
3.2, it is modeled using k as parameter using K-means clustering algorithm；
3.3, the Dunn index and silhouette coefficient of cluster result are calculated, Clustering Effect is assessed；
3.4, above step is repeated until Clustering Effect is best.
Step 4: the comentropy marking and filtering of clustering cluster
The comentropy of each clustering cluster generated in step 3, setting information entropy threshold are calculated, filtering is higher than entropy threshold
Clustering cluster, the specific steps are as follows:
4.1, the comentropy of each clustering cluster is calculated using comentropy calculation formula；
4.2, entropy threshold is set, sample of the entropy more than the clustering cluster of entropy threshold is filtered out；
4.3, subsequent random forests algorithm modeling is carried out；
4.4, it repeats the above steps, adjustment entropy threshold is until make subsequent random forest grader effect best.
If ratio shared by kth class sample is p in current sample set Dk(k=1,2 ..., | Y |), then the comentropy of D
It is defined as
The value of Ent (D) is smaller, then the purity of D is higher.The comentropy of each clustering cluster is calculated using above-mentioned formula, is clustered
Cluster comentropy bar shaped distribution map is as shown in Figure 3.Tradeoff data set utilization rate and classification accuracy can obtain, and information entropy threshold takes 3.0
Effect is best.
Step 5: training random forest grader
It will be used for the training of random forest grader by the data set of step 4 processing, ultimately generates mobile application type
Identification model, the specific steps are as follows:
5.1, training set will be randomly divided by the data set of step 3 and step 4 data prediction, verifying collects and test
Collect three parts；
5.2, using random forests algorithm using training set as data training classifier；
5.3, classifier is detected to the effect of mobile application type identification with verifying collection；
5.4, it adjusts base learner quantity in random forest, choose the ginseng such as measurement index of attribute node in base learner
Number；
5.5, the application type recognition effect that repetition 5.2 and 5.3 collects verifying up to classifier is best, finally uses test set
Detect the recognition effect of final mask.
In conclusion the invention proposes a kind of introducing K-means clustering algorithms and the marking of comentropy clustering cluster purity to think
The mobile application kind identification method thought calculates filtering interference sample by clustering and clustering cluster comentropy, to reduce
The erroneous judgement of interference sample may, improve application type recognition accuracy.
The basic principles, main features and advantages of the invention have been shown and described above.Those skilled in the art should
Understand, the present invention do not limited by above-mentioned specific embodiment, the description in above-mentioned specific embodiment and specification be intended merely into
One step illustrates the principle of the present invention, and under the premise of not departing from spirit of that invention range, the present invention also has various change and changes
Into these changes and improvements all fall within the protetion scope of the claimed invention.The scope of protection of present invention is by claim
Book and its equivalent thereof.
Claims (6)
1. a kind of mobile applications recognition methods based on K-means cluster and random forests algorithm, which is characterized in that including
Following steps:
Step S1, the encrypting traffic of a period is turned into several data flows, every number according to TCP session feature is discrete
It is indicated according to stream using input packet time sequence, output grouping time series and input and output packet time sequence；
Step S2, three time serieses corresponding to every data stream carry out mathematical statistics, obtain the statistical nature of data packet；
Step S3, by K-means clustering algorithm to the statistical nature clustering of encrypting traffic；
Step S4, it is given a mark, and filtered out pure by the purity of the calculation method of entropy each clustering cluster obtained to clustering
Spend the sample in lower clustering cluster；
Step S5, filtered clustering cluster is modeled by standing abreast algorithm at random as data set, is realized to encryption Liu institute
State the identification of mobile application type.
2. a kind of mobile applications based on K-means cluster and random forests algorithm according to claim 1 identify
Method, which is characterized in that specific step is as follows for discretization in the step S1:
Step S11, it by continuous refined net flow discretization as unit of burst, happens suddenly and is less than specified threshold for time interval
Grouping；
Step S12, multiple encrypting traffics are isolated from each burst, encrypting traffic is by a burst and the same as a pair of
The relevant grouping composition of quaternary ancestral；
Step S13, every data stream is indicated with three brother's packet time sequences, by the packet of each grouping flowed into data flow
The sequence that length is sequentially arranged, as input packet time sequence；Packet by each grouping flowed out in data flow is long
Spend the sequence to sort in chronological order, as output grouping time series；By each grouping for being flowed in and out in data flow by
The sequence of time sequencing arrangement, as input and output packet time sequence.
3. a kind of mobile applications based on K-means cluster and random forests algorithm according to claim 1 identify
Method, which is characterized in that specific step is as follows by the step S2:
Step S21, each packet time sequence corresponding to every data stream carries out statistical nature extraction, and statistical nature includes number
According to packet length minimum value, data packet length maximum value, data packet length average value, data packet length median absolute deviation, number
According to packet length standard deviation, data packet length variance, data packet length deflection, data packet length kurtosis, data packet length percentage
Data packet number in digit (from 10% to 90%) and the packet time sequence totally 18 statistical natures；
Step S22, by the corresponding statistical nature of above-mentioned each packet time sequence, by the corresponding statistics of input packet time sequence
Feature, the corresponding statistical nature of output grouping time series and the corresponding statistical nature of input and output packet time sequence it is suitable
Sequence is combined into the encrypting traffic feature vector that length is 54；
Step S23, every data stream is handled by step 22, until all Data Stream Processings finish.
4. a kind of mobile applications based on K-means cluster and random forests algorithm according to claim 1 identify
Method, which is characterized in that specific step is as follows by the step S3:
Step S31, it is searched for by line style and chooses clustering cluster quantity constant k；
Step S32, it is modeled using constant k as parameter by K-means clustering algorithm；
Step S33, the Dunn index and silhouette coefficient of cluster result are obtained, Clustering Effect is assessed；
Step S34, circulation step S31- step S33 is until Clustering Effect reaches best.
5. a kind of mobile applications based on K-means cluster and random forests algorithm according to claim 1 identify
Method, which is characterized in that specific step is as follows by the step S4:
Step S41, the comentropy of each clustering cluster is calculated by comentropy calculation formula；
Step S42, entropy threshold is set, and filtering is more than the sample of the clustering cluster of the entropy of threshold value；
Step S43, it is modeled by random forests algorithm；
Step S44, circulation appeal step, adjustment entropy threshold are until the classifying quality of random forests algorithm model is best.
6. a kind of mobile applications based on K-means cluster and random forests algorithm according to claim 1 identify
Method, which is characterized in that specific step is as follows by the step S5:
Step S51, training set will be randomly divided by the data set of step 3 and step 4 data prediction, verifying collects and test
Collect three parts；
Step S52, using random forests algorithm using training set as data training classifier；
Step S53, classifier is detected to the effect of mobile application type identification with verifying collection；
Step S54, it adjusts base learner quantity in random forest, choose the ginseng such as measurement index of attribute node in base learner
Number；
Step S55, circulation step S52 and step S53 is until classifier is best to the application type recognition effect of verifying collection, finally
With the recognition effect of test set detection final mask.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811422296.0A CN109726735A (en) | 2018-11-27 | 2018-11-27 | A kind of mobile applications recognition methods based on K-means cluster and random forests algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811422296.0A CN109726735A (en) | 2018-11-27 | 2018-11-27 | A kind of mobile applications recognition methods based on K-means cluster and random forests algorithm |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN109726735A true CN109726735A (en) | 2019-05-07 |
Family
ID=66295574
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811422296.0A Withdrawn CN109726735A (en) | 2018-11-27 | 2018-11-27 | A kind of mobile applications recognition methods based on K-means cluster and random forests algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109726735A (en) |
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110113349A (en) * | 2019-05-15 | 2019-08-09 | 北京工业大学 | A kind of malice encryption traffic characteristics analysis method |
| CN110348526A (en) * | 2019-07-15 | 2019-10-18 | 武汉绿色网络信息服务有限责任公司 | A kind of device type recognition methods and device based on semi-supervised clustering algorithm |
| CN110460502A (en) * | 2019-09-10 | 2019-11-15 | 西安电子科技大学 | Application rs traffic recognition methods under VPN based on distribution characteristics random forest |
| CN110493142A (en) * | 2019-07-05 | 2019-11-22 | 南京邮电大学 | Mobile applications Activity recognition method based on spectral clustering and random forests algorithm |
| CN110674010A (en) * | 2019-09-10 | 2020-01-10 | 西安电子科技大学 | Intelligent device application program identification method based on session length probability distribution |
| CN111058829A (en) * | 2019-12-05 | 2020-04-24 | 中国矿业大学 | Rock stratum analysis method based on image processing |
| CN111737374A (en) * | 2019-06-21 | 2020-10-02 | 腾讯科技（深圳）有限公司 | Position coordinate determination method and device, electronic equipment and storage medium |
| WO2020228527A1 (en) * | 2019-05-14 | 2020-11-19 | 华为技术有限公司 | Data stream classification method and message forwarding device |
| CN113420876A (en) * | 2021-06-29 | 2021-09-21 | 平安科技（深圳）有限公司 | Real-time operation data processing method, device and equipment based on unsupervised learning |
| CN113553499A (en) * | 2021-06-22 | 2021-10-26 | 杭州摸象大数据科技有限公司 | Cheating detection method and system based on marketing fission and electronic equipment |
| CN114500387A (en) * | 2022-02-14 | 2022-05-13 | 重庆邮电大学 | Mobile application traffic identification method and system based on machine learning |
| CN114866485A (en) * | 2022-03-11 | 2022-08-05 | 南京华飞数据技术有限公司 | Network traffic classification method and system based on aggregation entropy |
| CN116582363A (en) * | 2023-07-12 | 2023-08-11 | 江苏政采数据科技有限公司 | Industrial protocol based detection method for transmission flow abnormal attack |
| CN116936132A (en) * | 2023-09-18 | 2023-10-24 | 深圳市即达健康医疗科技有限公司 | Intelligent medical condition monitoring method and system based on big data |
-
2018
- 2018-11-27 CN CN201811422296.0A patent/CN109726735A/en not_active Withdrawn
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2020228527A1 (en) * | 2019-05-14 | 2020-11-19 | 华为技术有限公司 | Data stream classification method and message forwarding device |
| US12003609B2 (en) | 2019-05-14 | 2024-06-04 | Huawei Technologies Co., Ltd. | Data flow classification method and packet forwarding device |
| CN110113349A (en) * | 2019-05-15 | 2019-08-09 | 北京工业大学 | A kind of malice encryption traffic characteristics analysis method |
| CN111737374B (en) * | 2019-06-21 | 2023-05-09 | 腾讯科技（深圳）有限公司 | Position coordinate determination method, device, electronic equipment and storage medium |
| CN111737374A (en) * | 2019-06-21 | 2020-10-02 | 腾讯科技（深圳）有限公司 | Position coordinate determination method and device, electronic equipment and storage medium |
| CN110493142A (en) * | 2019-07-05 | 2019-11-22 | 南京邮电大学 | Mobile applications Activity recognition method based on spectral clustering and random forests algorithm |
| CN110348526A (en) * | 2019-07-15 | 2019-10-18 | 武汉绿色网络信息服务有限责任公司 | A kind of device type recognition methods and device based on semi-supervised clustering algorithm |
| CN110460502A (en) * | 2019-09-10 | 2019-11-15 | 西安电子科技大学 | Application rs traffic recognition methods under VPN based on distribution characteristics random forest |
| CN110674010A (en) * | 2019-09-10 | 2020-01-10 | 西安电子科技大学 | Intelligent device application program identification method based on session length probability distribution |
| CN110460502B (en) * | 2019-09-10 | 2022-03-04 | 西安电子科技大学 | Application program flow identification method under VPN based on distributed feature random forest |
| CN111058829A (en) * | 2019-12-05 | 2020-04-24 | 中国矿业大学 | Rock stratum analysis method based on image processing |
| CN111058829B (en) * | 2019-12-05 | 2021-06-25 | 中国矿业大学 | Rock stratum analysis method based on image processing |
| CN113553499A (en) * | 2021-06-22 | 2021-10-26 | 杭州摸象大数据科技有限公司 | Cheating detection method and system based on marketing fission and electronic equipment |
| CN113420876B (en) * | 2021-06-29 | 2023-10-27 | 平安科技（深圳）有限公司 | Method, device and equipment for processing real-time operation data based on unsupervised learning |
| CN113420876A (en) * | 2021-06-29 | 2021-09-21 | 平安科技（深圳）有限公司 | Real-time operation data processing method, device and equipment based on unsupervised learning |
| CN114500387A (en) * | 2022-02-14 | 2022-05-13 | 重庆邮电大学 | Mobile application traffic identification method and system based on machine learning |
| CN114866485A (en) * | 2022-03-11 | 2022-08-05 | 南京华飞数据技术有限公司 | Network traffic classification method and system based on aggregation entropy |
| CN114866485B (en) * | 2022-03-11 | 2023-09-29 | 南京华飞数据技术有限公司 | Network traffic classification method and classification system based on aggregation entropy |
| CN116582363A (en) * | 2023-07-12 | 2023-08-11 | 江苏政采数据科技有限公司 | Industrial protocol based detection method for transmission flow abnormal attack |
| CN116936132A (en) * | 2023-09-18 | 2023-10-24 | 深圳市即达健康医疗科技有限公司 | Intelligent medical condition monitoring method and system based on big data |
| CN116936132B (en) * | 2023-09-18 | 2024-04-26 | 深圳市即达健康医疗科技有限公司 | Intelligent medical condition monitoring method and system based on big data |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109726735A (en) | A kind of mobile applications recognition methods based on K-means cluster and random forests algorithm | |
| CN107330364B (en) | A kind of people counting method and system based on cGAN network | |
| Kasai et al. | Network volume anomaly detection and identification in large-scale networks based on online time-structured traffic tensor tracking | |
| CN104935600B (en) | A kind of mobile ad-hoc network intrusion detection method and equipment based on deep learning | |
| CN105718960B (en) | Based on convolutional neural networks and the matched image order models of spatial pyramid | |
| CN109582855B (en) | Enhance the anti-method, apparatus for climbing system identification performance and storage medium | |
| CN112381121A (en) | Unknown class network flow detection and identification method based on twin network | |
| CN109587350A (en) | A kind of sequence variation detection method of the telecommunication fraud phone based on sliding time window polymerization | |
| CN103200133A (en) | Flow identification method based on network flow gravitation cluster | |
| CN108540338B (en) | Application layer communication protocol identification method based on deep cycle neural network | |
| CN111385297A (en) | Wireless device fingerprint identification method, system, device and readable storage medium | |
| CN110377605A (en) | A kind of Sensitive Attributes identification of structural data and classification stage division | |
| CN108256449A (en) | A kind of Human bodys' response method based on subspace grader | |
| CN107154923A (en) | A kind of network inbreak detection method based on the very fast learning machine of multilayer | |
| CN110493142A (en) | Mobile applications Activity recognition method based on spectral clustering and random forests algorithm | |
| CN110149531A (en) | The method and apparatus of video scene in a kind of identification video data | |
| Liu et al. | A binary-classification-tree based framework for distributed target classification in multimedia sensor networks | |
| CN116150688A (en) | Lightweight Internet of things equipment identification method and device in smart home | |
| CN110414431A (en) | Face identification method and system based on elastic context relation loss function | |
| CN112383488B (en) | Content identification method suitable for encrypted and non-encrypted data streams | |
| CN112291226B (en) | Method and device for detecting abnormity of network flow | |
| CN109450876A (en) | A kind of DDos recognition methods and system based on various dimensions state-transition matrix feature | |
| CN117527295A (en) | Self-adaptive network threat detection system based on artificial intelligence | |
| Ma et al. | A Multi-Perspective Feature Approach to Few-Shot Classification of IoT Traffic | |
| Abdalla et al. | Impact of packet inter-arrival time features for online peer-to-peer (P2P) classification |
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 | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20190507 |