CN110460463A - Service quality prediction technique and its system based on deep neural network - Google Patents
Service quality prediction technique and its system based on deep neural network Download PDFInfo
- Publication number
- CN110460463A CN110460463A CN201910671422.4A CN201910671422A CN110460463A CN 110460463 A CN110460463 A CN 110460463A CN 201910671422 A CN201910671422 A CN 201910671422A CN 110460463 A CN110460463 A CN 110460463A
- Authority
- CN
- China
- Prior art keywords
- information
- module
- deep neural
- neural network
- network
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L41/00—Arrangements for maintenance, administration or management of data switching networks, e.g. of packet switching networks
- H04L41/14—Network analysis or design
- H04L41/145—Network analysis or design involving simulating, designing, planning or modelling of a network
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
- H04L43/00—Arrangements for monitoring or testing data switching networks
- H04L43/50—Testing arrangements
- H04L43/55—Testing of service level quality, e.g. simulating service usage
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Computing Systems (AREA)
- General Physics & Mathematics (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Quality & Reliability (AREA)
- Data Exchanges In Wide-Area Networks (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Compression, Expansion, Code Conversion, And Decoders (AREA)
- Computer And Data Communications (AREA)
Abstract
The invention discloses a kind of service quality prediction technique and its system based on deep neural network, are related to web services technologies field.Method includes the following steps: input request contexts variable information, and encoded by entity expression matrix in coding module, to obtain embedded request matrix；Feature extraction will be compressed and is carried out by the context variant information of coding, to obtain the embedded request vector according to timing of the request contexts variable information；The embedded request vector is sequentially input to LSTM network module, to export the deep neural network to the prediction data information for inputting request contexts variable information next time according to the timing；The prediction data information is perceived, to obtain the decoded prediction data information, the decoded prediction data information is reduced to quality of service data and is exported.
Description
Technical field
The present invention relates to web services technologies fields, pre- more particularly to a kind of service quality based on deep neural network
Survey method and its system.
Background technique
Fierce increase had occurred in the past during the decade for the quantity of WebService in internet, as public Web takes
The application of the increase for quantity of being engaged in, SOA (Service Oriented Architacture, Enterprise SOA) framework also becomes
It obtains very extensive.However, the network environment of internet complexity makes the state of service ever-changing, therefore effective service quality
Prediction just seems particularly significant at this time.
Traditional services qualitative forecasting method is mostly based on CF model (Collaborative Filtering, collaborative filtering mould
Type) and MF model (Matrix factorization, matrix decomposition model), the context data that can be utilized relatively has
Limit.Meanwhile both methods is difficult to carry out simply to extend to support polymorphic type QoS (Quality of Service, Service Quality
Amount) data prediction.It also lacks corresponding modeling means to the implicit variable for influencing QoS response sequence.
Summary of the invention
The service quality prediction technique based on deep neural network that the main purpose of the present invention is to provide a kind of and its it is
System, it is intended to the invisible factor for influencing QoS data is fitted in network, it can QoS service qualitative data to real world
It is predicted.
To achieve the above object, the present invention provides a kind of service quality prediction technique based on deep neural network, including
Following steps:
Request contexts variable information is inputted, and is encoded by entity expression matrix in coding module, it is embedding to obtain
Enter formula request matrix；
Feature extraction will be compressed and is carried out by the context variant information of coding, to obtain in the request
The hereafter embedded request vector according to timing of variable information；
According to the timing sequentially input the embedded request vector to LSTM (Long Short-Term Memory,
Shot and long term memory network) network module, to export the deep neural network to input request contexts variable information next time
Prediction data information；
The prediction data information is perceived, to obtain the decoded prediction data information, by the decoding
The prediction data information afterwards is reduced to quality of service data and exports.
Preferably, the context variant information includes context and its corresponding quality of service data information.
Preferably, the context variant information includes response time information and network throughput information；Institute after reduction
Stating quality of service data includes response time information and network throughput information.
Preferably, the step will carry out dimension compression and carry out feature to mention by the context variant information of coding
It takes further include:
After the context variant information of coding is compressed, the spy of multilayer fully-connected network composition will be input to
It levies extractor and carries out feature extraction.
Preferably, the embedded request vector is sequentially input to LSTM network module, to export according to the timing
Deep neural network is stated to the prediction data information for inputting request contexts variable information next time further include:
The embedded request vector sequentially input according to the timing successively obtains corresponding calculating knot by calculating
Fruit, using the last one calculated result in the calculated result as the deep neural network to inputting request contexts next time
The prediction data information of variable information.
Preferably, described to be reduced to quality of service data and export also wrap by the decoded prediction data information
It includes:
The prediction data information is inputted respectively in the recovery module that two fully-connected networks are constituted and is counted respectively
Calculate, with obtain include response time information and network throughput information quality of service data.
The service quality forecasting system based on deep neural network that the present invention also provides a kind of, the system comprises coding moulds
Block, compression module, LSTM network module, sensing module and recovery module；
The coding module exports embedded request matrix extremely for receiving request contexts variable information after being encoded
The compression module；
The compression module is used to be compressed and be carried out feature extraction by the context variant information of coding,
And embedded request vector is exported to the LSTM network module；
The LSTM network module calculates the embedded request vector, and exports the deep neural network pair
The prediction data information of request contexts variable information is inputted next time to the sensing module；
The sensing module perceives the prediction data information, and will obtain the decoded prediction data letter
Breath is sent to recovery module, and the decoded prediction data information is reduced to quality of service data by the recovery module
And it exports.
Preferably, the context variant information includes response time information and network throughput information；Institute after reduction
Stating quality of service data includes response time information and network throughput information.
Preferably, the recovery module includes the first recovery module and the second recovery module, when restoring response respectively
Between information and network throughput information.
The context variant information that technical solution of the present invention is collected by input, being fitted in network influences QoS data
Invisible factor, can the QoS service qualitative data to real world predict.LSTM network module can be to complexity
Unknown data is made prediction, and has certain correctness, can be used in the specific construction practices such as Services Composition.
Detailed description of the invention
Fig. 1 is that the present invention is based on the flow diagrams of the service quality prediction technique of deep neural network；
Fig. 2 is that the present invention is based on the schematic illustrations of the service quality forecasting system of deep neural network.
The embodiments will be further described with reference to the accompanying drawings for the realization, the function and the advantages of the object of the present invention.
Specific embodiment
It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention, it is not intended to limit the present invention.
The following further describes the present invention with reference to the drawings.
A kind of service quality prediction technique based on deep neural network, as shown in Figure 1, comprising the following steps:
Request contexts variable information is inputted, and is encoded by entity expression matrix in coding module, it is embedding to obtain
Enter formula request matrix.
In a particular embodiment, request contexts variable information includes numeric type and nonumeric type feature, is compiled being input to
It needs to be pre-processed respectively before code module.For continuous data, need to carry out sampling to be translated into discretization data, with
Quality of service data information is merged with request contexts information.It need to be to be asked for each for non-numerical variable
Preceding sextuple (u1, uas1, ur1, s1, sas1, sr1) in eight tuples for asking context variant information to obtain, in total entity number n=
We can be written as hexa-atomic group of (Isosorbide-5-Nitrae, 5,6,9,11) in the case where 13, be encoded to obtain N to it1=
(010011100101), i.e., to hexa-atomic X before eight tuples1=(x1, x2, x3, x4, x5, x6), by X1Corresponding N1XthiPosition
It is encoded to 1.Define the element expression matrix M of n*k dimensionn×k, wherein MiThe vectorization expression of i-th of entity is represented, k represents the table
Up to the dimension of vector, for hexa-atomic group (Isosorbide-5-Nitrae, 5,6,9,11), we select M respectively1, M4, M6, M9, M11Six k dimensional vectors
Form the matrix G of 6*k dimension1。G1It is the matrixing expression of hexa-atomic group of (Isosorbide-5-Nitrae, 5,6,9,11), expression matrix can be reserved for each
The vector of corresponding element is expressed.The operation definition is as follows:
s.t→len(N1)=n, wherein symbolAll row M are selected in expression from matrix Mi, and if only if (X1)i=1,
N is the number of all entities.
Response time and handling capacity for numeric type, we use independent two matrix Rrs×k, Tts×kAs it
Expression matrix.Wherein rs, ts are respectively the number of response time and handling capacity expression vector.For the response time, specific real
It applies in example, it is 0.01 that its step-length is arranged in discretization, and for handling capacity, it is 0.1 that its step-length, which is arranged, therefore, for arbitrary number
The response time of value and handling capacity can be mapped as corresponding integer expression, be expressed using the integer as its element
Mark.Specifically, corresponding (0.13,23.6), is expressed as (13,23), while selecting R from R matrix13, from T matrix
Select T26Form the matrix G of a 2*k2.It is directed to X2=(x7, x8), the corresponding N of the two can be obtained2、N3, select
Rn7, Tn8Form new matrix G2.The operation definition is as follows:
s.t→len(N2)=max (ResponseTime) * 100
s.t→len(N3)=max (ThroughOutput) * 10, wherein max (ResponseTime) max
(ThroughOutput) maximum value of response time and handling capacity in all data are respectively indicated.It is obtained for above two step
Expression matrix carries out direct splicing, can be obtained the expression matrix of eight tuples (Isosorbide-5-Nitrae, 5,6,9,11,13,26), G=
concat(G1, G2), the expression matrix that 8*k dimension matrix is as specifically requested.
Feature extraction will be compressed and is carried out by the context variant information of coding, to obtain in the request
The hereafter embedded request vector according to timing of variable information.
In a particular embodiment, the input for presetting network is sd=[sd1, sd2, sd3], wherein sdiIndicate LSTM network mould
The input of i-th of step of block.Wherein each sdiIt is represented by sdi=[sdi1, sdi2..., sdik], it indicates to be directed to and this time request
Expression matrix composed by k vector in context variant information i, sign extraction is by expression matrix sdiBoil down to
Amount expression Pi.As default i=3, i.e., the 4th request is predicted using first three request, then target is exactly by three-dimensional tensor
[sd1, sd2, sd3] boil down to two dimension tensor P=[P1, P2, P3]。
The embedded request vector is sequentially input to LSTM network module, to export the depth mind according to the timing
Through network to the prediction data information for inputting request contexts variable information next time.
In a particular embodiment, [P is sequentially input according to timing1, P2, P3] into LSTM network module, obtain three outputs
[h1, h2, h3], take wherein third output h3As the final output of network, here it is considered that h3Illustrate network under
The prediction result of request data.The calculating process is defined as follows:
ft=σ (Wf*[Ct-1,ht-1,Pt]+bf)
it=σ (Wi*[Ct-1,ht-1,Pt]+bi)
ot=σ (Wo*[Ct-1,ht-1,Pt]+bo)
Wherein htFor the output of LSTM network module in every step, ftFor the output for forgetting door, itFor the defeated of input gate
Out, otThe output of out gate is represented,It is the neoblast state also handled without input gate.
The prediction data information is perceived, to obtain the decoded prediction data information, by the decoding
The prediction data information afterwards is reduced to quality of service data and exports.Quality of service data after reduction can facilitate user to know
Not and use.
Preferably, the context variant information includes context and its corresponding quality of service data information.
Preferably, the context variant information includes response time information and network throughput information；Institute after reduction
Stating quality of service data includes response time information and network throughput information.
Preferably, the step will carry out dimension compression and carry out feature to mention by the context variant information of coding
It takes further include: multilayer fully-connected network composition after the context variant information of coding is compressed, will be input to
Feature extractor carries out feature extraction.
Preferably, the embedded request vector is sequentially input to LSTM network module, to export according to the timing
Deep neural network is stated to the prediction data information for inputting request contexts variable information next time further include: according to the timing
The embedded request vector sequentially input by calculate successively obtains corresponding calculated result, by the calculated result most
The latter calculated result believes the prediction data for inputting request contexts variable information next time as the deep neural network
Breath.
Preferably, described to be reduced to quality of service data and export also wrap by the decoded prediction data information
It includes: the prediction data information is inputted respectively in the recovery module that two fully-connected networks are constituted and is respectively calculated, with
Obtain include response time information and network throughput information quality of service data.
In a particular embodiment, by selecting absolute error loss function Loss1Loss letter as LSTM network module
Number, is trained parameter using Adam optimization algorithm.Wherein:
Wherein m is the number for needing the variable predicted, n is gross sample
This number, abs () expression take its absolute value,Indicate the value of i-th of the sample predicted for k-th of prediction task,
Indicate the true value that sample is corresponded in data set.λ is punishment term coefficient, L2(θ) is then regularization term.
As shown in Fig. 2, the embodiment of the present invention also provides a kind of service quality forecasting system based on deep neural network, institute
The system of stating includes coding module, compression module, LSTM network module, sensing module and recovery module；
The coding module exports embedded request matrix extremely for receiving request contexts variable information after being encoded
The compression module；
The compression module is used to be compressed and be carried out feature extraction by the context variant information of coding,
And embedded request vector is exported to the LSTM network module；
The LSTM network module calculates the embedded request vector, and exports the deep neural network pair
The prediction data information of request contexts variable information is inputted next time to the sensing module；
The sensing module perceives the prediction data information, and will obtain the decoded prediction data letter
Breath is sent to recovery module, and the decoded prediction data information is reduced to quality of service data by the recovery module
And it exports.
Preferably, the context variant information includes response time information and network throughput information；Institute after reduction
Stating quality of service data includes response time information and network throughput information.
Preferably, the recovery module includes the first recovery module and the second recovery module, when restoring response respectively
Between information and network throughput information.
It should be understood that the above is only a preferred embodiment of the present invention, the scope of the patents of the invention cannot be therefore limited,
It is all to utilize equivalent structure or equivalent flow shift made by description of the invention and accompanying drawing content, it is applied directly or indirectly in
Other related technical areas are included within the scope of the present invention.
Claims (9)
1. a kind of service quality prediction technique based on deep neural network, which comprises the following steps:
Request contexts variable information is inputted, and is encoded by entity expression matrix in coding module, it is embedded to obtain
Request matrix；
Feature extraction will be compressed and is carried out by the context variant information of coding, to obtain the request contexts
The embedded request vector according to timing of variable information；
The embedded request vector is sequentially input to LSTM network module, to export the depth nerve net according to the timing
Network is to the prediction data information for inputting request contexts variable information next time；
The prediction data information is perceived, it, will be described decoded to obtain the decoded prediction data information
The prediction data information is reduced to quality of service data and exports.
2. the service quality prediction technique according to claim 1 based on deep neural network, which is characterized in that on described
Hereafter variable information includes context and its corresponding quality of service data information.
3. the service quality prediction technique according to claim 2 based on deep neural network, which is characterized in that on described
Hereafter variable information includes response time information and network throughput information；The quality of service data after reduction includes response
Temporal information and network throughput information.
4. the service quality prediction technique according to claim 3 based on deep neural network, which is characterized in that the step
Suddenly dimension compression will be carried out by the context variant information of coding and carry out feature extraction further include:
The feature for after the context variant information of coding is compressed, being input to multilayer fully-connected network composition is mentioned
Device is taken to carry out feature extraction.
5. the service quality prediction technique according to claim 4 based on deep neural network, which is characterized in that according to institute
It states timing and sequentially inputs the embedded request vector to LSTM network module, to export the deep neural network to next time
Input the prediction data information of request contexts variable information further include:
The embedded request vector sequentially input according to the timing successively obtains corresponding calculated result by calculating, will
The last one calculated result is as the deep neural network to input request contexts variable next time in the calculated result
The prediction data information of information.
6. the service quality prediction technique according to claim 5 based on deep neural network, which is characterized in that described to incite somebody to action
The decoded prediction data information is reduced to quality of service data and exports further include:
The prediction data information is inputted respectively in the recovery module that two fully-connected networks are constituted and is respectively calculated, with
Obtain include response time information and network throughput information quality of service data.
7. a kind of service quality forecasting system based on deep neural network, which is characterized in that the system comprises coding module,
Compression module, LSTM network module, sensing module and recovery module；
The coding module exports embedded request matrix to described for receiving request contexts variable information after being encoded
Compression module；
The compression module is used to be compressed by the context variant information of coding and carries out feature extraction, and defeated
Embedded request vector is to the LSTM network module out；
The LSTM network module calculates the embedded request vector, and exports the deep neural network to next
The prediction data information of secondary input request contexts variable information is to the sensing module；
The sensing module perceives the prediction data information, and will obtain the decoded prediction data information hair
It send to recovery module, the decoded prediction data information is reduced to quality of service data and defeated by the recovery module
Out.
8. the service quality forecasting system according to claim 7 based on deep neural network, which is characterized in that on described
Hereafter variable information includes response time information and network throughput information；The quality of service data after reduction includes response
Temporal information and network throughput information.
9. the service quality forecasting system according to claim 8 based on deep neural network, which is characterized in that described to go back
Former module includes the first recovery module and the second recovery module, to restore response time information and network throughput letter respectively
Breath.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910671422.4A CN110460463B (en) | 2019-07-24 | 2019-07-24 | Service quality prediction method and system based on deep neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910671422.4A CN110460463B (en) | 2019-07-24 | 2019-07-24 | Service quality prediction method and system based on deep neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110460463A true CN110460463A (en) | 2019-11-15 |
| CN110460463B CN110460463B (en) | 2021-12-14 |
Family
ID=68483279
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910671422.4A Active CN110460463B (en) | 2019-07-24 | 2019-07-24 | Service quality prediction method and system based on deep neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110460463B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111541570A (en) * | 2020-04-22 | 2020-08-14 | 北京交通大学 | Cloud service QoS prediction method based on multi-source feature learning |
| CN114124554A (en) * | 2021-11-29 | 2022-03-01 | 燕山大学 | Virtual network service chain throughput prediction method |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108024158A (en) * | 2017-11-30 | 2018-05-11 | 天津大学 | There is supervision video abstraction extraction method using visual attention mechanism |
| CN109409499A (en) * | 2018-09-20 | 2019-03-01 | 北京航空航天大学 | One kind being based on deep learning and the modified track restoration methods of Kalman filtering |
| US20190102653A1 (en) * | 2017-09-29 | 2019-04-04 | Konica Minolta Laboratory U.S.A., Inc. | Local connectivity feature transform of binary images containing text characters for optical character/word recognition |
| CN109858972A (en) * | 2019-02-13 | 2019-06-07 | 重庆金窝窝网络科技有限公司 | The prediction technique and device of ad click rate |
-
2019
- 2019-07-24 CN CN201910671422.4A patent/CN110460463B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20190102653A1 (en) * | 2017-09-29 | 2019-04-04 | Konica Minolta Laboratory U.S.A., Inc. | Local connectivity feature transform of binary images containing text characters for optical character/word recognition |
| CN108024158A (en) * | 2017-11-30 | 2018-05-11 | 天津大学 | There is supervision video abstraction extraction method using visual attention mechanism |
| CN109409499A (en) * | 2018-09-20 | 2019-03-01 | 北京航空航天大学 | One kind being based on deep learning and the modified track restoration methods of Kalman filtering |
| CN109858972A (en) * | 2019-02-13 | 2019-06-07 | 重庆金窝窝网络科技有限公司 | The prediction technique and device of ad click rate |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111541570A (en) * | 2020-04-22 | 2020-08-14 | 北京交通大学 | Cloud service QoS prediction method based on multi-source feature learning |
| CN114124554A (en) * | 2021-11-29 | 2022-03-01 | 燕山大学 | Virtual network service chain throughput prediction method |
| CN114124554B (en) * | 2021-11-29 | 2022-08-30 | 燕山大学 | Virtual network service chain throughput prediction method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110460463B (en) | 2021-12-14 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108874914B (en) | Information recommendation method based on graph convolution and neural collaborative filtering | |
| CN110189192A (en) | A kind of generation method and device of information recommendation model | |
| CN113905391B (en) | Integrated learning network traffic prediction method, system, equipment, terminal and medium | |
| CN110933429B (en) | Video compression sensing and reconstruction method and device based on deep neural network | |
| CN110381524B (en) | Bi-LSTM-based large scene mobile flow online prediction method, system and storage medium | |
| CN108596902A (en) | The full reference image quality appraisement method of multitask based on gating convolutional neural networks | |
| CN111832570A (en) | Image semantic segmentation model training method and system | |
| CN109033107A (en) | Image search method and device, computer equipment and storage medium | |
| CN109635946A (en) | A kind of combined depth neural network and the clustering method constrained in pairs | |
| CN110460463A (en) | Service quality prediction technique and its system based on deep neural network | |
| CN116187555A (en) | Traffic flow prediction model construction method and prediction method based on self-adaptive dynamic diagram | |
| CN112995715A (en) | Video frame insertion processing method and device, electronic equipment and storage medium | |
| CN113255908A (en) | Method, neural network model and device for service prediction based on event sequence | |
| CN110059625A (en) | A kind of face training and recognition methods based on mixup | |
| CN113965659B (en) | HEVC (high efficiency video coding) video steganalysis training method and system based on network-to-network | |
| CN116306780B (en) | Dynamic graph link generation method | |
| CN111046213A (en) | Knowledge base construction method based on image recognition | |
| CN113345564A (en) | Early prediction method and device for patient hospitalization duration based on graph neural network | |
| CN115438753B (en) | Method for measuring security of federal learning protocol data based on generation | |
| Acosta et al. | Predicting city safety perception based on visual image content | |
| CN110020379B (en) | Link prediction method based on deep dynamic network embedded representation model | |
| CN114399901B (en) | Method and equipment for controlling traffic system | |
| CN116227562A (en) | Timing point process prediction method and system based on graph neural network and transducer | |
| CN114693352A (en) | Client access amount prediction method and device | |
| CN114547276A (en) | Three-channel diagram neural network-based session recommendation method |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |