CN107087161A - The Forecasting Methodology of user experience quality based on multilayer neural network in video traffic - Google Patents
The Forecasting Methodology of user experience quality based on multilayer neural network in video traffic Download PDFInfo
- Publication number
- CN107087161A CN107087161A CN201710291024.0A CN201710291024A CN107087161A CN 107087161 A CN107087161 A CN 107087161A CN 201710291024 A CN201710291024 A CN 201710291024A CN 107087161 A CN107087161 A CN 107087161A
- Authority
- CN
- China
- Prior art keywords
- mrow
- msubsup
- neural network
- layers
- layer
- 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
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N17/00—Diagnosis, testing or measuring for television systems or their details
- H04N17/004—Diagnosis, testing or measuring for television systems or their details for digital television systems
Abstract
The invention discloses the Forecasting Methodology of the user experience quality based on multilayer neural network in video traffic, including data prediction, that is, select to influence the characteristic parameter of Consumer's Experience in video traffic, the QoE of user is mapped as according to the report barrier of user/non-report barrier.The QoE forecast models of multilayer neural network are then set up, the neutral net includes five layers, is followed successively by from low to high：The hidden layer-second of input layer-first hidden layer-three hidden layers-output layer.Pretreated data are inputted, the optimum parameter value of model is obtained, the above-mentioned neural network model established is trained.Finally complete user experience quality QoE predictions.The present invention is pre-processed to data, choose important characteristic attribute, consider various parameters, the video quality of model prediction can be made real close to subjective experience of the user to video quality, help more preferably to predict user experience quality, be conducive to promptly and accurately feedback result, help service provider and Virtual network operator constantly improve video traffic and transmission to service.
Description
Technical field
The present invention relates to user experience quality prediction, the user's body based on multilayer neural network more particularly in video traffic
Test qualitative forecasting method.
Background technology
The fast development of Internet technology allows people to access various multimedia services, especially, and present IPTV is carried
For various services so that the life of people becomes rich and varied.But then, service provider and Virtual network operator be then more
It is concerned about the quality that Video service is provided, that is, how on earth is the IPTV Consumer's Experiences of watched video.This causes user's body to check the quality
The prediction and evaluation of amount turns into service provider and Virtual network operator focus of interest.The definition of user experience quality (QoE) is
" application or the overall acceptance of service that terminal user perceives ".QoE is not only influenceed by service itself, but also benefits from
The influence of environment residing for family.Because machine learning has the ability for intelligently solving big data information processing, it can be answered
Prediction for IPTV user experience qualities.However, traditional machine learning method, such as SVMs, decision tree it is pre-
Survey performance extremely limited, it is impossible to be accurately finished the task, thus need the new model of design and Forecasting Methodology, complete IPTV
The prediction and lifting of user experience quality.
The content of the invention
It is an object of the invention in view of the above-mentioned problems, proposing a kind of user experience quality based on multilayer neural network
Forecasting Methodology, accurately and efficiently to predict IPTV user experience qualities.
For this purpose, the technical solution adopted by the present invention is checked the quality for the user's body based on multilayer neural network in video traffic
The Forecasting Methodology of amount, specifically includes following steps：
Step 1：Data prediction：Select video traffic in influence Consumer's Experience characteristic parameter, according to the report of user barrier/
Non- report barrier, is mapped as the QoE of user, when QoE is 1, represents that user is satisfied with to business used, and when QoE is 0, uses
Family is unsatisfied with；
Step 2：Set up the QoE forecast models of multilayer neural network：The neutral net includes five layers, from low to high successively
For：The hidden layer-second of input layer-first hidden layer-three hidden layers-output layer, l layers of i-th of god of multilayer neural network
It is output as through member：
In above formula,The output of l layers of i-th of neuron of neutral net is represented,Represent the of neutral net
L-1 layers of output,The weight of l layers of i-th of neuron is represented,Represent the deviation of l i-th of neuron of layer, f
() represents activation primitive；
Step 3：Train QoE forecast models：Pretreated data are inputted, the optimum parameter value of model, training step is obtained
Rapid 2 neural network models established；
Step 4：User experience quality QoE predictions are completed, are specifically included：
(4-1) completes to pre-process first, the spy needed for the data of unknown subscriber's Quality of experience according to step 1
Levy data set；
This feature data set as input, is substituted into the multilayer neural network grader trained by (4-2), obtains prediction knot
Really.
Further, preferably, features described above parameter can include warning number of times, loss late, outlet download bandwidth, medium
Speed, delay, media loss rate, CPU usage, video transmission quality.
Preferably, above-mentioned each self-contained neuronal quantity of five layers of neutral net is respectively：Input layer has 10 nerves
Member, the first hidden layer has 30 neurons, and the second hidden layer has 100 neurons, and the 3rd hidden layer has 50 neurons, and output layer is 1
Individual neuron.
Above-mentioned activation primitive uses ReLU functions：
Wherein, z is input numerical value.
The detailed process of step 3 is as follows：
(3-1) determines that dropdout ratios are 0.2,
Wherein,It is to obey probability to be distributed for p Bernoulli Jacob, i.e. p dropout ratios herein, then, and l-1 layers
Output is changed intoSo l layers of i-th of neuron is output as
(3-2) determines that learning rate α is 0.1, and it is the step-length for controlling SGD algorithms；
(3-3) initializes the weight of each layer of input data,Illustrating l-1 layers has m
Output, initializes each layer of deviation,Wherein, n is the number of l layers of neuron, Obey just
State is distributed, in addition, making iterations m=1, sets total iterations as M；
(3-4) starts iteration, and each sample is chosen every time, is obtained using the Back Propagation Algorithm based on stochastic gradient descent method
Parameter in the model determined to step 2, training draws new model, and the model is output as 0 or 1；
(3-5) calculates loss function of the above-mentioned sample in "current" model：
Wherein,The output of l layers of i-th of neuron of neutral net is represented,Represent actual output valve, m tables
This total number of sample；
(3-6) updates weight w'j, its calculation formula is as follows：
Wherein, α represents learning rate, i.e., for controlling wjThe step-length of renewal, adjusts suitable learning rate, it is possible to achieve to mould
In the optimization of type, this model, learning rate is set to 0.2；
(3-7) judges whether to terminate iteration, if m<M, then jump to step (3-3), and iterations adds 1 (m=m+1),
Continue next iteration；Otherwise, iteration is terminated, weight w, deviation b is exported；Complete the training of multilayer neural network grader
Journey.
The invention has the advantages that：
1. the present invention is pre-processed to data, important characteristic attribute is chosen, various parameters is considered, mould can be made
The video quality of type prediction is real close to subjective experience of the user to video quality, helps preferably to predict that user's body is checked the quality
Amount, is conducive to promptly and accurately feedback result, so as to help service provider and Virtual network operator constantly improve video traffic and biography
Defeated service.
2. multilayer neural network model proposed by the present invention, efficient process lack of balance data set, effectively improves the standard of prediction
Exactness.
Brief description of the drawings
Fig. 1 is the Forecasting Methodology flow chart of the user experience quality based on multilayer neural network.
Fig. 2 is multilayer neural network structure chart.
Fig. 3 is the multilayer neural network structure chart after addition dropout.
Fig. 4 be yeast1 data sets under, MNN, SVM, the QoE prediction accuracy comparison diagrams of tri- kinds of algorithms of DT.
Fig. 5 be QoE predictive data sets under, MNN, SVM, the QoE predictablity rate comparison diagrams of tri- kinds of algorithms of DT.
Fig. 6 is the accuracy rate schematic diagram under different SGD learning rates.
Embodiment
With reference to Figure of description and instantiation, the present invention is described in further detail.
The Forecasting Methodology of user experience quality based on multilayer neural network in video traffic, this method, as shown in figure 1,
Comprise the following steps：
Step 1：Data prediction：Select to influence the characteristic parameter of Consumer's Experience in video traffic, it includes warning time
Number, loss late, outlet download bandwidth, media rate, delay, media loss rate, CPU usage, video transmission quality.In addition,
According to the report barrier of user in video traffic/non-report barrier, the QoE of user is mapped as, when QoE is 1, represents user to used
Business is satisfied with, and when QoE is 0, user is unsatisfied with；
Step 2：Set up QoE forecast models：Here multilayer neural network model is used.The neutral net includes five
Layer, as shown in Fig. 2 sorting from low to high：The hidden layer-second of input layer-first hidden layer-three hidden layers-output layer, wherein,
Input layer has 10 neurons, and the first hidden layer has 30 neurons, and the second hidden layer has 100 neurons, and the 3rd hidden layer has 50
Neuron, output layer is 1 neuron.L layers of i-th of neuron of multilayer neural network is output as：
In above formula,The output of l layers of i-th of neuron of neutral net is represented,Represent the of neutral net
L-1 layers of output,The weight of l layers of i-th of neuron is represented,Represent the deviation of l i-th of neuron of layer.f
() represents activation primitive, and ReLU functions are used herein：
Wherein, z is input numerical value, i.e.,
Step 3：Train QoE forecast models：Pretreated data are inputted, using based on stochastic gradient descent
Back-propagating (Back Propagation, the abbreviation BP) method of (Stochastic Gradient Descent, abbreviation SGD)
Obtain the optimum parameter value of model, the neural network model that training step 2 is established, the multilayer nerve net added after dropout
Network structure is as shown in figure 3, its detailed process is as follows：
(3-1) determines that dropdout ratios are 0.2,
Wherein,It is to obey probability to be distributed for p Bernoulli Jacob, i.e. p dropout ratios herein, then l-1 layers
Output is changed intoSo l layers of i-th of neuron is output as
(3-2) determines that learning rate α is 0.1, and it is the step-length for controlling SGD algorithms；
(3-3) initializes the weight of each layer of input data,Illustrating l-1 layers has m
Output.Each layer of deviation is initialized,Wherein, n is the number of l layers of neuron. Obey just
State is distributed.In addition, making iterations m=1, total iterations is set as M；
(3-4) starts iteration, each sample is chosen every time, using the back-propagating based on stochastic gradient descent method (SGD)
(BP) algorithm obtains the parameter in the model of step 2 determination, and training draws new model, it is necessary to which what is illustrated is the output of the model
For 0 or 1；
(3-5) calculates loss function of the above-mentioned sample in "current" model：
Wherein,The output of l layers of i-th of neuron of neutral net is represented,Represent actual output valve, m tables
This total number of sample.
(3-6) updates weight w'j, its calculation formula is as follows：
Wherein, α represents learning rate, i.e., for controlling wjThe step-length of renewal, adjusts suitable learning rate, it is possible to achieve to mould
The optimization of type.In this model, learning rate is set to 0.2；
(3-7) judges whether to terminate iteration.If m<M, then jump to step (3-3), and iterations adds 1 (m=m+1),
Continue next iteration；Otherwise, iteration is terminated, weight w, deviation b is exported；Complete the training of multilayer neural network grader
Journey；
Step 4：Complete user experience quality QoE predictions.
(4-1) completes to pre-process first, the spy needed for the data of unknown subscriber's Quality of experience according to step 1
Levy data set；
(4-2) as input, brings characteristic data set into train multilayer neural network grader, obtains prediction knot
Really.
Tested by above-mentioned flow, data prediction carried out first, determined characteristic attribute parameter, then set up model,
Training is completed using multilayer neural network model again and predicted.The data set used includes standard data set yeast1, and machine
The data record of the different user of top box collection.The training of model is completed by step 3, according to the BP algorithm based on SGD, is obtained
Optimal parameter needed for model --- weight w and deviation b, so as to train forecast model.Meanwhile, used in the training stage
Dropout method, random appropriate some neuron nodes of closing.It is predicted by step 4.Utilize ten folding cross validations
Method, does training sample using 9 groups of data therein in turn, and remaining one is test sample.Finally obtain the accuracy rate of model.For
Analysis result, contrasts the inventive method and two kinds of representative existing methods using accuracy rate --- SVMs
And decision tree (DT) estimated performance (SVM).
Fig. 4 compared for multilayer neural network (the inventive method), SVM, and DT algorithms under unbalanced dataset yeast1
Accuracy rate, yeast1 data sets include 1484 sample sets, each sample comprising 8 features, the unbalance factor used for
2.46.It is seen that the algorithm that this method is used is more preferable than the performance of other two kinds of algorithms, its accuracy rate is higher.This says
Bright the inventive method improves the prediction accuracy of unbalanced dataset to a certain extent.
Fig. 5 compared for multilayer neural network MNN (the inventive method), SVM, and the imbalance that DT algorithms are gathered in set top box
Accuracy rate under data set, in the algorithm, using five layer networks, (sorts from low to high：Hidden layer-the second of input layer-first
Hidden layer-three hidden layers-output layer), wherein, input layer has 10 neurons, and the first hidden layer has 30 neurons, the second hidden layer
There are 100 neurons, the 3rd hidden layer there are 50 neurons, and output layer is 1 neuron.Set in input layer and the first hidden layer
Dropout ratio is 0.2.With SVM, DT methods are compared, and the predictablity rate of this method is higher.
Fig. 6 shows different SGD learning rates, can cause the accuracy rate change of prediction, can be seen that in figure, work as learning rate
For 0.1 when, rate of accuracy reached, to highest, is 80.5%.
It should be noted that embodiment of above is merely illustrative of the technical solution of the present invention and unrestricted, although passing through
The present invention is described in detail for above preferred embodiment, it is to be understood by those skilled in the art that can be in shape
Various changes are made in formula and to it in details, without departing from limited range of the present invention.
Claims (5)
1. the Forecasting Methodology of the user experience quality based on multilayer neural network in video traffic, it is characterised in that methods described
Comprise the following steps：
Step 1：Data prediction：Select to influence the characteristic parameter of Consumer's Experience in video traffic, according to the report of user barrier/non-report
Barrier, is mapped as the QoE of user, when QoE is 1, represents that user is satisfied with to business used, and when QoE is 0, user is not
It is satisfied；
Step 2：Set up the QoE forecast models of multilayer neural network：The neutral net includes five layers, is followed successively by from low to high：It is defeated
Enter the hidden layer-second of layer-first hidden layer-three hidden layers-output layer, l layers of i-th of neuron of multilayer neural network
It is output as：
<mrow>
<msubsup>
<mi>y</mi>
<mi>i</mi>
<mi>l</mi>
</msubsup>
<mo>=</mo>
<mi>f</mi>
<mrow>
<mo>(</mo>
<msubsup>
<mi>w</mi>
<mi>i</mi>
<mi>l</mi>
</msubsup>
<msubsup>
<mi>y</mi>
<mrow>
<mi>i</mi>
<mo>-</mo>
<mn>1</mn>
</mrow>
<mrow>
<mi>l</mi>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msubsup>
<mo>+</mo>
<msubsup>
<mi>b</mi>
<mi>i</mi>
<mi>l</mi>
</msubsup>
<mo>)</mo>
</mrow>
</mrow>
In above formula,The output of l layers of i-th of neuron of neutral net is represented,Represent the l-1 of neutral net
The output of layer,The weight of l layers of i-th of neuron is represented,Represent the deviation of l i-th of neuron of layer, f ()
Represent activation primitive；
Step 3：Train QoE forecast models：Pretreated data are inputted, the optimum parameter value of model is obtained, training step 2 is built
The neural network model stood；
Step 4：User experience quality QoE predictions are completed, are specifically included：
(4-1) completes to pre-process first, the characteristic needed for the data of unknown subscriber's Quality of experience according to step 1
According to collection；
This feature data set as input, is substituted into the multilayer neural network grader trained, predicted the outcome by (4-2).
2. the Forecasting Methodology of the user experience quality based on multilayer neural network in video traffic according to claim 1,
It is characterized in that the characteristic parameter can include warning number of times, loss late, outlet download bandwidth, media rate, delay, media
Loss Rate, CPU usage, video transmission quality.
3. the Forecasting Methodology of the user experience quality based on multilayer neural network in video traffic according to claim 1,
It is characterized in that each self-contained neuronal quantity of five layers of neutral net is：Input layer has 10 neurons, and the first hidden layer has 30
Individual neuron, the second hidden layer has 100 neurons, and the 3rd hidden layer has 50 neurons, and output layer is 1 neuron.
4. the Forecasting Methodology of the user experience quality based on multilayer neural network in video traffic according to claim 1,
It is characterized in that the activation primitive uses ReLU functions：
<mrow>
<mi>f</mi>
<mrow>
<mo>(</mo>
<mi>z</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfenced open = "{" close = "">
<mtable>
<mtr>
<mtd>
<mi>z</mi>
<mo>,</mo>
<mi>z</mi>
<mo>&GreaterEqual;</mo>
<mn>0</mn>
</mtd>
</mtr>
<mtr>
<mtd>
<mn>0</mn>
<mo>,</mo>
<mi>z</mi>
<mo>&le;</mo>
<mn>0</mn>
</mtd>
</mtr>
</mtable>
</mfenced>
</mrow>
Wherein, z is input numerical value.
5. the Forecasting Methodology of the user experience quality based on multilayer neural network in video traffic according to claim 1,
It is characterized in that the detailed process of step 3 is as follows：
(3-1) determines that dropdout ratios are 0.2,
<mrow>
<msubsup>
<mi>r</mi>
<mi>j</mi>
<mrow>
<mi>l</mi>
<mo>-</mo>
<mn>1</mn>
</mrow>
</msubsup>
<mo>~</mo>
<mi>B</mi>
<mi>e</mi>
<mi>r</mi>
<mi>n</mi>
<mi>o</mi>
<mi>u</mi>
<mi>l</mi>
<mi>l</mi>
<mi>i</mi>
<mrow>
<mo>(</mo>
<mi>p</mi>
<mo>)</mo>
</mrow>
</mrow>
Wherein,It is to obey probability to be distributed for p Bernoulli Jacob, i.e. p dropout ratios herein, then, and l-1 layers of output
It is changed intoSo l layers of i-th of neuron is output as
(3-2) determines that learning rate α is 0.1, and it is the step-length for controlling SGD algorithms；
(3-3) initializes the weight of each layer of input data,Illustrating l-1 layers has m output,
Each layer of deviation is initialized,Wherein, n is the number of l layers of neuron,Obey normal state
Distribution, in addition, making iterations m=1, sets total iterations as M；
(3-4) starts iteration, and each sample is chosen every time, is walked using the Back Propagation Algorithm based on stochastic gradient descent method
Parameter in rapid 2 models determined, training draws new model, and the model is output as 0 or 1；
(3-5) calculates loss function of the above-mentioned sample in "current" model：
<mrow>
<mi>J</mi>
<mrow>
<mo>(</mo>
<mi>&theta;</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<mfrac>
<mn>1</mn>
<mrow>
<mn>2</mn>
<mi>m</mi>
</mrow>
</mfrac>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>m</mi>
</munderover>
<msup>
<mrow>
<mo>(</mo>
<msubsup>
<mi>y</mi>
<mo>*</mo>
<mi>i</mi>
</msubsup>
<mo>-</mo>
<msubsup>
<mi>y</mi>
<mi>i</mi>
<mi>l</mi>
</msubsup>
<mo>)</mo>
</mrow>
<mn>2</mn>
</msup>
</mrow>
Wherein,The output of l layers of i-th of neuron of neutral net is represented,Actual output valve is represented, m represents sample
This total number；
(3-6) updates weight w'j, its calculation formula is as follows：
<mfenced open = "" close = "">
<mtable>
<mtr>
<mtd>
<mrow>
<msubsup>
<mi>w</mi>
<mi>j</mi>
<mo>&prime;</mo>
</msubsup>
<mo>=</mo>
<msub>
<mi>w</mi>
<mi>j</mi>
</msub>
<mo>-</mo>
<mi>&alpha;</mi>
<mfrac>
<mrow>
<mo>&part;</mo>
<mi>J</mi>
<mrow>
<mo>(</mo>
<mi>w</mi>
<mo>)</mo>
</mrow>
</mrow>
<mrow>
<mo>&part;</mo>
<msub>
<mi>w</mi>
<mi>j</mi>
</msub>
</mrow>
</mfrac>
</mrow>
</mtd>
</mtr>
<mtr>
<mtd>
<mrow>
<mo>=</mo>
<msub>
<mi>w</mi>
<mi>j</mi>
</msub>
<mo>+</mo>
<mi>&alpha;</mi>
<mrow>
<mo>(</mo>
<msubsup>
<mi>y</mi>
<mo>*</mo>
<mi>i</mi>
</msubsup>
<mo>-</mo>
<msubsup>
<mi>y</mi>
<mi>i</mi>
<mi>l</mi>
</msubsup>
<mo>)</mo>
</mrow>
<msubsup>
<mi>x</mi>
<mi>i</mi>
<mi>l</mi>
</msubsup>
</mrow>
</mtd>
</mtr>
</mtable>
</mfenced>
Wherein, α represents learning rate, i.e., for controlling wjThe step-length of renewal, adjusts suitable learning rate, it is possible to achieve to model
In optimization, this model, learning rate is set to 0.2；
(3-7) judges whether to terminate iteration, if m<M, then jump to step (3-3), and iterations adds 1 (m=m+1), continues
Next iteration；Otherwise, iteration is terminated, weight w, deviation b is exported；Complete the training process of multilayer neural network grader.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710291024.0A CN107087161B (en) | 2017-04-28 | 2017-04-28 | The prediction technique of user experience quality in video traffic based on multilayer neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710291024.0A CN107087161B (en) | 2017-04-28 | 2017-04-28 | The prediction technique of user experience quality in video traffic based on multilayer neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107087161A true CN107087161A (en) | 2017-08-22 |
| CN107087161B CN107087161B (en) | 2019-05-03 |
Family
ID=59612313
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710291024.0A Active CN107087161B (en) | 2017-04-28 | 2017-04-28 | The prediction technique of user experience quality in video traffic based on multilayer neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107087161B (en) |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108768750A (en) * | 2018-06-22 | 2018-11-06 | 广东电网有限责任公司 | Communication network failure localization method and device |
| CN109040744A (en) * | 2018-07-27 | 2018-12-18 | 华为技术有限公司 | Predict the method, apparatus and storage medium of the Key Quality Indicator of video traffic |
| CN109189915A (en) * | 2018-09-17 | 2019-01-11 | 重庆理工大学 | A kind of information retrieval method based on depth relevant matches model |
| CN109934627A (en) * | 2019-03-05 | 2019-06-25 | 中国联合网络通信集团有限公司 | Establish the method and device of satisfaction degree estimation model |
| CN110446112A (en) * | 2019-07-01 | 2019-11-12 | 南京邮电大学 | IPTV user experience prediction technique based on two-way LSTM-Attention |
| CN110531163A (en) * | 2019-04-18 | 2019-12-03 | 中国人民解放军国防科技大学 | Bus capacitance state monitoring method for suspension chopper of maglev train |
| CN111031116A (en) * | 2019-12-02 | 2020-04-17 | 嘉兴学院 | Cloud service synthesis method, cloud server and cloud service synthesis system |
| CN111401637A (en) * | 2020-03-16 | 2020-07-10 | 湖南大学 | User experience quality prediction method fusing user behavior and expression data |
| CN112766482A (en) * | 2020-12-21 | 2021-05-07 | 北京航空航天大学 | Input layer structure and BP neural network |
| CN112884020A (en) * | 2021-01-29 | 2021-06-01 | 北京联合大学 | Service quality prediction method based on multi-scale circular convolution neural network |
| WO2023051318A1 (en) * | 2021-09-28 | 2023-04-06 | 中兴通讯股份有限公司 | Model training method, wireless resource scheduling method and apparatus therefor, and electronic device |
| CN117649153A (en) * | 2024-01-29 | 2024-03-05 | 南京典格通信科技有限公司 | Mobile communication network user experience quality prediction method based on information integration |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101188847A (en) * | 2007-11-28 | 2008-05-28 | 中讯邮电咨询设计院 | Experience evaluation method for mobile communication service user based on artificial neural network |
| US20100121798A1 (en) * | 2004-06-11 | 2010-05-13 | Canon Kabushiki Kaisha | Information processing apparatus, information processing method, pattern recognition apparatus, and pattern recognition method |
| CN102802089A (en) * | 2012-09-13 | 2012-11-28 | 浙江大学 | Shifting video code rate regulation method based on experience qualitative forecast |
| CN102944413A (en) * | 2012-11-29 | 2013-02-27 | 天津银河阀门有限公司 | Butterfly valve service life tester |
| CN104702666A (en) * | 2015-01-30 | 2015-06-10 | 北京邮电大学 | User experience quality confirmation method and system |
-
2017
- 2017-04-28 CN CN201710291024.0A patent/CN107087161B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20100121798A1 (en) * | 2004-06-11 | 2010-05-13 | Canon Kabushiki Kaisha | Information processing apparatus, information processing method, pattern recognition apparatus, and pattern recognition method |
| CN101188847A (en) * | 2007-11-28 | 2008-05-28 | 中讯邮电咨询设计院 | Experience evaluation method for mobile communication service user based on artificial neural network |
| CN102802089A (en) * | 2012-09-13 | 2012-11-28 | 浙江大学 | Shifting video code rate regulation method based on experience qualitative forecast |
| CN102944413A (en) * | 2012-11-29 | 2013-02-27 | 天津银河阀门有限公司 | Butterfly valve service life tester |
| CN104702666A (en) * | 2015-01-30 | 2015-06-10 | 北京邮电大学 | User experience quality confirmation method and system |
Cited By (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108768750A (en) * | 2018-06-22 | 2018-11-06 | 广东电网有限责任公司 | Communication network failure localization method and device |
| CN109040744A (en) * | 2018-07-27 | 2018-12-18 | 华为技术有限公司 | Predict the method, apparatus and storage medium of the Key Quality Indicator of video traffic |
| CN109040744B (en) * | 2018-07-27 | 2020-03-10 | 华为技术有限公司 | Method, device and storage medium for predicting key quality index of video service |
| CN109189915B (en) * | 2018-09-17 | 2021-10-15 | 重庆理工大学 | Information retrieval method based on depth correlation matching model |
| CN109189915A (en) * | 2018-09-17 | 2019-01-11 | 重庆理工大学 | A kind of information retrieval method based on depth relevant matches model |
| CN109934627A (en) * | 2019-03-05 | 2019-06-25 | 中国联合网络通信集团有限公司 | Establish the method and device of satisfaction degree estimation model |
| CN110531163A (en) * | 2019-04-18 | 2019-12-03 | 中国人民解放军国防科技大学 | Bus capacitance state monitoring method for suspension chopper of maglev train |
| CN110446112A (en) * | 2019-07-01 | 2019-11-12 | 南京邮电大学 | IPTV user experience prediction technique based on two-way LSTM-Attention |
| CN111031116A (en) * | 2019-12-02 | 2020-04-17 | 嘉兴学院 | Cloud service synthesis method, cloud server and cloud service synthesis system |
| CN111401637A (en) * | 2020-03-16 | 2020-07-10 | 湖南大学 | User experience quality prediction method fusing user behavior and expression data |
| CN111401637B (en) * | 2020-03-16 | 2023-06-16 | 湖南大学 | User experience quality prediction method integrating user behavior and expression data |
| CN112766482A (en) * | 2020-12-21 | 2021-05-07 | 北京航空航天大学 | Input layer structure and BP neural network |
| CN112884020A (en) * | 2021-01-29 | 2021-06-01 | 北京联合大学 | Service quality prediction method based on multi-scale circular convolution neural network |
| WO2023051318A1 (en) * | 2021-09-28 | 2023-04-06 | 中兴通讯股份有限公司 | Model training method, wireless resource scheduling method and apparatus therefor, and electronic device |
| CN117649153A (en) * | 2024-01-29 | 2024-03-05 | 南京典格通信科技有限公司 | Mobile communication network user experience quality prediction method based on information integration |
| CN117649153B (en) * | 2024-01-29 | 2024-04-16 | 南京典格通信科技有限公司 | Mobile communication network user experience quality prediction method based on information integration |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107087161B (en) | 2019-05-03 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107087161A (en) | The Forecasting Methodology of user experience quality based on multilayer neural network in video traffic | |
| Baños et al. | Resilience indexes for water distribution network design: a performance analysis under demand uncertainty | |
| CN102567391B (en) | Method and device for building classification forecasting mixed model | |
| CN104270281B (en) | A kind of Web Service QoS combination forecasting methods | |
| CN106570597A (en) | Content popularity prediction method based on depth learning under SDN architecture | |
| CN107480341A (en) | A kind of dam safety comprehensive method based on deep learning | |
| CN106357456A (en) | Prediction method of network traffic and device thereof | |
| CN109242149A (en) | A kind of student performance early warning method and system excavated based on educational data | |
| CN104900063B (en) | A kind of short distance running time Forecasting Methodology | |
| CN107464419A (en) | A kind of Short-time Traffic Flow Forecasting Methods for considering space-time characterisation | |
| CN104702666A (en) | User experience quality confirmation method and system | |
| CN107734512A (en) | A kind of network selecting method based on the analysis of gray scale relevance presenting levelses | |
| CN109460613A (en) | Model method of cutting out and device | |
| CN107346464A (en) | Operational indicator Forecasting Methodology and device | |
| CN106469340A (en) | A kind of compound Project Risk Assessment model based on neutral net | |
| CA2436352A1 (en) | Process and system for developing a predictive model | |
| CN107705556A (en) | A kind of traffic flow forecasting method combined based on SVMs and BP neural network | |
| CN111461500B (en) | Shared bicycle system tide phenomenon control method based on dynamic electronic fence and reinforcement learning | |
| CN111429737B (en) | Agent-based urban area boundary control method and system | |
| CN107832421A (en) | A kind of adaptive data splitting mining model system and method | |
| CN110213784B (en) | Flow prediction method and device | |
| CN106650022A (en) | Method for predicting fault of complex electronic device | |
| CN109508498A (en) | Rubber shock absorber formula designing system and method based on BP artificial neural network | |
| CN108898249A (en) | A kind of electric network fault prediction technique | |
| CN108510126A (en) | A kind of Predictive Methods of Road Accidents based on PCA and BP neural network |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| EE01 | Entry into force of recordation of patent licensing contract | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20170822 Assignee: Nanjing jinxuetang Information Technology Co.,Ltd. Assignor: NANJING University OF POSTS AND TELECOMMUNICATIONS Contract record no.: X2020980006913 Denomination of invention: Prediction method of user experience quality based on multilayer neural network in video service Granted publication date: 20190503 License type: Common License Record date: 20201020 |
|
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20201202 Address after: Room 214, building D5, No. 9, Kechuang Avenue, Zhongshan Science and Technology Park, Jiangbei new district, Nanjing, Jiangsu Province Patentee after: Nanjing Tian Gu Information Technology Co.,Ltd. Address before: 210003 Gulou District, Jiangsu, Nanjing new model road, No. 66 Patentee before: NANJING University OF POSTS AND TELECOMMUNICATIONS Effective date of registration: 20201202 Address after: Gulou District of Nanjing City, Jiangsu Province, Beijing Road No. 20 210024 Patentee after: STATE GRID JIANGSU ELECTRIC POWER Co.,Ltd. INFORMATION & TELECOMMUNICATION BRANCH Address before: Room 214, building D5, No. 9, Kechuang Avenue, Zhongshan Science and Technology Park, Jiangbei new district, Nanjing, Jiangsu Province Patentee before: Nanjing Tian Gu Information Technology Co.,Ltd. |
|
| EC01 | Cancellation of recordation of patent licensing contract | ||
| EC01 | Cancellation of recordation of patent licensing contract |
Assignee: Nanjing jinxuetang Information Technology Co.,Ltd. Assignor: NANJING University OF POSTS AND TELECOMMUNICATIONS Contract record no.: X2020980006913 Date of cancellation: 20220304 |