CN108960303B - Unmanned aerial vehicle flight data anomaly detection method based on LSTM - Google Patents
Unmanned aerial vehicle flight data anomaly detection method based on LSTM Download PDFInfo
- Publication number
- CN108960303B CN108960303B CN201810639367.6A CN201810639367A CN108960303B CN 108960303 B CN108960303 B CN 108960303B CN 201810639367 A CN201810639367 A CN 201810639367A CN 108960303 B CN108960303 B CN 108960303B
- Authority
- CN
- China
- Prior art keywords
- lstm
- aerial vehicle
- unmanned aerial
- flight data
- sample set
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/20—Administration of product repair or maintenance
Abstract
An unmanned aerial vehicle flight data anomaly detection method based on LSTM relates to the field of unmanned aerial vehicle anomaly detection and system health management. The method and the device aim to solve the problems that in the detection of the flight data abnormity of the unmanned aerial vehicle, functions generated in the working process of an unmanned aerial vehicle system are complex, and the approximation capability of the functions cannot meet the requirement of high fitting precision. Reconstructing a telemetering data phase space of an unmanned aerial vehicle, obtaining an input vector and an output vector, obtaining a training sample set and a testing sample set, building an LSTM basic prediction model by adopting a TensorFlow deep learning open source framework, optimizing parameters, obtaining an optimal LSTM model and further calculating an LSTM prediction result; and then abnormal point detection and abnormal sequence detection are respectively carried out, and finally, the abnormal detection of the flight data of the unmanned aerial vehicle is completed.
Description
Technical Field
The invention belongs to the field of unmanned aerial vehicle anomaly detection and system health management.
Background
The unmanned aerial vehicle anomaly detection method can be mainly divided into three categories, namely an anomaly detection method based on knowledge, an anomaly detection method based on a physical model and an anomaly detection method based on data driving. The first two methods need to fully understand domain knowledge or system structure to achieve better detection effect, have poor universality and poor anti-interference capability, and generally can only detect known types of abnormal detection. The unmanned aerial vehicle telemetering data has few abnormal labels and complicated abnormal types, and a uniform physical model is difficult to establish for each subsystem of the unmanned aerial vehicle telemetering data.
The anomaly detection method based on data driving belongs to the category of multivariate statistical analysis, and can be subdivided into a similarity-based method, a statistical-based method, a classification-based method, a prediction-based method and a probability-based method. The main form of the telemetering flight data of the unmanned aerial vehicle is a time sequence, the method based on prediction is widely applied to anomaly detection of the time sequence, has self-adaptability, and can carry out anomaly detection of off-line data or on-line data. Such methods are based on the prediction results of the data, so that the excellent detection effect depends on accurate and effective prediction models, such as ARMA, SVM, neural networks, and the like. However, the methods all belong to shallow learning algorithms, and for complex functions generated in the working process of the unmanned aerial vehicle system, the approximation capability of the functions cannot meet the requirement of high fitting precision. In recent years, deep learning is rapidly developed, and the problem is that an LSTM method in the deep learning field is adopted to complete a prediction model and complete unmanned aerial vehicle abnormity detection based on a prediction result.
Disclosure of Invention
The invention provides an LSTM-based unmanned aerial vehicle flight data anomaly detection method, aiming at solving the problems that in unmanned aerial vehicle flight data anomaly detection, functions generated in the working process of an unmanned aerial vehicle system are complex, and the approximation capability of the functions cannot meet the requirement of high fitting accuracy.
An unmanned aerial vehicle flight data anomaly detection method based on LSTM comprises an anomaly point detection method and an anomaly sequence detection method:
the abnormal point detection method specifically comprises the following steps:
the method comprises the following steps: reconstructing a telemetering data phase space of the unmanned aerial vehicle to obtain an input vector and an output vector, constructing a training sample set and a test sample set of an LSTM basic prediction model, and constructing the LSTM basic prediction model based on the training and test sample set;
step two: optimizing the parameters of the LSTM basic prediction model by using a grid search method, and substituting the optimized parameters into the LSTM basic prediction model to obtain an optimal LSTM model;
step three: inputting samples in the test sample set into the optimal LSTM model to obtain an LSTM prediction result;
step four: calculating residual errors between actual values of samples in the training sample set and LSTM prediction results, and taking the average value of the residual errors of the samples in the training sample set as a normal distribution center mu-mean (e)training) Taking the standard deviation of the sample residual in the training sample set as the normal distribution variance σ ═ std (e)training)，etrainingRepresenting the residual error of the samples in the training sample set, and obtaining a confidence interval when the confidence probability P is 99 percent:
[μ-2.6·σ,μ+2.6·σ]；
step five: judging whether the flight data of the unmanned aerial vehicle to be detected belong to a confidence interval, if so, determining that the flight data of the unmanned aerial vehicle to be detected is a normal point, otherwise, determining that the flight data of the unmanned aerial vehicle to be detected is an abnormal point;
the abnormal sequence detection method specifically comprises the following steps:
step six: dividing all unmanned aerial vehicle flight data to be detected into a plurality of time sequences according to a time sequence, wherein each time sequence contains n data, and n is more than or equal to 6 and less than or equal to 15;
step seven: and when the confidence probability is more than 99 percent, judging whether the number Q of the abnormal points in each time sequence belongs to [ n-4, n ], if so, determining that the time sequence is an abnormal time sequence, otherwise, determining that the time sequence is a normal time sequence.
Has the advantages that: according to the method for detecting the flight data abnormity of the unmanned aerial vehicle based on the LSTM, when the flight data abnormity of the unmanned aerial vehicle is detected, the detection effect on the segment abnormity is more accurate. For the unmanned aerial vehicle data with the abnormal segment [303,343] in fig. 3, the abnormal segment finally detected by the method is [303,346], so that the method has a good detection effect, and the approximation capability can meet the requirement of high fitting precision.
Drawings
FIG. 1 is a flow chart of a method for detecting anomalies in flight data of an unmanned aerial vehicle based on LSTM;
FIG. 2 is a schematic diagram of an LSTM network structure;
FIG. 3 is a graph of data for an unmanned aerial vehicle with anomalies;
FIG. 4 is a graph of a test set residual curve versus a decision threshold;
FIG. 5 is a graph of a test set anomaly detection result signature;
1: an LSTM node.
Detailed Description
Referring to fig. 1, the present embodiment, a method for detecting an abnormal flight data of an LSTM-based unmanned aerial vehicle according to the present embodiment,
the method comprises the following steps: and (5) carrying out unmanned aerial vehicle telemetry data phase space reconstruction.
In order to adapt to the characteristics of the LSTM network structure, for the telemetering data of the single-dimensional unmanned aerial vehicle, firstly, phase space reconstruction is carried out, and an input vector and an output vector are constructed to be used as a training sample and a test sample of the LSTM, as shown in a formula 1.
Wherein, x (t) is an input vector constructed from a single-dimensional time sequence, x (t) is a value of the single-dimensional unmanned aerial vehicle telemetering data existing in the form of the time sequence at the time t, the phase space reconstruction window length is D, and Y (t) is a real value corresponding to the output vector X (t). What the LSTM model needs to do is the learning of the x (t) → y (t) mapping.
Will be ahead of
Using the input vector and the output vector as a training sample set of the LSTM basic prediction model, and using the residual vector
The input vectors and the output vectors are used as a test sample set of the LSTM basic prediction model, and k represents the total number of the input vectors and the output vectors.
Step two: and (4) building an LSTM basic prediction model by using the training sample set and the test sample set obtained in the step one and adopting a TensorFlow deep learning open source framework. The LSTM network structure is shown in fig. 2.
The method specifically comprises four main points:
A. the input dimension is confirmed. And determining input dimension according to the number of input data points of the LSTM network at each moment, wherein the product of the time expansion times of the network and the input dimension is equal to the length of the phase space reconstruction window.
B. And determining the number of LSTM structural network layers according to the complexity and input dimension of the telemetering data of the unmanned aerial vehicle. The implementation mode adopts a three-layer LSTM structure, namely, only one layer of hidden nodes is adopted, so that overfitting caused by too deep longitudinal layers is avoided.
C. An activation function is selected. A linear function is chosen between the hidden layer and the output layer as the activation function.
D. And (5) outputting the layer design. LSTM is fitted with a mapping relation of [ x (t), x (t-1), …, x (t- (D-1)) ] → x (t +1), each round of input data only corresponds to one-dimensional output, so that results are not output at all moments before the last input data comes, namely, a network output layer is eliminated, and the output layer normally outputs a prediction result when the network is unfolded along time for the last time. Training and testing the output vectors in the sample.
Step three: and optimizing the parameters of the LSTM basic prediction model by using a grid search method, and substituting the optimized parameters into the LSTM basic prediction model to obtain the optimal LSTM model.
In the LSTM model, a number of preset parameters may have a significant influence on the final prediction effect. The three more critical parameters are the phase space reconstruction window length L, the number N of hidden layer nodes, and the learning rate η. And optimizing the three parameters by adopting a grid searching method. The grid search means that the three parameters respectively take different values and are mutually crossed to form a three-dimensional search space of the parameters. Traversing the set value ranges of the three parameters, finally giving the prediction precision of the test set of all parameter combinations, and arranging the test set according to the sequence from high to low, wherein the parameter combination arranged at the head is the optimal parameter combination.
Step four: inputting the samples in the test sample set into the optimal LSTM model to obtain LSTM prediction results
Step five: an outlier is detected.
The predicted value and the true value have residual errors, and the random noise is regarded as random noise which cannot be further refined. Random noise is determined by many factors, and generally satisfies a normal distribution with a mean value of 0. Calculating residual errors between actual values of samples in the training sample set and LSTM prediction results, and taking the average value of the residual errors of the samples in the training sample set as a normal distribution center mu-mean (e)training) Taking the standard deviation of the sample residual in the training sample set as the normal distribution variance σ ═ std (e)training)，etrainingRepresenting the residual error of the samples in the training sample set, and obtaining the residual error when the confidence probability P is 99 percentConfidence interval:
[μ-2.6·σ,μ+2.6·σ] (2)
and judging whether the flight data of the unmanned aerial vehicle to be detected belong to a confidence interval, if so, determining that the flight data of the unmanned aerial vehicle to be detected is a normal point, and otherwise, determining that the flight data of the unmanned aerial vehicle to be detected is an abnormal point.
Step six: and detecting an abnormal sequence.
Dividing all unmanned aerial vehicle flight data to be detected into a plurality of time sequences according to a time sequence, wherein each time sequence contains n data, and n is more than or equal to 6 and less than or equal to 15; and when the confidence probability is more than 99 percent, judging whether the number Q of the abnormal points in each time sequence belongs to [ n-4, n ], if so, determining that the time sequence is an abnormal time sequence, otherwise, determining that the time sequence is a normal time sequence.
Step seven: and taking a union set of all abnormal time sequences to obtain an abnormal point set of all the flight data of the unmanned aerial vehicle to be detected.
The unmanned aerial vehicle data with abnormal fragments are selected for effect evaluation of the method. Unmanned aerial vehicle data as shown in fig. 3, there is a segment anomaly within the interval [303,343 ]. With the proposed method, a confidence level of 99% is set, i.e. a confidence interval of [ mu-2.6 · σ, mu +2.6 · σ ], resulting in a confidence interval as shown in fig. 4, where the part of the residual beyond the confidence interval is an abnormal value. And finally, marking all abnormal points. According to the proposed method, when at least 6 abnormal values appear in the time series with the degree of 10, the segment is an abnormal segment, and the finally obtained abnormal segment is [303,346], so that the detection effect is better, as shown in fig. 5.
Claims (5)
1. An unmanned aerial vehicle flight data anomaly detection method based on LSTM is characterized by comprising an anomaly point detection method and an anomaly sequence detection method:
the abnormal point detection method specifically comprises the following steps:
the method comprises the following steps: reconstructing a telemetering data phase space of the unmanned aerial vehicle to obtain an input vector and an output vector, constructing a training sample set and a test sample set of an LSTM basic prediction model, and constructing the LSTM basic prediction model based on the training sample set and the test sample set;
step two: optimizing the parameters of the LSTM basic prediction model by using a grid search method, and substituting the optimized parameters into the LSTM basic prediction model to obtain an optimal LSTM model;
step three: inputting samples in the test sample set into the optimal LSTM model to obtain an LSTM prediction result;
step four: calculating residual errors between actual values of samples in the training sample set and LSTM prediction results, and taking the average value of the residual errors of the samples in the training sample set as a normal distribution center mu-mean (e)training) Taking the standard deviation of the sample residual in the training sample set as the normal distribution variance σ ═ std (e)training)，etrainingRepresenting the residual error of the samples in the training sample set, and obtaining a confidence interval when the confidence probability P is 99 percent:
[μ-2.6·σ,μ+2.6·σ]；
step five: judging whether the flight data of the unmanned aerial vehicle to be detected belong to a confidence interval, if so, determining that the flight data of the unmanned aerial vehicle to be detected is a normal point, otherwise, determining that the flight data of the unmanned aerial vehicle to be detected is an abnormal point;
the abnormal sequence detection method specifically comprises the following steps:
step six: dividing all unmanned aerial vehicle flight data to be detected into a plurality of time sequences according to a time sequence, wherein each time sequence contains n data, and n is more than or equal to 6 and less than or equal to 15;
step seven: and when the confidence probability is more than 99 percent, judging whether the number Q of the abnormal points in each time sequence belongs to [ n-4, n ], if so, determining that the time sequence is an abnormal time sequence, otherwise, determining that the time sequence is a normal time sequence.
2. The LSTM-based unmanned aerial vehicle flight data anomaly detection method according to claim 1, wherein after step seven, all anomaly time series are merged to obtain all anomaly point sets of unmanned aerial vehicle flight data to be detected.
3. The method for detecting the abnormal flight data of the unmanned aerial vehicle based on the LSTM according to claim 1 or 2, wherein in the first step, the input vector X (t) and the output vector Y (t) of the unmanned aerial vehicle telemetry data phase space have the following expressions:
wherein, x (t) is the value of the single-dimensional unmanned aerial vehicle telemetering data existing in a time sequence form at the time t, and D is the phase space reconstruction window length.
4. The unmanned aerial vehicle flight data anomaly detection method based on the LSTM according to claim 1 or 2, wherein the method for building the LSTM basic prediction model in the first step is as follows:
determining input dimensionality according to the number of input data points of the LSTM network at each moment, wherein the product of the time expansion times of the network and the input dimensionality is equal to the length of a phase space reconstruction window;
determining the number of LSTM structural network layers according to the complexity and input dimension of the telemetering data of the unmanned aerial vehicle;
selecting a linear function between the hidden layer and the output layer as the activation function;
and enabling the output layer to output the prediction result when the LSTM network is expanded along the time for the last time.
5. The method of claim 1 or 2, wherein the parameters of the LSTM basic prediction model in step two comprise: the method comprises the steps of phase space reconstruction window length, hidden layer node number and learning rate.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810639367.6A CN108960303B (en) | 2018-06-20 | 2018-06-20 | Unmanned aerial vehicle flight data anomaly detection method based on LSTM |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810639367.6A CN108960303B (en) | 2018-06-20 | 2018-06-20 | Unmanned aerial vehicle flight data anomaly detection method based on LSTM |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108960303A CN108960303A (en) | 2018-12-07 |
| CN108960303B true CN108960303B (en) | 2021-05-07 |
Family
ID=64490435
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810639367.6A Active CN108960303B (en) | 2018-06-20 | 2018-06-20 | Unmanned aerial vehicle flight data anomaly detection method based on LSTM |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108960303B (en) |
Families Citing this family (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109711440B (en) * | 2018-12-13 | 2022-02-08 | 新奥数能科技有限公司 | Data anomaly detection method and device |
| CN109934337B (en) * | 2019-03-14 | 2020-12-25 | 哈尔滨工业大学 | Spacecraft telemetry data anomaly detection method based on integrated LSTM |
| CN110458195B (en) * | 2019-07-10 | 2022-08-19 | 河海大学 | Abnormal data discrimination method based on multi-criterion fusion |
| CN112306722A (en) * | 2019-07-26 | 2021-02-02 | 北京京东乾石科技有限公司 | Method, device, equipment and computer readable medium for identifying fault |
| CN110705658B (en) * | 2019-08-15 | 2022-03-15 | 哈尔滨工业大学 | Multivariable regression-based unmanned aerial vehicle flight parameter data fault detection and recovery method |
| CN110569925B (en) * | 2019-09-18 | 2023-05-26 | 南京领智数据科技有限公司 | LSTM-based time sequence abnormality detection method applied to power equipment operation detection |
| CN111241744B (en) * | 2020-01-07 | 2022-08-09 | 浙江大学 | Low-pressure casting machine time sequence data abnormity detection method based on bidirectional LSTM |
| CN111260029A (en) * | 2020-01-13 | 2020-06-09 | 北京工业大学 | Credibility analysis method for air quality data |
| CN111562996B (en) * | 2020-04-11 | 2021-11-23 | 北京交通大学 | Method and system for detecting time sequence abnormality of key performance index data |
| CN111786738B (en) * | 2020-07-01 | 2022-06-03 | 中国人民解放军陆军工程大学 | Anti-interference learning network structure based on long-term and short-term memory and learning method |
| CN112085866B (en) * | 2020-08-14 | 2023-04-07 | 陕西千山航空电子有限责任公司 | Airplane abnormal state identification method based on flight parameter data |
| CN113218537B (en) * | 2021-05-25 | 2024-04-05 | 中国南方电网有限责任公司超高压输电公司广州局 | Training method, training device, training equipment and training storage medium for temperature anomaly detection model |
| CN114037012B (en) * | 2021-11-09 | 2023-04-07 | 四川大学 | Flight data anomaly detection method based on correlation analysis and deep learning |
| CN116088349B (en) * | 2023-03-07 | 2023-06-27 | 东方空间技术(山东)有限公司 | LSTM-based rocket semi-physical simulation data anomaly evaluation method and decision device |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106055885A (en) * | 2016-05-26 | 2016-10-26 | 哈尔滨工业大学 | Anomaly detection method of flight data of unmanned aerial vehicle based on over-sampling projection approximation basis pursuit |
| EP3261024A2 (en) * | 2016-06-23 | 2017-12-27 | Siemens Healthcare GmbH | Method and system for vascular disease detection using recurrent neural networks |
| CN107748899A (en) * | 2017-11-03 | 2018-03-02 | 哈尔滨工业大学深圳研究生院 | A kind of target classification of the two dimensional image based on LSTM sentences knowledge method |
| CN107833208A (en) * | 2017-10-27 | 2018-03-23 | 哈尔滨工业大学 | A kind of hyperspectral abnormity detection method based on changeable weight depth own coding |
| CN108053653A (en) * | 2018-01-11 | 2018-05-18 | 广东蔚海数问大数据科技有限公司 | Vehicle behavior Forecasting Methodology and device based on LSTM |
| CN108063698A (en) * | 2017-12-15 | 2018-05-22 | 东软集团股份有限公司 | Unit exception detection method and device, program product and storage medium |
-
2018
- 2018-06-20 CN CN201810639367.6A patent/CN108960303B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106055885A (en) * | 2016-05-26 | 2016-10-26 | 哈尔滨工业大学 | Anomaly detection method of flight data of unmanned aerial vehicle based on over-sampling projection approximation basis pursuit |
| EP3261024A2 (en) * | 2016-06-23 | 2017-12-27 | Siemens Healthcare GmbH | Method and system for vascular disease detection using recurrent neural networks |
| CN107833208A (en) * | 2017-10-27 | 2018-03-23 | 哈尔滨工业大学 | A kind of hyperspectral abnormity detection method based on changeable weight depth own coding |
| CN107748899A (en) * | 2017-11-03 | 2018-03-02 | 哈尔滨工业大学深圳研究生院 | A kind of target classification of the two dimensional image based on LSTM sentences knowledge method |
| CN108063698A (en) * | 2017-12-15 | 2018-05-22 | 东软集团股份有限公司 | Unit exception detection method and device, program product and storage medium |
| CN108053653A (en) * | 2018-01-11 | 2018-05-18 | 广东蔚海数问大数据科技有限公司 | Vehicle behavior Forecasting Methodology and device based on LSTM |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108960303A (en) | 2018-12-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108960303B (en) | Unmanned aerial vehicle flight data anomaly detection method based on LSTM | |
| CN112131760B (en) | CBAM model-based prediction method for residual life of aircraft engine | |
| CN109766583B (en) | Aircraft engine life prediction method based on unlabeled, unbalanced and initial value uncertain data | |
| CN111222549B (en) | Unmanned aerial vehicle fault prediction method based on deep neural network | |
| CN112926273B (en) | Method for predicting residual life of multivariate degradation equipment | |
| CN111680875B (en) | Unmanned aerial vehicle state risk fuzzy comprehensive evaluation method based on probability baseline model | |
| CN114266278B (en) | Dual-attention network-based equipment residual service life prediction method | |
| CN108427400B (en) | Aircraft airspeed head fault diagnosis method based on neural network analytic redundancy | |
| CN110309537B (en) | Intelligent health prediction method and system for aircraft | |
| CN116448419A (en) | Zero sample bearing fault diagnosis method based on depth model high-dimensional parameter multi-target efficient optimization | |
| CN116610092A (en) | Method and system for vehicle analysis | |
| CN114266289A (en) | Complex equipment health state assessment method | |
| Ye et al. | A deep learning-based method for automatic abnormal data detection: Case study for bridge structural health monitoring | |
| CN112257914A (en) | Aviation safety cause and effect prediction method based on random forest | |
| CN115051929A (en) | Network fault prediction method and device based on self-supervision target perception neural network | |
| Sun et al. | A wireless sensor network node fault diagnosis model based on belief rule base with power set | |
| CN111126477A (en) | Learning and reasoning method of hybrid Bayesian network | |
| CN114357372A (en) | Aircraft fault diagnosis model generation method based on multi-sensor data driving | |
| CN114169091A (en) | Method for establishing prediction model of residual life of engineering mechanical part and prediction method | |
| CN102788955A (en) | Remaining lifetime prediction method of ESN (echo state network) turbine generator classification submodel based on Kalman filtering | |
| CN111079348B (en) | Method and device for detecting slowly-varying signal | |
| CN104200269A (en) | Real-time fault diagnosis method based on online learning minimum embedding dimension network | |
| CN114841063A (en) | Aero-engine residual life prediction method based on deep learning | |
| CN115221973A (en) | Aviation bearing fault diagnosis method based on enhanced weighted heterogeneous ensemble learning | |
| JP2016520220A (en) | Hidden attribute model estimation device, method and program |
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 |