CN114863209B - Unsupervised domain adaptation modeling method, system, equipment and medium for category proportion guidance - Google Patents
Unsupervised domain adaptation modeling method, system, equipment and medium for category proportion guidance Download PDFInfo
- Publication number
- CN114863209B CN114863209B CN202210425027.XA CN202210425027A CN114863209B CN 114863209 B CN114863209 B CN 114863209B CN 202210425027 A CN202210425027 A CN 202210425027A CN 114863209 B CN114863209 B CN 114863209B
- Authority
- CN
- China
- Prior art keywords
- sample
- domain
- logging
- target domain
- training
- 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
- 238000000034 method Methods 0.000 title claims abstract description 34
- 230000006978 adaptation Effects 0.000 title claims description 16
- 239000011159 matrix material Substances 0.000 claims abstract description 43
- 230000009466 transformation Effects 0.000 claims abstract description 23
- 238000013507 mapping Methods 0.000 claims abstract description 21
- 230000003044 adaptive effect Effects 0.000 claims abstract description 10
- 238000002360 preparation method Methods 0.000 claims abstract description 7
- 208000035126 Facies Diseases 0.000 claims description 13
- 238000005457 optimization Methods 0.000 claims description 7
- 230000006870 function Effects 0.000 claims description 5
- 238000011478 gradient descent method Methods 0.000 claims description 5
- 230000005251 gamma ray Effects 0.000 claims description 3
- 238000003860 storage Methods 0.000 claims description 3
- 238000009826 distribution Methods 0.000 description 13
- 238000010801 machine learning Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000010365 information processing Effects 0.000 description 1
- 230000001537 neural effect Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000011426 transformation method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/77—Processing image or video features in feature spaces; using data integration or data reduction, e.g. principal component analysis [PCA] or independent component analysis [ICA] or self-organising maps [SOM]; Blind source separation
- G06V10/774—Generating sets of training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02A—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE
- Y02A10/00—TECHNOLOGIES FOR ADAPTATION TO CLIMATE CHANGE at coastal zones; at river basins
- Y02A10/40—Controlling or monitoring, e.g. of flood or hurricane; Forecasting, e.g. risk assessment or mapping
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Computing Systems (AREA)
- Software Systems (AREA)
- Databases & Information Systems (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Health & Medical Sciences (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Mathematical Analysis (AREA)
- Multimedia (AREA)
- Computational Mathematics (AREA)
- Medical Informatics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Algebra (AREA)
- General Engineering & Computer Science (AREA)
- Image Analysis (AREA)
Abstract
The application discloses an unsupervised field adaptive modeling method, a system, equipment and a medium for class proportion guidance, which comprise the following steps: data preparation and initialization: collecting training data set, wherein the training data set at least comprises a training sampleTraining source domain classification: performing random Fourier feature transformation on the source domain samples to obtain a source domain mapping sample matrixTraining the target domain classification: carrying out Fourier feature transformation on the target domain sample, wherein parameters of the Fourier feature transformation are consistent with the RFF in the random seed and source domain, and obtaining a target domain mapping sample matrix H t The target domain classifier is described as f t (x)=φ(x)B t Thereby obtaining an optimal source domain output weight matrixOutputting target domain classification: output ofAnd obtaining the target domain classifier.
Description
Technical Field
The application relates to the technical field of adaptive modeling, in particular to an unsupervised field adaptive modeling method, system, equipment and medium for class proportion guidance.
Background
Machine learning can be used to solve the modeling problem of complex unknown models, and is widely applied in many fields. Since modeling is data driven, model accuracy is closely related to data quality. In reality, all data cannot be obtained, so that the training data and the data generated by the real scene have larger probability distribution deviation, and the problem of model accuracy reduction and even failure is easily caused. For example, in geophysical well logging interpretation, the well log data of the new well deviates significantly from the well log data distribution of the interpreted well, presenting a significant challenge for machine learning applications in well logging interpretation model building. For the distribution difference problem, a domain adaptation method can be adopted to solve, and for the scene without any label of the target domain, the method is limited to unsupervised domain adaptation. The field adaptation method mainly comprises the following steps: sample weighting based, distribution alignment based, and model adjustment based domain adaptation methods. Wherein the sample weighting is mainly used for smaller distribution deviation scenes; distribution alignment depends on the quality of the pseudo tag of the target domain, and most aim at scenes with difference between edge distribution and conditional distribution; model adjustment has the widest use degree, but a certain label is needed to exist in a target domain, so that the model adjustment is difficult to be qualified for a scene without the label in the target domain. To sum up, existing techniques do not fit the scenario of a priori distribution differences.
Disclosure of Invention
The application mainly aims to provide an unsupervised domain adaptation modeling method, device, equipment and medium guided by category proportions, and aims to solve the technical problem of domain adaptation in which the edge distribution and priori distribution of a source domain and a target domain are changed in a classification task.
In order to achieve the above object, the present application provides an unsupervised domain adaptation modeling method of class ratio guidance, comprising the steps of:
data preparation and initialization: collecting training data set, wherein the training data set at least comprises a training sample
Training source domain classification: performing random Fourier feature transformation on the source domain samples to obtain a source domain mapping sample matrix
Training the target domain classification: performing target domain sampleFourier feature transformation, wherein parameters of the fourier feature transformation are consistent with the random seeds and RFF in the source domain, so as to obtain a target domain mapping sample matrix H t The target domain classifier is described as f t (x)=φ(x)B t Thereby obtaining an optimal source domain output weight matrix
Outputting target domain classification: output ofAnd obtaining the target domain classifier.
Preferably, the training sampleIn (a): d is the initial feature dimension of the sample, and the label corresponding to the sample is Representing a real number domain, wherein the label adopts single-heat coding;
let the collected source domain samples have n s Each sample has a label, and the source domain sample set is For the i-th sample of the source domain, +.>Is->A corresponding tag; the target domain sample has n t But without any tag, the target domain sample set is +.> An ith sample of the target domain;
manually setting a training balance coefficient gamma 1 ,η 1 ,η 2 >0, manually setting a proportional matrixΞ=diag(p),Is a target domain class scale vector.
Preferably, the source domain maps a sample matrixk is the mapped sample dimension, and the source domain classifier is described as f s (x)=φ(x)B S ,/>For the mapping function +.>To regenerate the kernel Hilbert space, the +.>Namely:
wherein ,
preferably, the characteristic transformation modes of the source domain and the target domain are completely consistent, and the target domain classifier f t (x)=φ(x)B t In (3) solving the following optimization problemNamely:
wherein L is according toThe resulting graph laplace matrix.
Preferably, the target domain classifier f t (x)=φ(x)B t Training by adopting a gradient descent method, namely:
wherein, delta represents the learning rate,representing B obtained by learning in the r-th step t ,/>Represents B obtained by learning in the r+1 step t And (2) and
the application also relates to an unsupervised domain adaptive modeling system guided by category proportions, comprising:
data preparation and initialization module: for collecting a training data set comprising at least one training sample
Training a source domain classifier, and carrying out random Fourier on a source domain sampleThen the matrix of source domain mapping samples is obtained by the transformation of the inner She Tezheng
Training a target domain classifier: performing Fourier feature transformation on the target domain samples to obtain a target domain mapping sample matrix H t The target domain classifier is described as f t (x)=φ(x)B t ;
Outputting a target domain classifier: output ofAnd obtaining the target domain classifier.
The application also relates to an unsupervised domain adaptation modeling device of class ratio guidance, the electronic device comprising:
at least one processor; the method comprises the steps of,
a memory communicatively coupled to the at least one processor; wherein,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the steps of the one type of class ratio guided unsupervised domain adaptation modeling method described above.
The present application also relates to a computer-readable storage medium having stored thereon a program for implementing a class-ratio guided unsupervised domain adapted modeling method, the program for implementing a class-ratio guided unsupervised domain adapted modeling method being executed by a processor to implement the steps of the above-described class-ratio guided unsupervised domain adapted modeling method.
Compared with the prior art, the method can effectively solve the technical problem of field adaptation in which the edge distribution and the prior distribution of the source domain and the target domain in the classification task are changed, and has the advantages of strong nonlinear fitting capability, high training speed and high model accuracy.
Drawings
The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the application and together with the description, serve to explain the principles of the application.
In order to more clearly illustrate the embodiments of the application or the technical solutions of the prior art, the drawings which are used in the description of the embodiments or the prior art will be briefly described, and it will be obvious to a person skilled in the art that other drawings can be obtained from these drawings without inventive effort.
FIG. 1 is a schematic flow chart of an unsupervised domain adaptive modeling method guided by class proportions.
The achievement of the objects, functional features and advantages of the present application will be further described with reference to the accompanying drawings, in conjunction with the embodiments.
Detailed Description
In order to make the above objects, features and advantages of the present application more comprehensible, the following description of the embodiments accompanied with the accompanying drawings will be given in detail. It will be apparent that the described embodiments are only some, but not all, embodiments of the application. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
Example 1
The embodiment of the application provides an unsupervised domain adaptation modeling method guided by category proportions, which comprises the following steps of:
step one, data preparation and initialization
Collecting training data set, wherein the training data set at least comprises a training sample and a sampled is the initial feature dimension of the sample, and the label corresponding to the sample is +.> Representing a real number domain, wherein the label adopts single-heat coding, and c is the total number of categories;
let the collected source domain samples have n s Each sample has a label, and the source domain sample set is For the i-th sample of the source domain, +.>Is->A corresponding tag; the target domain sample has n t But without any tag, the target domain sample set is +.> An ith sample of the target domain;
manually setting a training balance coefficient gamma 1 ,η 1 ,η 2 >0, manually setting a proportional matrixΞ=diag(p),A target domain class proportion vector;
step two, training a source domain classifier
Performing random Fourier feature (Random Fourier Features, RFF) transformation on the source domain samples to obtain a source domain mapping sample matrixk isThe mapped sample dimension, the source domain classifier can be described as f s (x)=φ(x)B s ,For the mapping function +.>To regenerate the nuclear Hilbert space, B s Outputting a weight matrix for the source domain, solving the following optimization problem by solving>Namely:
wherein ,B s the solution of (2) is resolved, and then an optimal source domain output weight matrix is obtained>
Step three, training a target domain classifier
RFF transformation is carried out on the target domain sample, wherein the parameters of the RFF transformation are consistent with the RFF in the source domain of the random seed, namely the characteristic transformation modes of the source domain and the target domain are completely consistent, and a target domain mapping sample matrix H is obtained t The target domain classifier can be described as f t (x)=φ(x)B t ,B t Outputting a weight matrix for the source domain, solving by solving the following optimization problemNamely:
wherein L is according toThe obtained graph Laplace matrix; b (B) t The solution of (1) needs to adopt a gradient descent method so as to obtain an optimal source domain output weight matrix +.>T represents the matrix transpose, tr is the trace of the matrix;
further, the target domain classifier f t (x)=φ(x)B t Training by adopting a gradient descent method, namely:
wherein, delta represents the learning rate,representing B obtained by learning in the r-th step t ,/>Represents B obtained by learning in the r+1 step t And (2) and
fourth, outputting a target domain classifier
Output ofAnd obtaining the target domain classifier.
Example 2
This embodiment is illustrated with geophysical well logging interpretation as an example: for example, the sedimentary facies of unexplained wells are deep lake facies, often developing large sections of mudstones, and setting classification targets as mudstones and sandstones, the steps are as follows:
step one, data preparation and initialization
Collecting a logging sample formed by geophysical logging curves (such as acoustic logging curves, gamma ray logging curves and natural potential logging curves) at a certain depthd represents the total number of logging categories used (i.e. feature dimension), the sample corresponds to the label +.> Representing a real number domain, wherein the label adopts single-heat coding, the physical meaning of the label can be mudstone and sandstone, and c is the total number of categories;
for interpreted wells, a labeled source field can be obtained if there is n along the depth s Logging values for a plurality of depth points, then the collected source domain samples have n s Each sample has a label, and the source domain sample set is For the i-th sample of the source domain, +.>Is->A corresponding tag; unexplained wells, i.e. unlabeled target fields, requiring prediction if there is n along the depth t Logging values for a plurality of depth points, the collected target domain samples have n t The target domain sample set is +.> An ith sample of the target domain;
manually setting a training balance coefficient gamma 1 ,η 1 ,η 2 >0, manually setting a proportional matrixΞ=diag(p),A target domain class proportion vector; for example, if the sedimentary facies of unexplained wells are deep lake facies, often develop large sections of mudstone, and if the classification targets are mudstone and sandstone, then +.>If the sedimentary facies of the unexplained well are in the form of a shallow beach, it is often the case that large sections of sandstone or mudstone sandstone are alternately present, then +.>The specific setting is based on the actual experience of the geologist.
Step two, training a source domain classifier
Performing random Fourier feature (Random Fourier Features, RFF) transformation on the source domain samples to obtain a source domain mapping sample matrixk is the mapped sample dimension, and the source domain classifier can be described as f s (x)=φ(x)B S ,For the mapping function +.>To regenerate the nuclear Hilbert space, B S Outputting a weight matrix for the source domain, solving the following optimization problem by solving>Namely:
wherein ,B S the solution of (2) is resolved, and then an optimal source domain output weight matrix is obtained>
Step three, training a target domain classifier
RFF transformation is carried out on the target domain sample, wherein the parameters of the RFF transformation are consistent with the RFF in the source domain of the random seed, namely the characteristic transformation modes of the source domain and the target domain are completely consistent, and a target domain mapping sample matrix H is obtained t The target domain classifier can be described as f t (x)=φ(x)B t ,B t Outputting a weight matrix for the source domain, solving by solving the following optimization problemNamely:
wherein L is according toThe obtained graph Laplace matrix; b (B) t The solution of (1) needs to adopt a gradient descent method so as to obtain an optimal source domain output weight matrix +.>T represents the matrix transpose, tr is the trace of the matrix;
further, the target domain classifier f t (x)=φ(x)B t By using laddersTraining by a degree-dropping method, namely:
wherein, delta represents the learning rate,representing B obtained by learning in the r-th step t ,/>Represents B obtained by learning in the r+1 step t And (2) and
fourth, outputting a target domain classifier
Output ofAnd obtaining the target domain classifier.
Further description in the above technical document is given below:
the random fourier feature transformation referred to in the present application is described in the paper Rahimi, ali, and Benjamin recht, "Random features for large-scale kernel machines," Advances in neural information processing systems (2007), and the feature transformation method referred to in algorithm 1 may be specifically used.
The sample similarity involved in the construction of the graph laplace matrix may be the euclidean distance, and normalization of the graph laplace matrix is required.
The foregoing description is only of the preferred embodiments of the present application, and is not intended to limit the scope of the application, but rather is intended to cover any equivalents of the structures or equivalent processes disclosed herein, or any application, directly or indirectly, in the field of other relevant technology.
Claims (5)
1. The non-supervision domain adaptive modeling method guided by the class proportion is used for geophysical logging interpretation, and classification targets are output through the obtained target domain classifier; the method is characterized by comprising the following steps of:
data preparation and initialization: collecting training data set, wherein the training data set at least comprises a training sampleThe training sample is a logging sample; the method specifically comprises the following steps:
combining geophysical log curves of a certain depth into a log sampled represents the total number of logging categories used, i.e., the characteristic dimension of the logging sample; the label corresponding to the logging sample is +.> Representing a real number domain, wherein the label adopts single-heat coding, the physical meaning of the label is mudstone and sandstone, and c is the total number of categories; the geophysical well logging curves comprise acoustic well logging curves, gamma ray well logging curves and natural potential well logging curves;
for interpreted wells, a tagged source domain can be obtained if there is n along the depth s Logging values for multiple depth points, then the collected source domain log n s Each logging sample is provided with a label, and the source domain logging sample set is For the ith log sample in the source domain log sample set,/for the source domain log sample set>Is->A corresponding tag; for an unexplained well, i.e., a target field without labels, a prediction is needed if there is n along the depth t Logging values of the depth points, the collected target domain logging samples have n t The target domain logging sample is unlabeled, and the target domain logging sample set is +.> An ith sample of the set of target domain log samples;
training source domain classification: performing random Fourier feature transformation on the source domain logging samples to obtain a source domain mapping sample matrixThe source domain mapping sample matrix->k is the dimension of the mapped log sample, and the source domain classifier is described as f s (x)=φ(x)B s ,/>For the mapping function +.>To regenerate the kernel Hilbert space, the +.>Namely:
wherein ,
training the target domain classification: performing Fourier feature transformation on the target domain logging sample, wherein parameters of the Fourier feature transformation are consistent with the random seeds and RFF in the source domain, and obtaining a target domain mapping sample matrix H t The target domain classifier is described as f t (x)=φ(x)B t Thereby obtaining an optimal source domain output weight matrixTarget domain classifier f t (x)=φ(x)B t Is solved by solving the following optimization problem>Namely:
wherein L is according toThe obtained graph Laplace matrix;
outputting target domain classification: output ofObtaining a target domain classifier;
γ 1 ,η 1 ,η 2 >the training balance coefficient is set for the human body,setting a proportion matrix for manual work>=diag(p),A target domain class proportion vector; if not explainedThe sedimentary facies of the well are deep lake facies, the classification targets of the target domain classifier are mudstone and sandstone, and the +.>If the sedimentary facies of the unexplained well is a shallow lake facies, then the method is set
2. The class-scale guided unsupervised domain adaptation modeling method of claim 1, wherein the target domain classifier f t (x)=φ(x)B t Training by adopting a gradient descent method, namely:
wherein, delta represents the learning rate,representing B obtained by learning in the r-th step t ,/>Represents B obtained by learning in the r+1 step t And (2) and
3. the non-supervision domain adaptive modeling system guided by the category proportion is used for geophysical logging interpretation, and classification targets are output through the obtained target domain classifier; characterized by comprising the following steps:
data preparation and initialization module: for collecting training data set comprising at least one training sampleThe training sample is a logging sample; the method specifically comprises the following steps:
combining geophysical log curves of a certain depth into a log sampled represents the total number of logging categories used, i.e., the characteristic dimension of the logging sample; the label corresponding to the logging sample is +.> Representing a real number domain, wherein the label adopts single-heat coding, the physical meaning of the label is mudstone and sandstone, and c is the total number of categories; the geophysical well logging curves comprise acoustic well logging curves, gamma ray well logging curves and natural potential well logging curves;
for interpreted wells, a tagged source domain can be obtained if there is n along the depth s Logging values for multiple depth points, then the collected source domain log n s Each logging sample is provided with a label, and the source domain logging sample set is For the ith log sample in the source domain log sample set,/for the source domain log sample set>Is->A corresponding tag; for an unexplained well, i.e., a target field without labels, a prediction is needed if there is n along the depth t Logging values of the depth points, the collected target domain logging samples have n t The target domain logging sample is unlabeled, and the target domain logging sample set is +.> An ith sample of the set of target domain log samples;
training a source domain classifier, and performing random Fourier feature transformation on a source domain logging sample to obtain a source domain mapping sample matrixThe source domain mapping sample matrix->k is the dimension of the mapped log sample, and the source domain classifier is described as f s (x)=φ(x)B s ,/>For the mapping function +.>To regenerate the kernel Hilbert space, the +.>Namely:
wherein ,
training a target domain classifier: fourier characteristic transformation is carried out on the target domain logging sample to obtain a target domain mapping sample matrix H t The target domain classifier is described as f t (x)=φ(x)B t The method comprises the steps of carrying out a first treatment on the surface of the Thereby obtaining the optimal source domain output weight matrixTarget domain classifier f t (x)=φ(x)B t Is solved by solving the following optimization problem>Namely:
wherein L is according toThe obtained graph Laplace matrix;
outputting a target domain classifier: output ofObtaining a target domain classifier;
γ 1 ,η 1 ,η 2 >the training balance coefficient is set for the human body,setting a proportion matrix for manual work>=diag(p),A target domain class proportion vector; if the sedimentary facies of the unexplained well are deep lake facies, the classification targets of the target domain classifier are mudstone and sandstone, setting +.>If the sedimentary facies of the unexplained well is a shallow lake facies, then the method is set
4. An unsupervised domain adaptation modeling apparatus of class ratio guidance, characterized in that the electronic apparatus comprises:
at least one processor; the method comprises the steps of,
a memory communicatively coupled to the at least one processor; wherein,
the memory stores instructions executable by the at least one processor to enable the at least one processor to perform the steps of a class scale guided unsupervised domain adaptation modeling method according to any one of claims 1 to 2.
5. A computer-readable storage medium, characterized in that a program for realizing a category-scale guided unsupervised domain adaptive modeling method is stored on the computer-readable storage medium, the program for realizing a category-scale guided unsupervised domain adaptive modeling method being executed by a processor to realize the steps of the category-scale guided unsupervised domain adaptive modeling method according to any one of claims 1 to 2.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210425027.XA CN114863209B (en) | 2022-04-21 | 2022-04-21 | Unsupervised domain adaptation modeling method, system, equipment and medium for category proportion guidance |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210425027.XA CN114863209B (en) | 2022-04-21 | 2022-04-21 | Unsupervised domain adaptation modeling method, system, equipment and medium for category proportion guidance |
Publications (2)
Publication Number | Publication Date |
---|---|
CN114863209A CN114863209A (en) | 2022-08-05 |
CN114863209B true CN114863209B (en) | 2023-10-20 |
Family
ID=82632471
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210425027.XA Active CN114863209B (en) | 2022-04-21 | 2022-04-21 | Unsupervised domain adaptation modeling method, system, equipment and medium for category proportion guidance |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114863209B (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN116167288B (en) * | 2023-04-19 | 2023-07-11 | 中国科学技术大学 | UA2FT semi-supervised transfer learning modeling method, medium and equipment |
CN117173350B (en) * | 2023-08-09 | 2024-07-05 | 中国科学技术大学 | Geological modeling method, system and medium based on active domain adaptive learning |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107490810A (en) * | 2017-08-17 | 2017-12-19 | 中国地质大学(北京) | Based on wind field, material resource, the reservoir sand body prediction method and apparatus of basin Systems |
CN113780346A (en) * | 2021-08-06 | 2021-12-10 | 中国科学技术大学 | Method and system for adjusting prior constraint classifier and readable storage medium |
CN113869451A (en) * | 2021-10-12 | 2021-12-31 | 中国民航大学 | Rolling bearing fault diagnosis method under variable working conditions based on improved JGSA algorithm |
CN114358123A (en) * | 2021-12-03 | 2022-04-15 | 华南理工大学 | Generalized open set fault diagnosis method based on deep countermeasure migration network |
Family Cites Families (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US11074495B2 (en) * | 2013-02-28 | 2021-07-27 | Z Advanced Computing, Inc. (Zac) | System and method for extremely efficient image and pattern recognition and artificial intelligence platform |
-
2022
- 2022-04-21 CN CN202210425027.XA patent/CN114863209B/en active Active
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107490810A (en) * | 2017-08-17 | 2017-12-19 | 中国地质大学(北京) | Based on wind field, material resource, the reservoir sand body prediction method and apparatus of basin Systems |
CN113780346A (en) * | 2021-08-06 | 2021-12-10 | 中国科学技术大学 | Method and system for adjusting prior constraint classifier and readable storage medium |
CN113869451A (en) * | 2021-10-12 | 2021-12-31 | 中国民航大学 | Rolling bearing fault diagnosis method under variable working conditions based on improved JGSA algorithm |
CN114358123A (en) * | 2021-12-03 | 2022-04-15 | 华南理工大学 | Generalized open set fault diagnosis method based on deep countermeasure migration network |
Non-Patent Citations (3)
Title |
---|
Cross-Well Lithology Identification Based on Wavelet Transform and Adversarial Learning;Longxiang Sun等;《Energies》;第16卷(第3期);1-17 * |
Generalize then Adapt: Source-Free Domain Adaptive Semantic Segmentation;Jogendra Nath Kundu等;《arXiv》;1-21 * |
LogRegX: An Explainable Regression Network for Cross-Well Geophysical Logs Generation;Wenjun Lv 等;《IEEE Transactions on Instrumentation and Measurement》;第72卷;1-11 * |
Also Published As
Publication number | Publication date |
---|---|
CN114863209A (en) | 2022-08-05 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN114863209B (en) | Unsupervised domain adaptation modeling method, system, equipment and medium for category proportion guidance | |
CN103400143B (en) | A kind of data Subspace clustering method based on various visual angles | |
CN103824088B (en) | SAR target variant recognition method based on multi-information joint dynamic sparse representation | |
CN112836802B (en) | Semi-supervised learning method, lithology prediction method and storage medium | |
CN109255289A (en) | A kind of across aging face identification method generating model based on unified formula | |
CN108897042A (en) | Organic matter content earthquake prediction method and device | |
Langer et al. | Advantages and pitfalls of pattern recognition: selected cases in geophysics | |
CN108154133A (en) | Human face portrait based on asymmetric combination learning-photo array method | |
CN103268607A (en) | Common object detection method on weak supervision condition | |
CN106250918B (en) | A kind of mixed Gauss model matching process based on improved soil-shifting distance | |
Xiao et al. | Deep-learning for ionogram automatic scaling | |
US20230306559A1 (en) | Point cloud denoising method based on multi-level attention perception | |
Wang et al. | Seismic stratum segmentation using an encoder–decoder convolutional neural network | |
CN114972904B (en) | Zero sample knowledge distillation method and system based on fighting against triplet loss | |
Huang et al. | Machine learning using U-net convolutional neural networks for the imaging of sparse seismic data | |
CN113780346B (en) | Priori constraint classifier adjustment method, system and readable storage medium | |
CN115019132A (en) | Multi-target identification method for complex background ship | |
CN104866905A (en) | Nonparametric sparse tensor dictionary learning method based on beta process | |
Kaur et al. | Deep-learning-based 3D fault detection for carbon capture and storage | |
CN104008373B (en) | SAR target identification methods based on multi information dictionary learning | |
CN108121964A (en) | Joint sparse locality preserving projections face identification method based on matrix | |
CN112766381A (en) | Attribute-guided SAR image generation method under limited sample | |
CN115639605B (en) | Automatic identification method and device for high-resolution fault based on deep learning | |
CN106095811A (en) | A kind of image search method of the discrete Hash of supervision based on optimum code | |
Dixit et al. | Genetic-evolutionary adaptive moment estimation-based semisupervised deep sequential convolution network for seismic impedance inversion: Application and uncertainty analysis |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |