CN111627505B - Cluster structure type identification method - Google Patents
Cluster structure type identification method Download PDFInfo
- Publication number
- CN111627505B CN111627505B CN202010498861.2A CN202010498861A CN111627505B CN 111627505 B CN111627505 B CN 111627505B CN 202010498861 A CN202010498861 A CN 202010498861A CN 111627505 B CN111627505 B CN 111627505B
- Authority
- CN
- China
- Prior art keywords
- atom
- cluster
- clusters
- atoms
- adjacent
- 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
Classifications
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16C—COMPUTATIONAL CHEMISTRY; CHEMOINFORMATICS; COMPUTATIONAL MATERIALS SCIENCE
- G16C20/00—Chemoinformatics, i.e. ICT specially adapted for the handling of physicochemical or structural data of chemical particles, elements, compounds or mixtures
- G16C20/70—Machine learning, data mining or chemometrics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Abstract
The invention relates to a cluster structure type identification method, which adopts the technical scheme that the most stable and local excellent structure obtained by optimizing a unitary, binary and ternary cluster based on multiple potential energy functions is taken as a cluster structure sample, the characteristic information of the structure type is extracted, the data is processed by applying a metrology technology, the processed sample is divided into a correction set and a verification set, and a correction model for cluster structure inspection is constructed by using a classification model technology and is used for identifying the verification set. The invention covers a sample database of unitary, binary and ternary clusters described by a plurality of potential functions, and lays a theoretical and method foundation for researching and developing a cluster structure type recognition technology; the invention extracts the characteristic information of the cluster structure and inputs the characteristic information into the neural network, and trains the neural network, and the trained neural network can realize accurate detection of the cluster structure type.
Description
Technical Field
The invention relates to the field of computational chemistry simulation, in particular to a cluster structure type identification method.
Background
Due to the ever-expanding scale of computational resources, computational simulation methods based on electronic structure methods or empirical force fields have advanced the atomic scale simulation of substances, including molecules, materials and biological systems. Nanoclusters have various structural and morphological characteristics, and the structure is closely related to specific properties (such as melting point, freezing point and the like). Therefore, determining the structure type of nanoclusters is of great significance to study the performance thereof. A huge number of cluster structures can be generated in computational simulation, especially in terms of identifying molecular patterns that constitute the basis of slowly evolving large system behavior. The traditional mode is to manually examine the classification and perform analysis based on iteratively designed structure descriptors. Such a schema is impractical and difficult to understand the potential relevance and complexity of hiding in a large amount of structure/property data. The method of cluster analysis or pattern recognition is expected to identify the most stable or locally excellent cluster structure type in a computer simulation mode. In addition, when a global optimization algorithm based on a potential energy function is used for optimizing the cluster structure, a huge number of structures with various structure types can be obtained. The change of the structure in the global optimization process is analyzed, the evolution process of the algorithm is facilitated to be researched, and the efficiency of the algorithm is improved.
First, fitting parameters for describing potential energy functions of clusters are different, and even if two cluster structures with completely identical shapes have different bond lengths, the extracted feature information is an important factor influencing the recognition rate; secondly, for the cluster composed of multiple elements, the nearest neighbor equilibrium distance between different kinds of atoms has difference in value, so that the cluster structure has certain distortion, thereby influencing the identification effect of the cluster structure type.
Disclosure of Invention
The invention aims to provide a method for quickly and accurately identifying the cluster structure type.
In order to achieve the purpose, the technical scheme adopted by the invention is as follows:
a cluster structure type identification method comprises the following specific steps:
s1: classification of Cluster Structure types
The cluster structure types are divided into decahedral, icosahedral, face-centered cubic, leary tetrahedron and amorphous structures, which are respectively represented by numerical values [1,2,3,4,5 ];
s2: acquisition of nanocluster structured samples
Taking the most stable structure obtained by a global optimization algorithm or the local excellent structure obtained by local optimization as a nanocluster structure sample for a unitary Lennard-Jones cluster, a binary Lennard-Jones cluster and a ternary Lennard-Jones cluster which are described based on a potential function energy function; then, at least 100 structural samples, at least 500 structural samples in total, are uniformly collected from the clusters of each structure according to the size distribution of the nanocluster structure;
s3: extraction of 14 eigenvalues of nanocluster structure
The 14 characteristic values are specifically as follows:
1) The total number of cluster atoms N;
2) Total number of bonds in cluster structure n: when the distance between any two atoms is less than 1.2 times of the nearest neighbor equilibrium distance, the two atoms are considered to be bonded;
3) Relative standard deviation of bond length RSD:
wherein x is i Is the bond length of the ith bonding bar,
is the average bond length;
4) The number of atoms for which the number of atom-adjacent links is 12;
5) The number of atoms for which the number of atom-adjacent links is 11;
6) The number of atoms for which the number of atom-adjacent links is 10;
7) The number of atoms for which the number of atom-adjacent links is 9;
8) The atom with the atom adjacent link number of 12 forms a bond angle with the other two atoms adjacent to the atom, which is less than the average of all bond angles of 90 degrees;
9) The average of all bond angles formed by an atom with the atom adjacent link number of 12 and the other two atoms adjacent to the atom is greater than 90 degrees and less than 170 degrees;
10 An atom having an atom adjacent link number of 11 forms a bond angle with two other atoms adjacent to the atom of less than an average of all bond angles of 90 degrees;
11 An atom having an atom adjacent link number of 11 forms a bond angle with the other two atoms adjacent to the atom of more than 90 degrees and less than 170 degrees of the average of all bond angles;
12 An atom with 10 adjacent links to the atom forms a bond angle with two other atoms adjacent to the atom that is less than 90 degrees;
13 An atom with 10 adjacent links to the atom forms a bond angle with two other atoms adjacent to the atom that is greater than 90 degrees and less than 170 degrees;
14 The number of triangular pyramids contained in the cluster;
s4: generation of correction set and verification set
Adopting a KS (Kennard-Stone) grouping algorithm to group the structure samples, taking 2/3-3/4 samples as a correction set, and taking 1/3-1/4 samples as a verification set;
s5: processing of sample data
The input dimension of the sample matrix is 14 characteristic numbers; and the data representing different characteristics have larger difference, carry on the normalization processing of the data, map the data to [0,1];
s6: classification recognition model construction
Constructing a classification recognition model for the cluster structure type, wherein the model is a BP neural network; the input of the model is a vector formed by 14 characteristic variables, so that the number of nodes of an input layer is 14; the model outputs a vector of a cluster structure type, so that the number of nodes of an output layer is set to be 1; the neural network is designed into two hidden layers, and the number of nodes of the first hidden layer and the second hidden layer is 8; the hidden layer response functions are nonlinear response functions, and both tan sig functions are adopted; the output layer adopts a linear response function; optimizing the model using a batch gradient descent algorithm, wherein model parameters are set as follows: the batch size is set to 200, the learning rate is set to 0.1, and the iteration number is 500;
s7: classification recognition model training
Training the classification recognition model by using the correction set, finishing the training when the iteration times are reached, returning to the step S6 if the model can not meet the requirements, and modifying the parameters of the model until the requirements are met;
s8: verification set testing
And identifying the cluster structure type of the verification set according to the parameters and the response function of the classification identification model, wherein the correct identification rate of the cluster structure type of the verification set sample is 96.00%.
Preferably, the size of the nanocluster structure sample collected in step S2 is from 13 atoms to 300 atoms.
Preferably, the unary clusters in step S2 include unary Lennard-Jones clusters, aluminum clusters, silver clusters, gold clusters, copper clusters, and cobalt clusters; the binary clusters include binary Lennard-Jones clusters, copper-gold clusters, and cobalt-palladium clusters; ternary clusters include ternary Lennard-Jones clusters, silver-palladium-platinum clusters, and gold-palladium-platinum clusters.
Preferably, the aluminum cluster is described based on an NP-B potential function, the silver cluster is described based on a Gupta potential function, the gold cluster is described based on a Gupta potential function, the copper cluster is described based on a Sutton-Chen function, and the cobalt cluster is described based on a Gupta potential function; the copper-gold cluster is described based on a Gupta potential function, and the cobalt-palladium cluster is described based on a Gupta potential function; the silver-palladium-platinum cluster is described based on a Gupta potential function, and the gold-palladium-platinum cluster is described based on a Gupta potential function.
The invention has the beneficial effects that:
(1) The invention covers the sample database of the unitary, binary and ternary clusters described by a plurality of potential functions, and lays a theoretical and method foundation for researching and developing a cluster structure type identification technology;
(2) The invention extracts the characteristic information of the cluster structure and inputs the characteristic information into the neural network, and trains the neural network, and the trained neural network can realize accurate detection of the cluster structure type.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in detail and fully with reference to the following embodiments.
The embodiment is a cluster structure type identification method, which comprises the following specific implementation steps:
s1: classification of Cluster Structure types
The cluster structure types are divided into decahedral, icosahedral, face-centered cubic, leary tetrahedron and amorphous structures, which are respectively represented by numerical values [1,2,3,4,5 ];
s2: acquisition of nanocluster structured samples
Taking the most stable structure obtained by a global optimization algorithm or the local excellent structure obtained by local optimization as a nanocluster structure sample for a unitary Lennard-Jones cluster, a binary Lennard-Jones cluster and a ternary Lennard-Jones cluster which are described based on a potential function energy function; then, 100 structural samples were uniformly collected from the cluster of each structure in accordance with the size distribution of the nanocluster structure, for a total of 500 structural samples;
the size of the collected nanocluster structure sample is from 13 atoms to 300 atoms.
The monobasic clusters include monobasic Lennard-Jones clusters, aluminum clusters, silver clusters, gold clusters, copper clusters and cobalt clusters; the binary clusters include binary Lennard-Jones clusters, copper-gold clusters, and cobalt-palladium clusters; ternary clusters include ternary Lennard-Jones clusters, silver-palladium-platinum clusters, and gold-palladium-platinum clusters.
The aluminum cluster is described based on an NP-B potential function, the silver cluster is described based on a Gupta potential function, the gold cluster is described based on a Gupta potential function, the copper cluster is described based on a Sutton-Chen function, and the cobalt cluster is described based on a Gupta potential function; the copper-gold cluster is described based on a Gupta potential function, and the cobalt-palladium cluster is described based on a Gupta potential function; the silver-palladium-platinum cluster is described based on a Gupta potential function, and the gold-palladium-platinum cluster is described based on a Gupta potential function.
S3: extraction of 14 eigenvalues of nanocluster structure
The 14 characteristic values are specifically as follows:
1) The total number of cluster atoms N;
2) Total number of bonds in cluster structure n: when the distance between any two atoms is less than 1.2 times the nearest neighbor equilibrium distance, the two atoms are considered to be bonded;
3) Relative standard deviation of bond length RSD:
wherein x is i Is the bond length of the ith key,
is the average bond length;
4) The number of atoms for which the number of atom-adjacent links is 12;
5) The number of atoms for which the number of atom-adjacent links is 11;
6) The number of atoms for which the number of atom-adjacent links is 10;
7) The number of atoms for which the number of atom-adjacent links is 9;
8) The atom with the atom adjacent link number of 12 forms a bond angle with the other two atoms adjacent to the atom that is less than the average of all bond angles of 90 degrees;
9) The average of all bond angles formed by an atom with 12 atom-adjacent chain numbers and the other two atoms adjacent to the atom is greater than 90 degrees and less than 170 degrees;
10 An atom with an atom adjacent link number of 11) forms a bond angle with two other atoms adjacent to the atom of less than 90 degrees on average of all bond angles;
11 An atom with an atom adjacent link number of 11 forms a bond angle with two other atoms adjacent to the atom of more than 90 degrees and less than 170 degrees of all bond angles;
12 An atom having 10 atom-adjacent linkages forms a bond angle with two other atoms adjacent to the atom of less than 90 degrees on average of all bond angles;
13 An atom having 10 atom-adjacent linkages forms a bond angle with two other atoms adjacent to the atom of more than 90 degrees and less than 170 degrees of the average of all bond angles;
14 The number of triangular pyramids contained in the cluster;
s4: generation of correction set and verification set
Adopting a KS (Kennard-Stone) grouping algorithm to group the structural samples, taking 2/3-3/4 samples as a correction set, and taking 1/3-1/4 samples as a verification set;
s5: processing of sample data
The input dimension of the sample matrix is 14 characteristic numbers; the data representing different characteristics have larger difference, data normalization processing is carried out, and the data are mapped to [0,1];
s6: classification recognition model construction
Constructing a classification recognition model for the cluster structure type, wherein the model is a BP neural network; the input of the model is a vector formed by 14 characteristic variables, so that the number of nodes of an input layer is 14; the model outputs a vector of a cluster structure type, so that the number of nodes of an output layer is set to be 1; the neural network is designed into two hidden layers, and the number of nodes of the first hidden layer and the second hidden layer is 8; hidden layer response functions are nonlinear response functions, and tansig functions are adopted; the output layer adopts a linear response function; optimizing the model using a batch gradient descent algorithm, wherein model parameters are set as follows: the batch size is set to 200, the learning rate is set to 0.1, and the iteration number is 500;
s7: classification recognition model training
Training the classification recognition model by using the correction set, finishing training after the iteration times are reached, and returning to the step S6 to modify the model parameters until the requirements are met if the model cannot meet the requirements;
s8: verification set testing
And identifying the cluster structure type of the verification set according to the parameters and the response function of the classification identification model, wherein the correct identification rate of the cluster structure type of the verification set sample is 96.00%.
Claims (4)
1. A cluster structure type identification method is characterized in that: the method comprises the following specific steps:
s1: classification of Cluster Structure types
The cluster structure types are divided into decahedral, icosahedral, face-centered cubic, leary tetrahedron and amorphous structures, which are respectively represented by numerical values [1,2,3,4,5 ];
s2: acquisition of nanocluster structured samples
For the unitary clusters, binary clusters and ternary clusters described based on the potential function energy function, obtaining the most stable structure through a global optimization algorithm or obtaining a local excellent structure through local optimization as a nanocluster structure sample; then, at least 100 structural samples, at least 500 structural samples in total, are uniformly collected from the clusters of each structure according to the size distribution of the nanocluster structure;
s3: extraction of 14 eigenvalues of nanocluster structure
The 14 characteristic values are specifically as follows:
1) The total number of cluster atoms N;
2) Total number of bonds in cluster structure n: when the distance between any two atoms is less than 1.2 times of the nearest neighbor equilibrium distance, the two atoms are considered to be bonded;
3) Relative standard deviation of bond length RSD:
wherein x is i Is the bond length of the ith bonding bar,
is the average bond length;
4) The number of atoms for which the number of atom-adjacent links is 12;
5) The number of atoms for which the number of atom-adjacent links is 11;
6) The number of atoms for which the number of atom-adjacent links is 10;
7) The number of atoms for which the number of atom-adjacent links is 9;
8) The atom with the atom adjacent link number of 12 forms a bond angle with the other two atoms adjacent to the atom, which is less than the average of all bond angles of 90 degrees;
9) The average of all bond angles formed by an atom with 12 atom-adjacent chain numbers and the other two atoms adjacent to the atom is greater than 90 degrees and less than 170 degrees;
10 An atom having an atom adjacent link number of 11 forms a bond angle with two other atoms adjacent to the atom of less than an average of all bond angles of 90 degrees;
11 An atom having an atom adjacent link number of 11 forms a bond angle with the other two atoms adjacent to the atom of more than 90 degrees and less than 170 degrees of the average of all bond angles;
12 An atom with 10 adjacent links to the atom forms a bond angle with two other atoms adjacent to the atom that is less than 90 degrees;
13 An atom with 10 adjacent links to the atom forms a bond angle with two other atoms adjacent to the atom that is greater than 90 degrees and less than 170 degrees;
14 The number of triangular pyramids contained in the cluster;
s4: generation of correction set and verification set
Adopting a KS (Kennard-Stone) grouping algorithm to group the structural samples, taking 2/3-3/4 samples as a correction set, and taking 1/3-1/4 samples as a verification set;
s5: processing of sample data
The input dimension of the sample matrix is 14 characteristic numbers; and the data representing different characteristics have larger difference, carry on the normalization processing of the data, map the data to [0,1];
s6: classification recognition model construction
Constructing a classification recognition model for the cluster structure type, wherein the model is a BP neural network; the input of the model is a vector formed by 14 characteristic variables, so that the number of nodes of an input layer is 14; the model outputs a vector of a cluster structure type, so that the number of nodes of an output layer is set to be 1; the neural network is designed into two hidden layers, and the number of nodes of the first hidden layer and the second hidden layer is 8; hidden layer response functions are nonlinear response functions, and tansig functions are adopted; the output layer adopts a linear response function; optimizing the model using a batch gradient descent algorithm, wherein model parameters are set as follows: the batch size is set to 200, the learning rate is set to 0.1, and the iteration number is 500;
s7: classification recognition model training
Training the classification recognition model by using the correction set, finishing training after the iteration times are reached, and returning to the step S6 to modify the model parameters until the requirements are met if the model cannot meet the requirements;
s8: verification set testing
And identifying the cluster structure type of the verification set according to the parameters and the response function of the classification identification model, wherein the correct identification rate of the cluster structure type of the verification set sample is 96.00%.
2. The cluster structure type identification method according to claim 1, wherein: the size of the nanocluster structure sample collected in step S2 is from 13 atoms to 300 atoms.
3. The cluster structure type identification method according to claim 1, wherein: the monobasic clusters in the step S2 comprise monobasic Lennard-Jones clusters, aluminum clusters, silver clusters, gold clusters, copper clusters and cobalt clusters; the binary clusters include binary Lennard-Jones clusters, copper-gold clusters, and cobalt-palladium clusters; ternary clusters include ternary Lennard-Jones clusters, silver-palladium-platinum clusters, and gold-palladium-platinum clusters.
4. The cluster structure type identification method according to claim 3, wherein: the aluminum cluster is described based on an NP-B potential function, the silver cluster is described based on a Gupta potential function, the gold cluster is described based on a Gupta potential function, the copper cluster is described based on a Sutton-Chen function, and the cobalt cluster is described based on a Gupta potential function; the copper-gold cluster is described based on a Gupta potential function, and the cobalt-palladium cluster is described based on a Gupta potential function; the silver-palladium-platinum clusters are described based on a Gupta potential function, and the gold-palladium-platinum clusters are described based on a Gupta potential function.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010498861.2A CN111627505B (en) | 2020-06-04 | 2020-06-04 | Cluster structure type identification method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202010498861.2A CN111627505B (en) | 2020-06-04 | 2020-06-04 | Cluster structure type identification method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111627505A CN111627505A (en) | 2020-09-04 |
| CN111627505B true CN111627505B (en) | 2023-04-07 |
Family
ID=72260145
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202010498861.2A Active CN111627505B (en) | 2020-06-04 | 2020-06-04 | Cluster structure type identification method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111627505B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112216355B (en) * | 2020-10-22 | 2024-01-26 | 哈尔滨理工大学 | Multi-component crystal configuration energy prediction method based on machine learning |
| CN113270149B (en) * | 2021-04-30 | 2022-10-04 | 上海大学 | Molecular dynamics-based characterization method for growth and structure of titanyl clusters in molten iron |
| CN113326664B (en) * | 2021-06-28 | 2022-10-21 | 南京玻璃纤维研究设计院有限公司 | Method for predicting dielectric constant of glass based on M5P algorithm |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2003106989A (en) * | 2001-09-27 | 2003-04-09 | Asahi Kasei Corp | Method and device for predicting adhesion of high polymer material and method and device for determining parameter describing roughly visualized model |
| CN102472750A (en) * | 2009-08-19 | 2012-05-23 | 皇家飞利浦电子股份有限公司 | Detection of different target components by cluster formation |
| CN107563104A (en) * | 2017-10-18 | 2018-01-09 | 安庆师范大学 | Binary Clusters structural optimization method based on simulated annealing optimization algorithm |
-
2020
- 2020-06-04 CN CN202010498861.2A patent/CN111627505B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2003106989A (en) * | 2001-09-27 | 2003-04-09 | Asahi Kasei Corp | Method and device for predicting adhesion of high polymer material and method and device for determining parameter describing roughly visualized model |
| CN102472750A (en) * | 2009-08-19 | 2012-05-23 | 皇家飞利浦电子股份有限公司 | Detection of different target components by cluster formation |
| CN107563104A (en) * | 2017-10-18 | 2018-01-09 | 安庆师范大学 | Binary Clusters structural optimization method based on simulated annealing optimization algorithm |
Non-Patent Citations (1)
| Title |
|---|
| 银金钯铂组成对三元金属团簇结构的影响;吴夏等;《计算机与应用化学》(第10期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111627505A (en) | 2020-09-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111627505B (en) | Cluster structure type identification method | |
| CN106845717B (en) | Energy efficiency evaluation method based on multi-model fusion strategy | |
| CN106874688B (en) | Intelligent lead compound based on convolutional neural networks finds method | |
| Wang et al. | Identification of apple leaf diseases by improved deep convolutional neural networks with an attention mechanism | |
| CN110197205A (en) | A kind of image-recognizing method of multiple features source residual error network | |
| CN109241995B (en) | Image identification method based on improved ArcFace loss function | |
| CN111681718B (en) | Medicine relocation method based on deep learning multi-source heterogeneous network | |
| CN113392748B (en) | Remote sensing image cultivated land information extraction method based on convolutional neural network | |
| CN112529638B (en) | Service demand dynamic prediction method and system based on user classification and deep learning | |
| CN112766283A (en) | Two-phase flow pattern identification method based on multi-scale convolution network | |
| CN105046323A (en) | Regularization-based RBF network multi-label classification method | |
| CN114821340A (en) | Land utilization classification method and system | |
| Rhee | Composition of visual feature vector pattern for deep learning in image forensics | |
| Ullah et al. | Machine health surveillance system by using deep learning sparse autoencoder | |
| CN107220320B (en) | Emerging technology identification method based on patent citation | |
| CN113313830A (en) | Encoding point cloud feature extraction method based on multi-branch graph convolutional neural network | |
| CN109409407A (en) | A kind of industry monitoring data clustering method based on LE algorithm | |
| CN116883364A (en) | Apple leaf disease identification method based on CNN and Transformer | |
| CN116108751A (en) | Material stress-strain curve prediction model based on graph neural network, construction method and prediction method thereof | |
| CN114926702B (en) | Small sample image classification method based on depth attention measurement | |
| CN110309528A (en) | A kind of radar Design Method based on machine learning | |
| CN115688990A (en) | Nonferrous metal industry electric carbon data matching early warning method based on BP neural network | |
| CN109472319B (en) | Three-dimensional model classification method and retrieval method | |
| CN110175625B (en) | WeChat information identification and management method based on improved SSD algorithm | |
| CN116698410B (en) | Rolling bearing multi-sensor data monitoring method based on convolutional 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 |