CN114882270A - Aortic dissection CT image classification method based on particle swarm optimization algorithm - Google Patents
Aortic dissection CT image classification method based on particle swarm optimization algorithm Download PDFInfo
- Publication number
- CN114882270A CN114882270A CN202210395179.XA CN202210395179A CN114882270A CN 114882270 A CN114882270 A CN 114882270A CN 202210395179 A CN202210395179 A CN 202210395179A CN 114882270 A CN114882270 A CN 114882270A
- Authority
- CN
- China
- Prior art keywords
- neural network
- classification
- layer
- convolutional neural
- aortic dissection
- 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.)
- Pending
Links
- 239000002245 particle Substances 0.000 title claims abstract description 97
- 238000005457 optimization Methods 0.000 title claims abstract description 53
- 208000002251 Dissecting Aneurysm Diseases 0.000 title claims abstract description 46
- 206010002895 aortic dissection Diseases 0.000 title claims abstract description 46
- 238000000034 method Methods 0.000 title claims abstract description 28
- 238000013527 convolutional neural network Methods 0.000 claims abstract description 70
- 238000012549 training Methods 0.000 claims description 38
- 238000011176 pooling Methods 0.000 claims description 15
- 230000003044 adaptive effect Effects 0.000 claims description 13
- 238000012795 verification Methods 0.000 claims description 8
- 238000012854 evaluation process Methods 0.000 claims description 7
- 230000004913 activation Effects 0.000 claims description 6
- 238000012360 testing method Methods 0.000 claims description 5
- 238000011156 evaluation Methods 0.000 claims description 4
- 230000008859 change Effects 0.000 claims description 3
- 239000010410 layer Substances 0.000 abstract description 53
- 238000002474 experimental method Methods 0.000 abstract description 6
- 210000000709 aorta Anatomy 0.000 abstract description 2
- 239000011229 interlayer Substances 0.000 abstract description 2
- 230000008569 process Effects 0.000 abstract description 2
- 230000006870 function Effects 0.000 description 10
- 238000013528 artificial neural network Methods 0.000 description 5
- 230000009286 beneficial effect Effects 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 238000013135 deep learning Methods 0.000 description 2
- 230000007547 defect Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 238000010200 validation analysis Methods 0.000 description 2
- 241000208340 Araliaceae Species 0.000 description 1
- 206010028980 Neoplasm Diseases 0.000 description 1
- 235000005035 Panax pseudoginseng ssp. pseudoginseng Nutrition 0.000 description 1
- 235000003140 Panax quinquefolius Nutrition 0.000 description 1
- 230000006399 behavior Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 235000008434 ginseng Nutrition 0.000 description 1
- 238000003384 imaging method Methods 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 230000003211 malignant effect Effects 0.000 description 1
- 230000002028 premature Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
Images
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
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- 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
-
- 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
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
Abstract
The invention discloses an aortic dissection CT image classification method based on a particle swarm optimization algorithm, and mainly relates to the field of intelligent optimization algorithms and image classification. Aiming at a specific aorta interlayer CT image classification task, the optimal classification convolutional neural network is automatically searched based on a particle swarm optimization algorithm. In the searching process, each particle position is set to represent a classified convolutional neural network, and a classified convolutional neural network with the best performance is finally searched through continuous iteration of a particle swarm optimization algorithm. The classification accuracy of 100% is obtained by performing an image classification experiment on the aortic dissection CT image data set, and higher classification accuracy can be obtained based on the aortic dissection CT image classification method provided by the invention compared with the classification accuracy of 98.97% obtained by the existing 110-layer classification convolutional neural network ResNet and the classification accuracy of 99.48% obtained by the 20-layer classification convolutional neural network SHEDA-CNN.
Description
Technical Field
The invention relates to the field of intelligent calculation and image classification, in particular to an aortic dissection CT image classification method based on a particle swarm optimization algorithm.
Background
In recent years, deep learning represented by a classified convolutional neural network has been widely applied to medical image data including aortic dissection CT images. Classifying the malignant epidermal tumor image by using a classification convolutional neural network; and classifying the fundus inspection image by using a classification convolutional neural network. However, the above-mentioned classification convolutional neural network is designed and constructed by computer experts according to experience, which is time-consuming and labor-consuming. Due to various problems of noise, unclear imaging and the like of medical images, a great challenge exists in manually designing a classification convolutional neural network for classifying the medical images.
In order to solve the limitation of manually designing and classifying the convolutional neural network, the method for automatically designing and classifying the convolutional neural network by utilizing an algorithm becomes a hot topic of deep learning field research. As an efficient swarm intelligence optimization algorithm, the particle swarm optimization algorithm mainly searches an optimal solution by simulating predation behaviors of natural birds, and has good global search capability and good convergence. Because the algorithm is simple and efficient, the method is widely applied to solving optimization problems in various real worlds. Therefore, considering the challenges of the aortic dissection CT image classification task and the effective optimization capability of the particle swarm optimization algorithm, it is urgently needed to provide a method for automatically searching the classification convolutional neural network for the aortic dissection CT image classification task by using the particle swarm optimization algorithm.
Disclosure of Invention
The invention aims to solve the above challenges in the prior art and provide an aortic dissection CT image classification method based on a particle swarm optimization algorithm. The method utilizes the particle swarm optimization algorithm to automatically search the classification convolutional neural network aiming at the aortic dissection CT image classification task, and is beneficial to improving the classification accuracy of the aortic dissection CT image classification task and avoiding consuming a large amount of labor energy and time. In addition, the invention further utilizes evolution state information of the particle swarm optimization algorithm to adjust the training times of the classification convolutional neural networks corresponding to the particles in different evolution stages, which is beneficial to improving the searching capability of the particle swarm optimization algorithm and finding the classification convolutional neural network with higher classification accuracy on the aortic dissection CT image classification task.
The purpose of the invention can be achieved by implementing the following technical scheme:
an aortic dissection CT image classification method based on a particle swarm optimization algorithm comprises the following steps:
s1, initializing a particle swarm population, wherein the position of each particle represents a classified convolutional neural network, and the total number of the particles is N;
s2, evaluating the performance of the classification convolutional neural network represented by the position of the particle according to the aortic dissection CT image classification task, taking the evaluation result as the adaptive value of the particle, and carrying out the following evaluation process:
s201, acquiring an aortic dissection CT image data set to be detected, and dividing the aortic dissection CT image data set into a training set, a verification set and a test set;
s202, training a classification convolutional neural network in a training set and verifying the classification convolutional neural network in a verification set, wherein the obtained classification accuracy is used as an adaptive value, and the training times of the classification convolutional neural network in the training set are E;
s3, comparing the adaptive values of the N particles, wherein the position of the particle with the largest adaptive value is recorded as a global optimal position Gbest;
s4, updating the speed and the position of the particles according to the search strategy of the particle swarm optimization algorithm, wherein the speed updating formula is as follows:
whereinRepresents the d-dimensional velocity of the ith particle in the g +1 th generation,representing the d-dimensional velocity of the ith particle in the g-th generation,represents the position of the ith particle in the d-dimension of the g-th generation, represents the velocity weight,representing the historical optimum position, Gbest, of the ith particle in dimension d d Representing the global optimum position in d-th dimension, c 1 And c 2 Is two update coefficients, r 1 And r 2 Is the d-dimension is [0,1 ]]The particle position value is updated by adding a new speed updated by the formula (1) on the basis of the original position as a new particle position;
s5, evaluating and comparing new positions after updating and iteration of the particle swarm optimization algorithm according to the evaluation process, and searching for a better classified convolutional neural network continuously, however, the traditional particle swarm optimization algorithm has the defects of premature convergence and easy falling into local optimization. The process is as follows:
s501, updating evolution state information S of the particles according to the change condition of the global optimal position, wherein if the global optimal position of the g generation is different from the global optimal position of the g-1 generation, an updating formula of S is as follows:
s g =s g-1 +1 (2)
wherein s is g-1 Represents evolution status information of the g-1 th generation, s g Denotes the g thEvolution state information of generations; if the global optimal position of the g-th generation is the same as the global optimal position of the g-1 th generation, the update formula of s is as follows:
s g =0; (3)
s502, updating the training times E of the classification convolutional neural network in the particle evaluation process according to the evolution state information S, and if the value of S is less than or equal to 3, keeping E unchanged; otherwise, the update formula for E is as follows:
E=E+5 (4)
and S6, when the particle swarm optimization algorithm iterates until the training frequency E of the classification convolutional neural network is a specified maximum value, continuously training the classification convolutional neural network corresponding to the global optimal position until convergence, taking the trained classification convolutional neural network as an aortic dissection CT image classification method, and otherwise, continuously executing the steps S3-S5 until the training frequency E meets the maximum value preset by the algorithm.
Further, the classification convolutional neural network is used for position representation of the particles.
Further, the basic structure of the classified convolutional neural network is as follows: the first layer is an input layer, the second layer is a convolution layer, the third layer is a pooling layer, the fourth layer to the seventeenth layer are convolution layers, the eighteenth layer is a pooling layer, the nineteenth layer is a full-connection layer, and the twentieth layer is an output layer; all layers, except the input layer and the output layer, contain an activation function; and in the convolutional layers from the fourth layer to the seventeenth layer, a residual error connection exists between every two convolutional layers.
Further, the convolution kernel size of the convolution layer of the classified convolutional neural network is 3 × 3 or 5 × 5; the type of the pooling layer is average pooling or maximum pooling; the type of the activation function is a ReLU function, a sigmoid function, or a tanh function.
Further, in step S2, the initial number E of training times of the classified convolutional neural network is 5.
Further, in step S3, the training times of the classification convolutional neural network are too small to evaluate the corresponding performance, and the initial value of the training times E of the classification convolutional neural network is 5.
Further, in step S5, when the global optimal position of the particle swarm optimization algorithm is not updated continuously, the particle swarm optimization algorithm falls into a dead state, so the present invention records the evolution state information S of the particle swarm optimization algorithm, and the value of the evolution state information S is an algebra in which the global optimal position is not updated continuously.
Further, in the step S5, in order to avoid the defects that the conventional particle swarm optimization algorithm is too early to converge and is easily trapped in local optimization, the present invention adjusts the training times of the classified convolutional neural networks corresponding to the particles at different evolution stages by using the evolution state information S of the particle swarm optimization algorithm, so as to improve the search capability of the particle swarm optimization algorithm.
Further, in step S6, the training times for classifying the convolutional neural network are too large, which requires more calculation time and calculation resources, and the maximum value of the training times E predetermined in the present invention is 15.
Compared with the prior art, the invention has the following advantages and effects:
1. the invention utilizes the particle swarm optimization algorithm to automatically design the classification convolutional neural network for classifying the aortic dissection CT images, thereby avoiding the time and energy consumed by manually designing the classification convolutional neural network for classifying the aortic dissection CT images.
2. The invention enables the classification convolutional neural network to represent the positions of the particles of the particle swarm optimization algorithm, and is beneficial to continuously searching a new classification convolutional neural network by utilizing the iteration step of the particle swarm optimization algorithm.
3. In the particle swarm optimization algorithm provided by the invention, the update information of the global optimal position of the particle swarm is recorded as the state information of the particle swarm optimization algorithm, so that the judgment of the evolution stage of the particle swarm is facilitated.
4. In the particle swarm optimization algorithm provided by the invention, the training times E of the classified convolutional neural network used in the adaptive value evaluation stage of the particle swarm are continuously adjusted by using the evolution state information of the particle swarm optimization algorithm, so that the particle swarm is helped to jump out of local optimization, the exploration capability of the particle swarm optimization algorithm is improved, and a better classified convolutional neural network is found.
Drawings
The accompanying drawings, which are included to provide a further understanding of the invention and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the invention and together with the description serve to explain the invention without limiting the invention. In the drawings:
FIG. 1 is a flow chart of an aortic dissection CT image classification method based on a particle swarm optimization algorithm, which is provided by the invention;
FIG. 2 is an exemplary aortic dissection CT image in an embodiment of the present invention;
FIG. 3 is another example aortic dissection CT image in an embodiment of the invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention are clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some embodiments of the present invention, but not all embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Examples
Referring to fig. 1, the embodiment discloses an aortic dissection CT image classification method based on a particle swarm optimization algorithm, which includes the following steps:
s1, initializing a particle swarm population, wherein the position of each particle represents a classified convolutional neural network, and the total number of the particles is N; the basic structure of the classified convolutional neural network is as follows: the first layer is an input layer, the second layer is a convolution layer, the third layer is a pooling layer, the fourth layer to the seventeenth layer are convolution layers, the eighteenth layer is a pooling layer, the nineteenth layer is a full-connection layer, and the twentieth layer is an output layer; all layers, except the input layer and the output layer, contain an activation function; in the convolutional layers from the fourth layer to the seventeenth layer, a residual error is connected between every two convolutional layers; the convolution kernel size of the convolution layer is 3 × 3 or 5 × 5; the type of the pooling layer is average pooling or maximum pooling; the type of the activation function is a ReLU function, a sigmoid function, or a tanh function.
S2, evaluating the performance of the classification convolution neural network represented by the particles according to the aorta interlayer CT image classification task, taking the evaluation result as the adaptive value of the particles, and the evaluation process is as follows:
s201, acquiring an aortic dissection CT image data set to be detected, and dividing the aortic dissection CT image data set into a training set, a verification set and a test set;
s202, training a classification convolutional neural network in a training set and verifying the classification convolutional neural network in a verification set, wherein the obtained classification accuracy is used as an adaptive value, and the training times of the classification convolutional neural network in the training set are E;
s3, comparing the adaptive values of the N particles, wherein the position of the particle with the largest adaptive value is recorded as a global optimal position Gbest;
s4, updating the speed and the position of the particles according to the search strategy of the particle swarm optimization algorithm, wherein the speed updating formula is as follows:
whereinRepresents the d-dimensional velocity of the ith particle in the g +1 th generation,representing the d-dimensional velocity of the ith particle in the g-th generation,represents the position of the ith particle in the d-dimension of the g-th generation, represents the velocity weight,representing the historical optimum position, Gbest, of the ith particle in dimension d d Representing the global optimum position in d-th dimension, c 1 And c 2 Is two update coefficients, r 1 And r 2 Is the d-dimension is [0,1 ]]The particle position value is updated by adding a new speed updated by the formula (1) on the basis of the original position as a new particle position;
s5, updating the evolution state information S of the particles according to the change condition of the global optimal position, wherein if the global optimal position of the g generation is different from the global optimal position of the g-1 generation, the updating formula of S is as follows:
s g =s g-1 +1 (2)
wherein s is g-1 Represents evolution status information of the g-1 th generation, s g Representing evolution state information of the g generation; if the global optimal position of the g-th generation is the same as the global optimal position of the g-1 th generation, the update formula of s is as follows: s g =0; (3)
S6, updating the training times E of the classification convolutional neural network in the particle evaluation process according to the evolution state information S, and if the value of S is less than or equal to 3, keeping E unchanged; otherwise, the update formula for E is as follows:
E=E+5 (4)
s7, when the training frequency E of the classification convolution neural network is a specified maximum value, continuing to train the classification convolution neural network corresponding to the global optimal position until convergence, and using the classification convolution neural network as an aortic dissection CT image classification method, otherwise continuing to execute the steps S3-S6 until the training frequency E meets the maximum value preset by the algorithm.
In order to verify that the aortic dissection CT image classification method based on the particle swarm optimization algorithm can find the optimal classification convolutional neural network aiming at a specific aortic dissection CT image classification task, a verification experiment is carried out as follows:
in a verification experiment, the classification convolution neural network of the particle swarm optimization algorithm provided by the invention is applied to classification application of an aortic dissection CT image data set. Example pictures are shown in fig. 2 and 3.
The aortic dissection CT image dataset comprises a training set, a validation set and a test set, comprising 3139 training samples, 348 validation samples and 386 test samples, respectively, of which there are two categories. The parameters set in this experiment are shown in table 1 below:
TABLE 1 table of parameters set in experiment
To illustrate the advantages of the method of the present application over the prior art by comparison, the present experiment simultaneously classified on aortic dissection CT image datasets using a ResNet classification convolutional neural network and a sheeda-CNN classification convolutional neural network, respectively. The final experimental results are shown in table 2:
TABLE 2 Experimental comparison results
Classification method | Number of layers | Amount of ginseng | Accuracy of classification |
ResNet | 110 | 10.78M | 98.97% |
SHEDA-CNN | 20 | 4.18M | 99.48% |
PSO-CNN | 20 | 1.91M | 100% |
As can be seen from table 2, classification using the 110-layer classification convolutional neural network ResNet in the aortic dissection CT image data set only obtained a classification accuracy of 98.97%; the use of a 20-layer SHEDA-CNN classification convolutional neural network resulted in a 99.48% classification accuracy. The classification method provided by the invention has 100% classification accuracy on the aortic dissection CT image data set, uses less parameters and is easier to train.
The experimental result shows that the aortic dissection CT image classification method based on the particle swarm optimization algorithm has higher classification accuracy and fewer parameters on a specific aortic dissection CT image classification task.
The above embodiments are preferred embodiments of the present invention, but the present invention is not limited to the above embodiments, and any other changes, modifications, substitutions, combinations, and simplifications which do not depart from the spirit and principle of the present invention should be construed as equivalents thereof, and all such changes, modifications, substitutions, combinations, and simplifications are intended to be included in the scope of the present invention.
Claims (6)
1. An aortic dissection CT image classification method based on a particle swarm optimization algorithm is characterized by comprising the following steps:
s1, initializing a particle swarm population, wherein the position of each particle represents a classified convolutional neural network, and the total number of the particles is N;
s2, evaluating the performance of the classification convolutional neural network represented by the position of the particle according to the aortic dissection CT image classification task, taking the evaluation result as the adaptive value of the particle, and carrying out the following evaluation process:
s201, acquiring an aortic dissection CT image data set to be detected, and dividing the aortic dissection CT image data set into a training set, a verification set and a test set;
s202, training a classification convolutional neural network in a training set and verifying the classification convolutional neural network in a verification set, wherein the obtained classification accuracy is used as an adaptive value, and the training times of the classification convolutional neural network in the training set are E;
s3, comparing the adaptive values of the N particles, wherein the position of the particle with the largest adaptive value is recorded as a global optimal position Gbest;
s4, updating the evolution state information S of the particles according to the change condition of the global optimal position, wherein if the global optimal position of the g generation is different from the global optimal position of the g-1 generation, the updating formula of S is as follows:
s g =s g-1 +1 (1)
wherein s is g-1 Represents evolution status information of the g-1 th generation, s g Representing evolution state information of the g generation; if the global optimal position of the g-th generation is the same as the global optimal position of the g-1 th generation, the update formula of s is as follows: s g =0; (2)
S5, updating the training times E of the classification convolutional neural network in the particle evaluation process according to the evolution state information S, and if the value of S is less than or equal to 3, keeping E unchanged; otherwise, the update formula for E is as follows:
E=E+5 (3)
s6, when the particle swarm optimization algorithm iterates until the training frequency E of the classification convolutional neural network is a specified maximum value, continuously training the classification convolutional neural network corresponding to the global optimal position until convergence, taking the trained classification convolutional neural network as the classification convolutional neural network for classifying the aortic dissection CT images, and otherwise, continuously executing the steps S3-S5 until the training frequency E meets the maximum value preset by the particle swarm optimization algorithm.
2. The particle swarm optimization algorithm-based aortic dissection CT image classification method according to claim 1, characterized in that the classification convolutional neural network is used for the position representation of the particles.
3. The aortic dissection CT image classification method based on particle swarm optimization algorithm according to claim 1, characterized in that the basic structure of the classification convolutional neural network is: the first layer is an input layer, the second layer is a convolution layer, the third layer is a pooling layer, the fourth layer to the seventeenth layer are convolution layers, the eighteenth layer is a pooling layer, the nineteenth layer is a full-connection layer, and the twentieth layer is an output layer; all layers, except the input layer and the output layer, contain an activation function; and in the convolutional layers from the fourth layer to the seventeenth layer, a residual error connection exists between every two convolutional layers.
4. The particle swarm optimization algorithm-based aortic dissection CT image classification method according to claim 3, characterized in that the convolution kernel size of the convolution layer of the classification convolutional neural network is 3 x 3 or 5 x 5; the type of the pooling layer is average pooling or maximum pooling; the type of the activation function is a ReLU function, a sigmoid function, or a tanh function.
5. The method for classifying an aortic dissection CT image based on particle swarm optimization according to claim 1, wherein in the step S2, the initial value of the training time E of the classification convolutional neural network is 5.
6. The method for classifying an aortic dissection CT image based on particle swarm optimization according to claim 1, wherein in step S6, the maximum value of the training times E is set in advance to 15.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210395179.XA CN114882270A (en) | 2022-04-15 | 2022-04-15 | Aortic dissection CT image classification method based on particle swarm optimization algorithm |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210395179.XA CN114882270A (en) | 2022-04-15 | 2022-04-15 | Aortic dissection CT image classification method based on particle swarm optimization algorithm |
Publications (1)
Publication Number | Publication Date |
---|---|
CN114882270A true CN114882270A (en) | 2022-08-09 |
Family
ID=82669495
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210395179.XA Pending CN114882270A (en) | 2022-04-15 | 2022-04-15 | Aortic dissection CT image classification method based on particle swarm optimization algorithm |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114882270A (en) |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108038507A (en) * | 2017-12-13 | 2018-05-15 | 太原理工大学 | Local receptor field extreme learning machine image classification method based on particle group optimizing |
CN109544511A (en) * | 2018-10-25 | 2019-03-29 | 广州大学 | Convolutional neural networks based on particle swarm algorithm optimization know method for distinguishing to Lung neoplasm |
CN113283573A (en) * | 2021-06-25 | 2021-08-20 | 江南大学 | Automatic search method for optimal structure of convolutional neural network |
CN113627606A (en) * | 2020-05-06 | 2021-11-09 | 江苏科技大学 | RBF neural network optimization method based on improved particle swarm optimization |
CN113989543A (en) * | 2021-09-28 | 2022-01-28 | 南京邮电大学 | COVID-19 medical image detection and classification method and device |
CN114118369A (en) * | 2021-11-19 | 2022-03-01 | 西南石油大学 | Image classification convolution neural network design method based on group intelligent optimization |
-
2022
- 2022-04-15 CN CN202210395179.XA patent/CN114882270A/en active Pending
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108038507A (en) * | 2017-12-13 | 2018-05-15 | 太原理工大学 | Local receptor field extreme learning machine image classification method based on particle group optimizing |
CN109544511A (en) * | 2018-10-25 | 2019-03-29 | 广州大学 | Convolutional neural networks based on particle swarm algorithm optimization know method for distinguishing to Lung neoplasm |
CN113627606A (en) * | 2020-05-06 | 2021-11-09 | 江苏科技大学 | RBF neural network optimization method based on improved particle swarm optimization |
CN113283573A (en) * | 2021-06-25 | 2021-08-20 | 江南大学 | Automatic search method for optimal structure of convolutional neural network |
CN113989543A (en) * | 2021-09-28 | 2022-01-28 | 南京邮电大学 | COVID-19 medical image detection and classification method and device |
CN114118369A (en) * | 2021-11-19 | 2022-03-01 | 西南石油大学 | Image classification convolution neural network design method based on group intelligent optimization |
Non-Patent Citations (4)
Title |
---|
刘杨 等: "粒子群优化算法惯量权重控制方法的研究", 《南京大学学报(自然科学)》, vol. 47, no. 4, 31 July 2011 (2011-07-31), pages 364 - 371 * |
彭建新 等: "全局信息引导的改进粒子群优化算法", 《小型微型计算机系统》, vol. 37, no. 7, 31 July 2016 (2016-07-31), pages 1518 - 1521 * |
彭建新 等: "自适应步长和发现概率的布谷鸟搜索算法", 《济南大学学报( 自然科学版)》, vol. 30, no. 5, 30 September 2016 (2016-09-30), pages 328 - 333 * |
王子佳 等: "基于概率评估差分进化的多峰值优化", 《智能系统学报》, vol. 17, no. 2, 31 March 2022 (2022-03-31), pages 427 - 439 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111666836B (en) | High-resolution remote sensing image target detection method of M-F-Y type light convolutional neural network | |
CN111882040B (en) | Convolutional neural network compression method based on channel number search | |
WO2022121289A1 (en) | Methods and systems for mining minority-class data samples for training neural network | |
CN109740734B (en) | Image classification method of convolutional neural network by optimizing spatial arrangement of neurons | |
CN107392919B (en) | Adaptive genetic algorithm-based gray threshold acquisition method and image segmentation method | |
CN108549718B (en) | A kind of general theme incorporation model joint training method | |
CN112633403A (en) | Graph neural network classification method and device based on small sample learning | |
CN111984817A (en) | Fine-grained image retrieval method based on self-attention mechanism weighting | |
CN114999635A (en) | circRNA-disease association relation prediction method based on graph convolution neural network and node2vec | |
CN112270405A (en) | Filter pruning method and system of convolution neural network model based on norm | |
CN110738362A (en) | method for constructing prediction model based on improved multivariate cosmic algorithm | |
Yanxia et al. | Threshold image target segmentation technology based on intelligent algorithms | |
CN106934398A (en) | Image de-noising method based on super-pixel cluster and rarefaction representation | |
CN114037014A (en) | Reference network clustering method based on graph self-encoder | |
CN111860265B (en) | Multi-detection-frame loss balanced road scene understanding algorithm based on sample loss | |
CN116452904B (en) | Image aesthetic quality determination method | |
CN114882270A (en) | Aortic dissection CT image classification method based on particle swarm optimization algorithm | |
CN109063749B (en) | Robust convolution kernel number adaptation method based on angular point radiation domain | |
CN111582442A (en) | Image identification method based on optimized deep neural network model | |
WO2023273934A1 (en) | Method for selecting hyper-parameter of model, and related apparatus | |
CN112434790B (en) | Self-interpretation method for distinguishing part of black box problem of convolutional neural network | |
CN115034262A (en) | Bearing fault diagnosis method based on fusion of convolutional neural network and optimization SVM | |
CN115170599A (en) | Method and device for vessel segmentation through link prediction of graph neural network | |
CN113744210A (en) | Heart segmentation method based on multi-scale attention U-net network | |
CN108932550B (en) | Method for classifying images based on fuzzy dense sparse dense algorithm |
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 |