Brain effect connection network learning method based on automatic variational auto-encoder
An autoencoder and network connection technology, which is applied in the field of neural network deep learning theory and application research, and brain science research. It can solve the problems of algorithm performance dependence and difficulty in accurately learning the brain effect connection network of algorithms, so as to alleviate the problem of manual adjustment of a large number of parameters. , the effect of less hyperparameters and high accuracy
Pending Publication Date: 2022-03-22
BEIJING UNIV OF TECH
View PDF0 Cites 0 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0037] The Sim3 simulation data set was selected from the Smith simulation data set, and the fMRI data of 15 brain regions were used as input, and the automatic variational autoencoder was used to learn the effect connection network of brain regions. The basic structure of the method is as figure 1 As shown, its specific implementation steps are as follows:
[0038]Step (1): Parameter setting: including the number of brain regions n=15, the brain effect connection network parameter matrix A is initialized through the Pearson correlation coefficient of the brain interval, the hyperparameter λ=0.5 of the network sparse loss function, the expected KL dispersion degree value V KL =1.5, proportional controller coefficient K P =0.005, integral controller coefficient K I = 0.01.
[0039] Step (2): Use the encoder to learn latent variables from fMRI data, the specific steps are as follows:
[0040] Step (2.1): Use structural equation modeling to encode the brain-effect connectiv...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention discloses a brain effect connection network learning method based on an automatic variational auto-encoder, and belongs to the field of deep learning algorithms. Firstly, parameter initialization is carried out on the model, then a latent variable is learned from fMRI data of each brain region by using a coding network of an automatic variational auto-encoder, and generated fMRI data is obtained from the latent variable through a decoding network. And finally, when the generated fMRI data is highly similar to the real fMRI data, the model can learn an optimal brain effect connection network in the iterative training process. According to the method, the parameters of the model are adaptively adjusted by using the variational auto-encoder fused with the proportional-integral controller, and the effect connection network of the human brain is automatically and accurately learned in the end-to-end training process. Therefore, the method has the advantages of few parameters, high accuracy, strong generalization ability and the like, and can effectively relieve the problem of difficulty in manual parameter adjustment in the existing brain effect connection network deep learning method.
Description
technical field [0001] The invention belongs to the field of brain science research, neural network deep learning theory and application research, and specifically relates to a brain effect connection network learning method based on an automatic variational autoencoder. Background technique [0002] The human brain connectome research attempts to establish a brain network group map that describes different living human brain functions and structures from multiple levels. The brain effect connection network is a graph model composed of nodes and directed edges. Nodes are usually defined as Areas, directed edges describe the causal effect of one brain area on the neural activity of another brain area. At present, using computational methods to learn brain effect connection networks from functional magnetic resonance imaging (fMRI) data of the human brain has become a frontier hotspot in this research. [0003] In recent years, with the continuous integration of information s...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More Patent Type & Authority Applications(China)
IPC IPC(8): G06N3/04G06N3/08
CPCG06N3/084G06N3/045
Inventor 冀俊忠邹爱笑
Owner BEIJING UNIV OF TECH