CN110765978B - Channel selection method based on fractal dimension - Google Patents
Channel selection method based on fractal dimension Download PDFInfo
- Publication number
- CN110765978B CN110765978B CN201911065029.7A CN201911065029A CN110765978B CN 110765978 B CN110765978 B CN 110765978B CN 201911065029 A CN201911065029 A CN 201911065029A CN 110765978 B CN110765978 B CN 110765978B
- Authority
- CN
- China
- Prior art keywords
- electroencephalogram
- brain
- fractal dimension
- source component
- time
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/12—Classification; Matching
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
- G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
- G06F3/011—Arrangements for interaction with the human body, e.g. for user immersion in virtual reality
- G06F3/015—Input arrangements based on nervous system activity detection, e.g. brain waves [EEG] detection, electromyograms [EMG] detection, electrodermal response detection
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Signal Processing (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Biomedical Technology (AREA)
- Dermatology (AREA)
- General Health & Medical Sciences (AREA)
- Neurology (AREA)
- Neurosurgery (AREA)
- Human Computer Interaction (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The invention relates to the field of brain-computer interface technology and information processing technology, in particular to a channel selection method based on fractal dimension. The method solves the problems of high computational complexity, dependence on early-stage feature engineering and dependence on the existing neurophysiological cognition in the prior art. The method adopted by the invention comprises the following steps: (1) using a DEAP public emotion data set as an electroencephalogram signal to be analyzed; (2) in the electroencephalogram source component extraction stage, source component extraction is carried out on the multichannel electroencephalogram signals through an independent component analysis method; (3) a brain power component quantitative analysis stage, converting the plurality of brain power components extracted in the step (2) into time-frequency images through Wigner distribution, and quantitatively evaluating the information content in the time-frequency images through fractal dimension; (4) and (4) in a brain power component back projection stage, sequencing the values of the brain power components obtained in the step (3) through fractal dimension calculation from large to small in sequence, and realizing back projection through an equivalent dipole analysis method.
Description
Technical Field
The invention relates to the field of brain-computer interface technology and information processing technology, in particular to a channel selection method based on fractal dimension.
Background
Brain-computer interface (BCI) technology is a system for realizing mutual control and communication between the human brain and a computer or other electronic devices by using bio-electric signals detected in the inside of the brain, the surface of the cortex, and the scalp, thereby allowing a user to effectively communicate without the involvement of peripheral nervous system and muscular tissue. Electroencephalogram signals are widely applied to brain-computer interface technology due to the characteristics of non-invasiveness, convenience in acquisition, relative low price, high time resolution and the like. Compared with emotion recognition methods based on facial expressions, body postures, voices and the like, emotion recognition based on electroencephalogram signals is gradually a research hotspot because the emotion recognition is not easy to be manually dominant and has more objectivity.
In most existing documents, researchers acquire electroencephalogram signals by using channels as many as possible so as to improve emotion recognition precision and obtain a relatively accurate monitoring effect. However, using more channel signals can have negative effects such as reducing user comfort, increasing computational load and cost in signal processing. In order to solve the problems, a channel selection technology is developed, and the technology is combined with an actual application scene to remove channels irrelevant to tasks and redundant to simplify a system and improve the calculation efficiency.
In emotion recognition application based on electroencephalogram signals, the existing channel selection strategies are mainly divided into three categories:
firstly, researchers select channels according to basic cognition on correlation of electroencephalogram signals and emotional states, for example: in the specification of the Estimation of the existence of using two front EEG channels, Wu et al take data collected by Fp1 and Fp2 channels in the International 10-20 system as the input of EEG signal analysis according to the correlation between emotion and the characteristic of asymmetry of EEG activity in the frontal lobe area of the brain. The method depends on the existing neurophysiological cognition, and no reliable experimental result support exists, so that the accuracy and the effectiveness need to be verified. Therefore, with the development of brain science, researchers have proposed the second and third types of methods for automatic channel selection.
The second method is to find the first N characteristics related to the recognition task in the electroencephalogram signal based on a characteristic selection algorithm and map the first N characteristics to corresponding channels so as to realize channel selection. The identification performance of the channel selection method completely depends on whether the feature selection is reasonable or not, and if the feature selection is not reasonable, the selected channel cannot achieve the expected effect in practical application.
The third method is to evaluate candidate channel subsets generated by search algorithms such as a genetic algorithm, a particle swarm algorithm and the like through a classification algorithm to realize channel selection. This method, although obtaining a subset of channels with better performance by performing a full search among all channels, is computationally complex.
Disclosure of Invention
In view of this, the invention provides a channel selection method based on fractal dimension, so as to overcome the problems of high computational complexity, dependence on early stage feature engineering and dependence on the existing neurophysiological cognition in the prior art.
In order to solve the problems in the prior art, the technical scheme of the invention is as follows: an electroencephalogram channel selection method based on fractal dimension comprises the following steps:
(1) using a DEAP public emotion data set as an electroencephalogram signal to be analyzed;
(2) in the electroencephalogram source component extraction stage, source component extraction is carried out on multi-channel electroencephalogram signals through an independent component analysis method;
(3) a brain power component quantitative analysis stage, converting the plurality of brain power components extracted in the step (2) into a time-frequency diagram through Wigner distribution, and quantitatively evaluating the information content in the time-frequency diagram through fractal dimension;
(4) back projection brain power composition: the back projection module of the brain electrical source component back projects the brain electrical source component with more information content after quantization to the corresponding channel according to the inverse localization theory of the brain electrical signal so as to realize the final selection of the channel.
Further, in the step (2), the extraction of the brain electrical source components is completed by an independent component analysis method, which is represented as follows:
the electroencephalogram signal can be expressed as X ═ X in a matrix form 1 (t),x 2 (t),x 3 (t),...,x n (t)} T N represents the number of channels for acquiring the electroencephalogram signals, T is the transposition representation of X, and X is represented by m independent electroencephalogram source component vectors S ═ S 1 (t),s 2 (t),s 3 (t),...,s m (t)} T Mixed by a matrix W:
X＝WS (1)
in the above-mentioned X ═ WS model, the source component vector and the mixing matrix are obtained by the electroencephalogram data matrix X, that is:
U＝W -1 X (2)
wherein, W -1 Called the unmixing matrix, is the inverse of the mixing matrix W, defining W m ＝(w 1m ,w 2m ,...,w nm ) T Is a matrix W -1 A column vector of (2), wherein w 1m The mth brain-electrical source component, w, representing the first channel 2m The mth brain electrical source component representing the second channel, and so on, w nm Representing the mth brain electrical source component of the nth channel.
Further, the quantitative analysis of the brain power components in the step (3) is completed by both the wigner distribution and the fractal dimension, and is represented as:
by equation (3), any brain power component S is represented using the time-frequency domain, S (t) and S * (t) are complex conjugate of each other, τ is the time difference variable, f is the frequency, j is the imaginary unit, j 2 The gray-scale time-frequency image is converted into a binary image;
TFI(pix(k))＞thrd→1，TFI(pix(k))≤thrd (4)
TFI represents a time-frequency diagram of any brain power component, pix (k) represents the kth pixel, and thrd represents a threshold; the time-frequency graph after the binarization processing is represented as F, and the fractal dimension calculation is shown as a formula (5);
D α is the fractal dimension value of F, N r (F) The number of boxes required to cover F is shown, and r is the side length of the box.
Further, the back projection of the brain power source component in the step (4) is realized by an equivalent dipole analysis method, which is expressed as follows:
the back projection of the brain electrical source component regards any brain electrical source component as a current dipole, and the information such as the position, the intensity and the like of the current dipole is determined by using the minimized residual error, as shown in a formula (6).
φ 2 ＝||Zj-U|| 2 (6)
Wherein j is the transition dipole moment, Z is the lead field matrix, U is the source component vector obtained by formula (2), and Φ is the information representation of the solved current dipole.
Compared with the prior art, the invention has the following advantages:
1. the method completely accords with the generation mechanism of the electroencephalogram signals, namely the electroencephalogram signals collected by any scalp electrode are generally the result of comprehensively reflecting a plurality of brain power components, and are not formed by a single electroencephalogram signal.
2. Carrying out quantitative evaluation through Wigner analysis and fractal dimension so as to screen out electroencephalogram source components closely related to the emotion recognition task;
3. the method uses the independent component analysis method for emotion recognition scenes, optimizes an electroencephalogram signal channel, improves the convenience of wearable equipment, and subsequently simplifies subsequent electroencephalogram signal processing.
Drawings
Fig. 1 is a schematic diagram of a channel selection method based on fractal dimension according to the present invention;
fig. 2 is a flow chart of a channel selection method based on fractal dimension according to the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not limit the invention.
The invention provides a channel selection method based on fractal dimension, which comprises three key steps of extracting brain power components, quantitatively analyzing the brain power components and back projecting the brain power components, as shown in figure 1, and comprises the following specific steps:
step one, a DEAP public emotion data set is used as an electroencephalogram signal to be analyzed;
step two, extracting brain power components: the electroencephalogram source components with the same quantity as the number of the channels of the original acquired electroencephalogram signal are obtained by an independent component analysis method.
Specifically, an electroencephalogram signal generation mechanism is taken as a theoretical basis, an electroencephalogram signal acquired by any scalp electrode is generally a result of comprehensively reflecting a plurality of brain power components, and is not formed by a single electroencephalogram signal, and electroencephalogram source components of each electroencephalogram signal acquisition channel are acquired by an independent component analysis method.
Step three, quantitatively analyzing brain power components: firstly, a time-frequency diagram of a plurality of brain power components obtained by an extraction module of brain power components is obtained according to time-frequency analysis of Wigner distribution. And quantizing the time-frequency graphs of the brain power components through the fractal dimension to obtain the information content of the brain power components in a specific application scene.
Step four, back-projecting brain power components: the back projection module of the brain electrical source components back projects the brain electrical source components with more information content after quantization to the corresponding channels according to the inverse localization theory of the brain electrical signals so as to realize the final selection of the channels.
As shown in fig. 2, the invention provides a channel selection method based on fractal dimension, which specifically comprises the following steps:
step one, a DEAP public emotion data set is used as an electroencephalogram signal to be analyzed;
step two, extracting a plurality of electroencephalogram source components by using an algorithm based on the maximum principle of information transmission: the input EEG signal to be selected is output and compared with the channel number of the original EEG signal by an independent component analysis methodThe same brain electrical source components. The acquired brain electrical signals can be expressed as X ═ X in a matrix form 1 (t),x 2 (t),x 3 (t),...,x n (t)} T N represents the number of channels for acquiring the electroencephalogram signal, and T is a transposed representation of X (for example, each experiment is performed on one experiment of one person, namely 32 channels, so that in the analysis process of one experiment, X is composed of 32X (T), and n is equal to 32). X is formed by m independent brain electrical source component vectors S ═ S 1 (t),s 2 (t),s 3 (t),...,s m (t)} T The electroencephalogram source component is formed by mixing a matrix W (for example, 32 electroencephalogram source components can be obtained after 32 channels of electroencephalogram signals are processed by an independent component analysis method, m is equal to 32, the mixing relation of the 32 channels of electroencephalogram signals and the 32 electroencephalogram source components is realized by a W matrix, and W is called as a mixing matrix):
X＝WS (1)
in the above model, a source component vector and a mixing matrix are obtained through an electroencephalogram data matrix X, that is:
U＝W -1 X (2)
the matrix U is an approximate estimate of m mutually independent brain electrical source components. Wherein, W -1 Called the unmixing matrix, is the inverse of the mixing matrix W, defining W m ＝(w 1m ,w 2m ,...,w nm ) T Is a matrix W -1 Of a column vector of, wherein w 1m The mth brain-electrical source component, w, representing the first channel 2m The mth brain electrical source component representing the second channel, and so on, w nm Representing the mth brain electrical source component of the nth channel.
The Infmax algorithm based on the maximum principle of information transmission extracts a plurality of brain power components, namely when mutual information of an input brain electrical signal and the plurality of output brain power components is larger, redundant information among the output brain electrical source components is smaller.
Step three, quantitative analysis of electroencephalogram source components: in order to remove electroencephalogram source components irrelevant to the emotion recognition task, a plurality of mutually independent source components are subjected to Wigner distribution transformation to obtain a time-frequency diagram of the corresponding electroencephalogram source components. And performing time-frequency analysis on a plurality of brain power components through the Wigner distribution, wherein any brain power component S is a continuous time signal.
(3)
By the formula (3), any brain power component is expressed by a time-frequency domain, s (t) and s * (t) are complex conjugate of each other, τ is the time difference variable, f is the frequency, j is the imaginary unit, j 2 And (4) outputting the signal in a time frequency diagram form so as to measure the information content of the brain electrical source component by a box counting method.
And (3) using the fractal dimension as a quantization index of information contained in the time-frequency diagram, thereby preferably selecting electroencephalogram source components closely related to the emotion recognition task, namely quantizing the electroencephalogram source components. The grayscale time-frequency map is converted to a binary image before applying the box-counting method.
TFI(pix(k))＞thrd→1，TFI(pix(k))≤thrd (4)
TFI represents the time-frequency diagram of any brain power component, pix (k) represents the kth pixel, and thrd represents the threshold. If the pixel value is less than or equal to the threshold, the pixel is set to 0, otherwise it is 1.
The time-frequency graph after the binarization processing is represented by F, and the fractal dimension calculation is shown as a formula (5).
D α Is the fractal dimension value of F, N r (F) The number of boxes required to cover F is shown, and r is the side length of the box.
And step four, mapping the electroencephalogram source components which are screened by the box counting method and are related to the emotion recognition task to corresponding electroencephalogram channels through a back projection technology. The method uses an equivalent dipole analysis method to regard any brain power source component as a current dipole, and uses the minimized residual error to determine the information such as the position and the intensity of the current dipole, as shown in a formula (6).
φ 2 ＝||Zv-U|| 2 (6)
v is transition dipole moment, Z is lead field matrix, U is source component vector obtained by formula (2), and phi is information representation of the solved current dipole.
In order to verify the rationality of the proposed channel selection method, the selected channel subset is used as the input of a 1-D convolutional neural network, and the four-classification of emotion is realized by automatically extracting features through deep learning.
The above description is only a preferred embodiment of the present invention, and is not intended to limit the scope of the present invention.
Claims (4)
1. A fractal dimension-based electroencephalogram channel selection method comprises the following steps:
(1) using a DEAP public emotion data set as an electroencephalogram signal to be analyzed;
(2) in the electroencephalogram source component extraction stage, source component extraction is carried out on multi-channel electroencephalogram signals through an independent component analysis method;
(3) a brain power component quantitative analysis stage, converting the plurality of brain power components extracted in the step (2) into a time-frequency diagram through Virger distribution, and quantitatively evaluating the information content in the time-frequency diagram through fractal dimension to realize quantification;
(4) back projection stage of brain power components:
and (4) sequencing the values obtained by the fractal dimension calculation of the brain electrical source components in the step (3) from large to small in sequence, and back-projecting the brain electrical source components with more information content after quantization to the corresponding channels according to the inverse localization theory of brain electrical signals so as to realize the final selection of the channels, wherein the back-projection of the brain electrical source components is realized by an equivalent dipole analysis method.
2. The fractal dimension-based electroencephalogram channel selection method as claimed in claim 1, wherein in the step (2), the electroencephalogram source component extraction part is completed by an independent component analysis method, and is represented as:
the electroencephalogram signal can be expressed as X ═ { X in a matrix form 1 (t),x 2 (t),x 3 (t),...,x n (t)} T N represents the number of channels for acquiring the electroencephalogram signals, T is the transposition representation of X, and X is represented by m independent electroencephalogram source component vectors S ═ S 1 (t),s 2 (t),s 3 (t),...,s m (t)} T Mixed by a matrix W:
X＝WS (1)
in the above-mentioned X ═ WS model, the source component vector and the mixing matrix are obtained by the electroencephalogram data matrix X, that is:
U＝W -1 X (2)
wherein, W -1 Called the unmixing matrix, is the inverse of the mixing matrix W, defining W m ＝(w 1m ,w 2m ,...,w nm ) T Is a matrix W -1 A column vector of (2), wherein w 1m The mth brain-electrical source component, w, representing the first channel 2m The mth brain electrical source component representing the second channel, and so on, w nm Representing the mth brain electrical source component of the nth channel.
3. The fractal dimension-based electroencephalogram channel selection method according to claim 1 or 2, wherein the quantitative analysis of brain power components in the step (3) is jointly completed by the Virgener distribution and the fractal dimension, and is represented as:
by equation (3), any brain power component S is represented using the time-frequency domain, S (t) and S * (t) are complex conjugate of each other, τ is the time difference variable, f is the frequency, j is the imaginary unit, j 2 And (4) outputting the gray-scale time-frequency image in a time-frequency image form, and converting the gray-scale time-frequency image into a binary image through a formula (4):
TFI(pix(k))＞thrd→1，TFI(pix(k))≤thrd→0 (4)
TFI represents a time-frequency diagram of any brain power component, pix (k) represents the kth pixel, and thrd represents a threshold; if TFI (pix (k)) is less than or equal to the threshold, then set the pixel to 0, otherwise to 1;
the time-frequency graph after the binarization processing is represented as F, and the fractal dimension calculation is shown as a formula (5):
D α is the fractal dimension value of F, N r (F) The number of boxes required to cover F is shown, and r is the side length of the box.
4. The fractal dimension-based electroencephalogram channel selection method according to claim 3, wherein the back projection of brain power components in the step (4) is realized by an equivalent dipole analysis method, which is expressed as follows:
the electroencephalogram source component back projection regards any electroencephalogram source component as a current dipole, and information such as the position, the intensity and the like of the current dipole is determined by using the minimized residual error, as shown in a formula (6):
φ 2 ＝||Zv-U|| 2 (6)
wherein v is transition dipole moment, Z is lead field matrix, U is source component vector obtained by formula (2), and Φ is information representation of the solved current dipole.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911065029.7A CN110765978B (en) | 2019-11-04 | 2019-11-04 | Channel selection method based on fractal dimension |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911065029.7A CN110765978B (en) | 2019-11-04 | 2019-11-04 | Channel selection method based on fractal dimension |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110765978A CN110765978A (en) | 2020-02-07 |
| CN110765978B true CN110765978B (en) | 2022-08-16 |
Family
ID=69335490
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911065029.7A Active CN110765978B (en) | 2019-11-04 | 2019-11-04 | Channel selection method based on fractal dimension |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110765978B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015054681A1 (en) * | 2013-10-11 | 2015-04-16 | Marcio Marc Abreu | Method and apparatus for biological evaluation |
| CN108937968A (en) * | 2018-06-04 | 2018-12-07 | 安徽大学 | The Conduction choice method of emotion EEG signals based on independent component analysis |
| WO2018235076A1 (en) * | 2017-06-21 | 2018-12-27 | Hadasit Medical Research Services And Development Ltd. | Method and system for predicting response to pharmacological treatment from eeg |
| CN110353673A (en) * | 2019-07-16 | 2019-10-22 | 西安邮电大学 | A kind of brain electric channel selection method based on standard mutual information |
-
2019
- 2019-11-04 CN CN201911065029.7A patent/CN110765978B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015054681A1 (en) * | 2013-10-11 | 2015-04-16 | Marcio Marc Abreu | Method and apparatus for biological evaluation |
| WO2018235076A1 (en) * | 2017-06-21 | 2018-12-27 | Hadasit Medical Research Services And Development Ltd. | Method and system for predicting response to pharmacological treatment from eeg |
| CN108937968A (en) * | 2018-06-04 | 2018-12-07 | 安徽大学 | The Conduction choice method of emotion EEG signals based on independent component analysis |
| CN110353673A (en) * | 2019-07-16 | 2019-10-22 | 西安邮电大学 | A kind of brain electric channel selection method based on standard mutual information |
Non-Patent Citations (4)
| Title |
|---|
| " EEG Channel Selection using Fractal Dimension and Artificial Bee Colony Algorithm";Garro B A;《2018 IEEE Symposium Series on Computational Intelligence (SSCI)》;20181231;全文 * |
| "A novel approach to emotion recognition using local subset feature selection and modified Dempster-Shafer theory";Zangeneh Soroush;《Behavioral and Brain Functions》;20181031;全文 * |
| "基于独立成分分析的优选N200和P300特征通道算法";李文轩;《计算机工程与科学》;20170930;全文 * |
| 小波变换结合盲源分离的EEG情感识别;沈成业等;《计算机工程与应用》;20170719(第10期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110765978A (en) | 2020-02-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109165556B (en) | Identity recognition method based on GRNN | |
| CN109199414B (en) | Audio-visual evoked emotion recognition method and system based on electroencephalogram signals | |
| Meyers | The neural decoding toolbox | |
| Ilyas et al. | A survey of analysis and classification of EEG signals for brain-computer interfaces | |
| CN112244873A (en) | Electroencephalogram time-space feature learning and emotion classification method based on hybrid neural network | |
| CN112800998B (en) | Multi-mode emotion recognition method and system integrating attention mechanism and DMCCA | |
| Tung et al. | Entropy-assisted multi-modal emotion recognition framework based on physiological signals | |
| CN113157094B (en) | Electroencephalogram emotion recognition method combining feature migration and graph semi-supervised label propagation | |
| CN110367976B (en) | Brain wave signal detection method, related device and storage medium | |
| Yang et al. | Blind image quality assessment based on multi-scale KLT | |
| Jiang et al. | Analytical comparison of two emotion classification models based on convolutional neural networks | |
| Arivazhagan et al. | Skin disease classification by extracting independent components | |
| CN116250849A (en) | Electroencephalogram signal identification method based on information separator and regional convolution network | |
| CN115238796A (en) | Motor imagery electroencephalogram signal classification method based on parallel DAMSCN-LSTM | |
| Tosato et al. | EEG synthetic data generation using probabilistic diffusion models | |
| Hurri | Independent component analysis of image data | |
| CN114242202A (en) | Cerebral apoplexy hand rehabilitation training method, device and system | |
| CN113974627A (en) | Emotion recognition method based on brain-computer generated confrontation | |
| CN113627391A (en) | Cross-mode electroencephalogram signal identification method considering individual difference | |
| CN110765978B (en) | Channel selection method based on fractal dimension | |
| CN116531001A (en) | Method and device for generating multi-listener electroencephalogram signals and identifying emotion of cross-listener | |
| Zhao et al. | Automatic windowing for mri with convolutional neural network | |
| CN113974625B (en) | Emotion recognition method based on brain-computer cross-modal migration | |
| CN113842151B (en) | Cross-test EEG cognitive state detection method based on efficient multi-source capsule network | |
| CN114818822A (en) | Electroencephalogram migration emotion recognition method combining semi-supervised regression and icon label propagation |
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 |