CN109330613A - Human body Emotion identification method based on real-time brain electricity - Google Patents
Human body Emotion identification method based on real-time brain electricity Download PDFInfo
- Publication number
- CN109330613A CN109330613A CN201811259812.2A CN201811259812A CN109330613A CN 109330613 A CN109330613 A CN 109330613A CN 201811259812 A CN201811259812 A CN 201811259812A CN 109330613 A CN109330613 A CN 109330613A
- Authority
- CN
- China
- Prior art keywords
- sample
- eeg signals
- emotion identification
- classification
- real
- 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
Classifications
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/16—Devices for psychotechnics; Testing reaction times ; Devices for evaluating the psychological state
- A61B5/165—Evaluating the state of mind, e.g. depression, anxiety
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/24—Detecting, measuring or recording bioelectric or biomagnetic signals of the body or parts thereof
- A61B5/316—Modalities, i.e. specific diagnostic methods
- A61B5/369—Electroencephalography [EEG]
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7203—Signal processing specially adapted for physiological signals or for diagnostic purposes for noise prevention, reduction or removal
-
- A—HUMAN NECESSITIES
- A61—MEDICAL OR VETERINARY SCIENCE; HYGIENE
- A61B—DIAGNOSIS; SURGERY; IDENTIFICATION
- A61B5/00—Measuring for diagnostic purposes; Identification of persons
- A61B5/72—Signal processing specially adapted for physiological signals or for diagnostic purposes
- A61B5/7235—Details of waveform analysis
- A61B5/7264—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems
- A61B5/7267—Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems involving training the classification device
Landscapes
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Engineering & Computer Science (AREA)
- Psychiatry (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Heart & Thoracic Surgery (AREA)
- Medical Informatics (AREA)
- Molecular Biology (AREA)
- Surgery (AREA)
- Animal Behavior & Ethology (AREA)
- Pathology (AREA)
- Public Health (AREA)
- Veterinary Medicine (AREA)
- Signal Processing (AREA)
- Psychology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physiology (AREA)
- Educational Technology (AREA)
- Developmental Disabilities (AREA)
- Child & Adolescent Psychology (AREA)
- Hospice & Palliative Care (AREA)
- Social Psychology (AREA)
- Evolutionary Computation (AREA)
- Fuzzy Systems (AREA)
- Mathematical Physics (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
Abstract
The present invention discloses a kind of human body Emotion identification method based on real-time brain electricity, includes the following steps: 1) eeg signal acquisition: including using multichannel brain electric acquisition equipment acquisition subject EEG signals;2) EEG signals pre-process: comparison step 1 gained EEG signals are pre-processed, and to reduce tail interference, improve final classification discrimination;3) EEG signals sample characteristics are extracted;4) to eeg signal classification;5) Emotion identification is carried out.The invention reside in providing, a kind of eeg signal acquisition is simple, and intelligence degree is high, can accurately and effectively identify the human body Emotion identification method based on real-time brain electricity of EEG signals comprehensively.
Description
Technical field
The present invention relates to the EEG signals tagsort technical fields more particularly to one kind in living things feature recognition field to be based on
The human body Emotion identification method of real-time brain electricity.
Background technique
Mood is able to reflect the cognition and attitude of a people, can influence the psychology and behavior of people, be people's daily life
In important component part.It is auxiliary it is desirable to there is more humanized computer with the fast development that human-computer interaction is applied
Help others complete task, this requires computer have certain Emotion identification ability.In human-computer interaction process, if
Computer can rapidly and accurately identify emotional state locating for people, then it can adjust its work according to the emotional state of people
Content and mode improve the experience of human-computer interaction, so that human-computer interaction process is more friendly and natural.
In recent years, with the application of EEG signals equipment, the Emotion identification research based on brain electricity has become human-computer interaction
Using with a highly important research contents in artificial intelligence field.A kind of physiology of the EEG signals as central nervous system
Signal by many studies demonstrate that it has biggish correlation with mood, there is stronger mood to characterize ability, Ke Yizuo
Emotion identification is carried out for a kind of effective means.Currently, commonly the method for acquisition EEG signals is to place electricity on scalp
Pole records the potential change on scalp by electrode.Due to there is one layer of cuticula on scalp, and cuticula is non-conductive, so logical
It often cannot directly collect EEG signals.Therefore gluey conductive paste is smeared between electrode and scalp, reduce electrode and scalp
Between resistance, to collect EEG signals.The placement of electrode is covered with according to the distribution of electrodes figure of international 10-20 system
Entire skull.Although this method can collect more stable EEG signals, but have a disadvantage in that acquisition will every time
Conductive paste is smeared to subject, the workload of this process is very huge, and preparation process is also very complicated, and conductive paste and scalp
Contact can allow people to generate sense of discomfort for a long time.In addition, the EEG signals in all electrodes may have the redundancy unrelated with mood
Information, if all of the complexity that not only will increase algorithm, and interference can be generated to Emotion identification, reduce Emotion identification
Precision.Therefore, crucial brain area relevant to mood is found, the cost and complexity for reducing brain wave acquisition become most important.
Chinese Patent Application No. are as follows: 201710372387.7, the applying date is: on May 24th, 2017, publication date was: 2017
Years 15 days 09 month, patent name are as follows: EEG signals Feature Recognition System and method, a kind of EEG signals feature of the disclosure of the invention
Identifying system and method acquire different people under different emotional states using four conductive electrodes for being set to temporal lobe area above ear
Original EEG signals and form sample set;Then by pretreatment and feature extraction, it is special that brain electricity is obtained from sample set
Levy data;Training sample is obtained after being finally smoothed to brain electrical characteristic data, for being trained to support vector machines,
To obtain Emotion identification classifier.The present invention remains under the premise of the acquisition cost and complexity of brain electricity is greatly lowered
Higher Emotion identification accuracy rate is enough kept, and provides feasible foundation to carry out Emotion identification using wearable device.
Above patent document discloses a kind of EEG signals Feature Recognition System and method, but the system identification method is adopted
It is complicated to collect EEG signals mode, it is at high cost, EEG signals can not be effectively accurately identified comprehensively, and being unable to satisfy contemporary work life needs
It asks.
Summary of the invention
In view of this, simple the invention reside in a kind of eeg signal acquisition is provided, intelligence degree is high, can be comprehensively accurate
The human body Emotion identification method based on real-time brain electricity of effective identification EEG signals.
In order to achieve the object of the present invention, following technical scheme can be taken:
A kind of human body Emotion identification method based on real-time brain electricity, includes the following steps:
Step 1) eeg signal acquisition: including using multichannel brain electric acquisition equipment acquisition subject EEG signals;
The pretreatment of step 2) EEG signals: comparison step 1 gained EEG signals are pre-processed, to reduce tail interference,
Improve final classification discrimination;
Step 3) extracts EEG signals sample characteristics;
Step 4) is to eeg signal classification;
Step 5) carries out Emotion identification.
It includes by Sample Entropy algorithm to EEG signals sample characteristics that the step 3), which extracts EEG signals sample characteristics,
It extracts;The Sample Entropy algorithm is obtained particular by following algorithmic formula:
If initial data is the time series that length is N, indicate are as follows: { u (i): 1≤i≤N };
2) the vector X (1), X (2) ..., X (N-m+1) of one group of m-dimensional space are constructed, wherein X (i)={ u (i), u (i+
1),...,u(i+m)}.;
2) the distance between definition vector X (i) and X (j) d [X (i), X (j)] are that difference is maximum in two vector corresponding elements
One, it may be assumed that
3) for each { i:1≤i≤N-m+1 }, in the case of allowable deviation is r, statistics
The number of d [X (i), X (j)] < r, is calculated as Nm(i), and this number and the ratio apart from sum are calculated, are counted as:
4) it averages to all i and is counted as φm(r), i.e.,
5) dimension m is increased by 1, become m+1 repeat it is above-mentioned 1)~4) process obtainsφm+1(r);
6) the theoretically Sample Entropy SampEn (N, m, r) of this sequence are as follows:
N can not take ∞ in practice, when N takes finite value, estimation:
SampEn (N, m, r)=- 1n [φm+1(r)/φm(r)]
The value and parameter N of SampEn (N, m, r), the selection of m, r are related.Different insertion dimension m and similar tolerance r is corresponding
Sample entropy it is also different;The r takes 0.1~0.25 times of initial data standard deviation, SampEn when m=1 or m=2 (N, m,
R) value is best to the dependence of sequence length N, calculates resulting Sample Entropy at this time with relatively reasonable statistical property.
The step 3) further includes apart from battle array Sample Entropy algorithm by two-value to brain electricity to the extraction of EEG signals sample characteristics
Sample of signal feature extraction;The two-value is obtained apart from battle array Sample Entropy algorithm particular by following algorithmic formula:
Step 1: first calculating N × N two-value Distance matrix D=[d to N point sequenceij]N×N。
Second step, using the element in matrix D, according to the incremental sequence of row, every two row (as m=2) or every three row (when
When m=3) matrix element carry out AND operation by the combination of oblique line directions, the result of the oblique line "AND" of every a line it is cumulative after
Divided by N- (m+1), can be obtainedWith
Step 3: byWithCalculate separately φ2(r) and φ3(r)。
Step 4: calculating SampEn (N, m, r).
The step 4) includes support vector cassification method and genetic algorithm class method to eeg signal classification.
Step 5) the Emotion identification method includes unsupervised learning method of identification and supervised learning method of identification;It is described unsupervised
Study method of identification refers to does not specify classification information when carrying out pattern drill to sample to it, but from sample itself to characteristic close
Sample it is close, it is separate with the sample of different characteristics, to reach the aggregation of similar sample, the effect of foreign peoples's sample separation, finally
Implementation pattern classification;The supervised learning method of identification, which refers to, to be needed to be labeled the classification of sample, in the finger of classification information
Lower continuous correction model parameter is led, then obtained training pattern is used for the classification of test sample.
The beneficial effect of the technical scheme provided by the present invention is that: 1) present invention pass through by Sample Entropy algorithm pattern to brain electric
Sample of signal feature extraction, to more comprehensively accurate by eeg signal classification and changeable in mood identification, make by EEG signals come
The mood of identification people has obtained real utilization;2) present invention reduces costs, greatly improve work efficiency;3) present invention is logical
Unsupervised learning method of identification and supervised learning method of identification are crossed to carry out Emotion identification, the recognition methods is simpler and more direct, it is comprehensive, can
Row.
Detailed description of the invention
Fig. 1 is human body Emotion identification method flow diagram of the embodiment of the present invention based on real-time brain electricity;
Fig. 2 is that the two-value of human body Emotion identification method of the embodiment of the present invention based on real-time brain electricity is calculated apart from battle array Sample Entropy
Method matrix schematic diagram.
Specific embodiment
With reference to the accompanying drawing and the embodiment of the present invention is described in further detail invention.
Embodiment 1
Referring to Fig. 1, it is somebody's turn to do the human body Emotion identification method based on real-time brain electricity, is included the following steps:
Step 1) eeg signal acquisition S1: including using multichannel brain electric acquisition equipment acquisition subject EEG signals;
Step 2) EEG signals pre-process S2: comparison step 1 gained EEG signals are pre-processed, dry to reduce tail
It disturbs, improves final classification discrimination;
Step 3) extracts S3 to EEG signals sample characteristics;
Step 4) is to eeg signal classification S4;
Step 5) carries out Emotion identification S5.
It includes by Sample Entropy algorithm to EEG signals sample characteristics that the step 3), which extracts EEG signals sample characteristics,
It extracts;The Sample Entropy algorithm is obtained particular by following algorithmic formula:
If initial data is the time series that length is N, indicate are as follows: { u (i): 1≤i≤N };
3) the vector X (1), X (2) ..., X (N-m+1) of one group of m-dimensional space are constructed, wherein X (i)={ u (i), u (i+
1),...,u(i+m)}.;
2) the distance between definition vector X (i) and X (j) d [X (i), X (j)] are that difference is maximum in two vector corresponding elements
One, it may be assumed that
3) for each { i:1≤i≤N-m+1 }, in the case of allowable deviation is r, statistics
The number of d [X (i), X (j)] < r, is calculated as Nm(i), and this number and the ratio apart from sum are calculated, are counted as:
4) it averages to all i and is counted as φm(r), i.e.,
5) dimension m is increased by 1, become m+1 repeat it is above-mentioned 1)~4) process obtainsφm+1(r);
6) the theoretically Sample Entropy SampEn (N, m, r) of this sequence are as follows:
N can not take ∞ in practice, when N takes finite value, estimation:
SampEn (N, m, r)=- 1n [φm+1(r)/φm(r)]
The value and parameter N of SampEn (N, m, r), the selection of m, r are related.Different insertion dimension m and similar tolerance r is corresponding
Sample entropy it is also different;The r takes 0.1~0.25 times of initial data standard deviation, SampEn when m=1 or m=2 (N, m,
R) value is best to the dependence of sequence length N, calculates resulting Sample Entropy at this time with relatively reasonable statistical property.
It is actually to data length N, similar tolerance r that formula, which can be seen that the Sample Entropy, in analysis, and m point data section is mutual
The approximation of the m+1 point data section negative average natural logrithm of similar conditional probability CP mutually under phase similar situation.Sample Entropy exists
Improvement on algorithm relative to approximate entropy algorithm has the property that (1) Sample Entropy does not include the comparison of data section, because
This it be conditional probability negative average natural logrithm exact value, therefore the calculating of Sample Entropy does not depend on data length;(2) sample
Entropy has better consistency.I.e. if a time series has higher value than another time series, that is for other m and r
Value, it may have higher value;(3) Sample Entropy is insensitive for losing data.Even if loss of data up to 1/3, to calculated value shadow
Ring still very little.
Embodiment 2
The difference is that, in the present embodiment, the step 3) mentions EEG signals sample characteristics with above-described embodiment
Take further includes being extracted apart from battle array Sample Entropy algorithm to EEG signals sample characteristics by two-value;The two-value is apart from battle array Sample Entropy
Algorithm is obtained particular by following algorithmic formula:
Step 1: first calculating N × N two-value Distance matrix D=[d to N point sequenceij]N×N。
Second step, using the element in matrix D, according to the incremental sequence of row, every two row (as m=2) or every three row (when
When m=3) matrix element carry out AND operation by the combination of oblique line directions, the result of the oblique line "AND" of every a line it is cumulative after
Divided by N- (m+1), can be obtainedWith
It is as shown in Figure 2:
For example, we will judge whether d [X (2), X (4) < r] is true when m=2, it is equivalent to judge d [u (2), u (4) <
R] and d [u (3), u (5) < r] whether set up simultaneously.That is d24*d35Whether=1 is true, and here it is the mistakes that above-mentioned oblique line seeks "AND"
Journey.M=2 and m=3 can be placed in the same circulation and carry out in practical calculating process, and only seek "AND" in m=2 oblique line
As a result for the ground of " 1 ", it is necessary to carry out the oblique line of m=3 to seek "AND" process just now.
Step 3: byWithCalculate separately φ2(r) and φ3(r);
Step 4: calculating SampEn (N, m, r).
Embodiment 3
Referring to Fig. 1, the difference is that, in the present embodiment, the step 4) is to EEG signals point with above-described embodiment
Class includes support vector cassification method and genetic algorithm class method.
The support vector machines is a kind of machine that the one kind that developed on the basis of Statistical Learning Theory by Vapnik is new
Device learning method, it is based on structural risk minimization, ensure that Learning machine has good generalization ability, small in solution
The problems such as sample, high dimension, non-linear, local minimum point, is upper relatively good.Least square method supporting vector machine is by Suykens et al.
The novel support vector machines of the one kind put forward, it is that least square linear method is entered in support vector machines, by standard
Support vector machines in quadratic programming problem be transformed into linear equation solution, therefore simplify the complexity of calculating.
The genetic algorithm classifies to characteristic signal, and big measure feature letter is extracted from EEG signals to be processed
Number (including pseudo-characteristic signal and useful feature signal) then removes pseudo-characteristic signal by genetic algorithm, retains useful
Characteristic signal as driving signal.Genetic algorithm searching speed is slow, because it could not utilize the feedback letter of network in time
Breath.Therefore, the more training time is needed to obtain more accurately to solve.The potential ability of its parallel mechanism does not obtain sufficiently
It utilizes, this is also a research hotspot of current genetic algorithm.In addition, decision tree, probabilistic model, Logic Neural Networks, multilayer
The relevant classifications method such as perceptron is also widely used.
Embodiment 4
The difference is that, in the present embodiment, step 5) the Emotion identification method includes unsupervised with above-described embodiment
Learn method of identification and supervised learning method of identification;
The unsupervised learning method of identification refers to does not specify classification information to it when carrying out pattern drill to sample, but by
Sample itself is close to the sample of characteristic close, separate with the sample of different characteristics, to reach similar sample aggregation, foreign peoples's sample
The effect of this separation, final implementation pattern classification;
The supervised learning method of identification, which refers to, to be needed to be labeled the classification of sample, under the guidance of classification information not
Disconnected correction model parameter, then obtained training pattern is used for the classification of test sample.
Regularity of the unsupervised learning recognition methods in searching data set, this regularity are not necessarily to reach and draw
The purpose of divided data collection, that is to say, that be not necessarily intended to " classify ".This point is more extensive than the purposes of supervised learning method.Example
Such as analyze the principal component of a heap data, or the scope what feature analysis data set has can be attributed to unsupervised learning method.
The unsupervised learning recognition methods is segmented into two major classes, and one kind is based on the direct of PDF estimation
Method refers to that try to find the distribution parameter of all categories in feature space classifies again.It is another kind of to be known as based on similar between sample
Property measurement indirect clustering method, principle is to try to make different classes of core or initial nucleoid, then according to sample with
Sample is gathered into different classes of by the similarity measurement between these cores.
Commonly the example of the direct method based on Multilayer networks is histogram method.Such as we count one and are learned
Histogram method just often can be used in middle school student's height distribution in school, and height is divided into a section, and such as 1 meter to 1 meter 75 is calculated one section,
Then the number of students of height within this range is counted to each section, obtains histogram.If the male and female students number of this school
It is close, then we it finds that the histogram can embody there are two be distributed peak.The valley point in two peaks is so found, it will
Student is divided into two classes.
Supervised learning recognition methods must have training set and test sample.Rule is looked in training set, and to test
Sample uses this rule;And unsupervised learning does not have training set that this is said, only one group of data are found in the group data set
Rule.
Supervised learning recognition methods includes support vector cassification method and neural network classification method.
The purpose of supervised learning recognition methods is exactly to identify things, and the result of identification is shown to be added to data to be identified
Label.Therefore training sample set must be made of the sample of tape label.And what unsupervised learning recognition methods only to be analyzed
Data set itself, in advance without what label.If it find that certain aggregation is presented in data set, then it can be by natural aggregation point
Class, but for the purpose of not squared with by the classification designator preparatory with certain.
In supervised learning method of identification, sample set distribution, which is presented, is folded situation, and unsupervised learning recognition methods is not due to having
There is the guidance of classification sample, can not determine them is folded situation, can only be divided by the cluster situation of distribution.
Currently, brain-computer interface technology has emerged many effective applications, such as based on the wheelchair control of Mental imagery, base
In the cursor control of Hz ERP and the car steering under 3D virtual environment based on Mental imagery etc..But at present
Brain-machine interaction mode remains in some primary demands for realizing some disabled persons, and demand higher for disabled person without
Method is realized.Emotion identification based on brain electricity is on the basis of brain-computer interface, the certain more advanced demands of further satisfaction disabled person.Such as
For the disabled person of specified disease, its preference degree to things can be gone out by brain electricity analytical, for example, the selection of canteen, TV
The selection of program, selection of preferred music etc., brain electricity Emotion identification method, it is more advanced to can satisfy disabled person through the invention
Demand, reach preferably nursing as a result, improve disabled person quality of life.
The foregoing is only a preferred embodiment of the present invention, is not intended to limit the scope of the present invention.
Claims (5)
1. a kind of human body Emotion identification method based on real-time brain electricity, characterized by the following steps:
Step 1) eeg signal acquisition: including using multichannel brain electric acquisition equipment acquisition subject EEG signals;
The pretreatment of step 2) EEG signals: comparison step 1 gained EEG signals are pre-processed, and to reduce tail interference, are improved
Final classification discrimination;
Step 3) extracts EEG signals sample characteristics;
Step 4) is to eeg signal classification;
Step 5) carries out Emotion identification.
2. the human body Emotion identification method according to claim 1 based on real-time brain electricity, it is characterised in that: the step 3) is right
It includes being extracted by Sample Entropy algorithm to EEG signals sample characteristics that EEG signals sample characteristics, which extract,;The Sample Entropy algorithm
It is obtained particular by following algorithmic formula:
If initial data is the time series that length is N, indicate are as follows: { u (i): 1≤i≤N };
1) construct one group of m-dimensional space vector X (1), X (2) ..., X (N-m+1), wherein X (i)=u (i), u (i+1) ...,
u(i+m)}.;
2) the distance between definition vector X (i) and X (j) d [X (i), X (j)] are difference maximum one in two vector corresponding elements
It is a, it may be assumed that
3) for each { i:1≤i≤N-m+1 }, in the case of allowable deviation is r, statistics
The number of d [X (i), X (j)] < r, is calculated as Nm(i), and this number and the ratio apart from sum are calculated, are counted as:
4) it averages to all i and is counted as φm(r), i.e.,
5) dimension m is increased by 1, become m+1 repeat it is above-mentioned 1)~4) process obtainsφm+1(r);
6) the theoretically Sample Entropy SampEn (N, m, r) of this sequence are as follows:
N can not take ∞ in practice, when N takes finite value, estimation:
SampEn (N, m, r)=- 1n [φm+1(r)/φm(r)]
The value and parameter N of SampEn (N, m, r), the selection of m, r are related.Different insertion dimension m and the corresponding sample of similar tolerance r
This entropy is also different;The r takes 0.1~0.25 times of initial data standard deviation, SampEn (N, m, r) when m=1 or m=2
Value is best to the dependence of sequence length N, calculates resulting Sample Entropy at this time with relatively reasonable statistical property.
3. the human body Emotion identification method according to claim 1 based on real-time brain electricity, it is characterised in that: the step 3) is right
The extraction of EEG signals sample characteristics further includes being extracted apart from battle array Sample Entropy algorithm to EEG signals sample characteristics by two-value;It is described
Two-value obtained apart from battle array Sample Entropy algorithm particular by following algorithmic formula:
Step 1: first calculating N × N two-value Distance matrix D=[d to N point sequenceij]N×N。
Second step, using the element in matrix D, according to the incremental sequence of row, every two row (as m=2) or every three row (work as m=3
When) matrix element carry out AND operation by the combination of oblique line directions, the result of the oblique line "AND" of every a line it is cumulative after divided by
N- (m+1), can be obtainedWith
Step 3: byWithCalculate separately φ2(r) and φ3(r)。
Step 4: calculating SampEn (N, m, r).
4. the human body Emotion identification method according to claim 1 based on real-time brain electricity, it is characterised in that: the step 4) is right
Eeg signal classification includes support vector cassification method and genetic algorithm class method.
5. the human body Emotion identification method according to claim 1 based on real-time brain electricity, it is characterised in that: the step 5) feelings
Thread method of identification includes unsupervised learning method of identification and supervised learning method of identification;The unsupervised learning method of identification refers to sample
It carries out not specifying it classification information when pattern drill, but it is close to the sample of characteristic close from sample itself, with characteristic phase
Different sample is separate, thus reach similar sample aggregation, the effect of foreign peoples's sample separation, final implementation pattern classification;It is described to have
Supervised learning method of identification, which refers to, to be needed to be labeled the classification of sample, the continuous correction model ginseng under the guidance of classification information
It counts, then obtained training pattern is used for the classification of test sample.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811259812.2A CN109330613A (en) | 2018-10-26 | 2018-10-26 | Human body Emotion identification method based on real-time brain electricity |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201811259812.2A CN109330613A (en) | 2018-10-26 | 2018-10-26 | Human body Emotion identification method based on real-time brain electricity |
Publications (1)
Publication Number | Publication Date |
---|---|
CN109330613A true CN109330613A (en) | 2019-02-15 |
Family
ID=65312097
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201811259812.2A Pending CN109330613A (en) | 2018-10-26 | 2018-10-26 | Human body Emotion identification method based on real-time brain electricity |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN109330613A (en) |
Cited By (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110070105A (en) * | 2019-03-25 | 2019-07-30 | 中国科学院自动化研究所 | Brain electricity Emotion identification method, the system quickly screened based on meta learning example |
CN110110584A (en) * | 2019-03-14 | 2019-08-09 | 杭州电子科技大学 | A kind of emotional characteristics classification method based on CNN |
CN111466930A (en) * | 2020-04-16 | 2020-07-31 | 江西科技学院 | Audio-visual evoked emotion recognition method and system based on electroencephalogram signals |
CN112084935A (en) * | 2020-09-08 | 2020-12-15 | 南京邮电大学 | Emotion recognition method based on expansion of high-quality electroencephalogram sample |
CN113208633A (en) * | 2021-04-07 | 2021-08-06 | 北京脑陆科技有限公司 | Emotion recognition method and system based on EEG brain waves |
CN113869289A (en) * | 2021-12-02 | 2021-12-31 | 西北工业大学深圳研究院 | Multi-channel ship radiation noise feature extraction method based on entropy |
Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102915447A (en) * | 2012-09-20 | 2013-02-06 | 西安科技大学 | Binary tree-based SVM (support vector machine) classification method |
US20150196244A1 (en) * | 2012-08-07 | 2015-07-16 | Japan Science And Technology Agency | Emotion identification device, emotion identification method, and emotion identification program |
CN105930864A (en) * | 2016-04-15 | 2016-09-07 | 杭州电子科技大学 | EEG (electroencephalogram) signal feature classification method based on ABC-SVM |
CN106803095A (en) * | 2016-12-22 | 2017-06-06 | 辽宁师范大学 | Based on the brain electricity emotion identification method that assemblage characteristic is extracted |
CN106886792A (en) * | 2017-01-22 | 2017-06-23 | 北京工业大学 | A kind of brain electricity emotion identification method that Multiple Classifiers Combination Model Based is built based on layering |
CN107157477A (en) * | 2017-05-24 | 2017-09-15 | 上海交通大学 | EEG signals Feature Recognition System and method |
CN107411739A (en) * | 2017-05-31 | 2017-12-01 | 南京邮电大学 | EEG signals Emotion identification feature extracting method based on dual-tree complex wavelet |
CN107518894A (en) * | 2017-10-12 | 2017-12-29 | 公安部南昌警犬基地 | A kind of construction method and device of animal brain electricity disaggregated model |
CN108280414A (en) * | 2018-01-17 | 2018-07-13 | 重庆大学 | A kind of recognition methods of the Mental imagery EEG signals based on energy feature |
CN108363493A (en) * | 2018-03-20 | 2018-08-03 | 山东建筑大学 | User characteristics method for establishing model, system and storage medium based on brain-computer interface |
-
2018
- 2018-10-26 CN CN201811259812.2A patent/CN109330613A/en active Pending
Patent Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US20150196244A1 (en) * | 2012-08-07 | 2015-07-16 | Japan Science And Technology Agency | Emotion identification device, emotion identification method, and emotion identification program |
CN102915447A (en) * | 2012-09-20 | 2013-02-06 | 西安科技大学 | Binary tree-based SVM (support vector machine) classification method |
CN105930864A (en) * | 2016-04-15 | 2016-09-07 | 杭州电子科技大学 | EEG (electroencephalogram) signal feature classification method based on ABC-SVM |
CN106803095A (en) * | 2016-12-22 | 2017-06-06 | 辽宁师范大学 | Based on the brain electricity emotion identification method that assemblage characteristic is extracted |
CN106886792A (en) * | 2017-01-22 | 2017-06-23 | 北京工业大学 | A kind of brain electricity emotion identification method that Multiple Classifiers Combination Model Based is built based on layering |
CN107157477A (en) * | 2017-05-24 | 2017-09-15 | 上海交通大学 | EEG signals Feature Recognition System and method |
CN107411739A (en) * | 2017-05-31 | 2017-12-01 | 南京邮电大学 | EEG signals Emotion identification feature extracting method based on dual-tree complex wavelet |
CN107518894A (en) * | 2017-10-12 | 2017-12-29 | 公安部南昌警犬基地 | A kind of construction method and device of animal brain electricity disaggregated model |
CN108280414A (en) * | 2018-01-17 | 2018-07-13 | 重庆大学 | A kind of recognition methods of the Mental imagery EEG signals based on energy feature |
CN108363493A (en) * | 2018-03-20 | 2018-08-03 | 山东建筑大学 | User characteristics method for establishing model, system and storage medium based on brain-computer interface |
Non-Patent Citations (1)
Title |
---|
吕杰 等: "基于样本熵的生物医学信号复杂度分析方法及其优化算法", 《生物医学工程学进展》 * |
Cited By (7)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110110584A (en) * | 2019-03-14 | 2019-08-09 | 杭州电子科技大学 | A kind of emotional characteristics classification method based on CNN |
CN110070105A (en) * | 2019-03-25 | 2019-07-30 | 中国科学院自动化研究所 | Brain electricity Emotion identification method, the system quickly screened based on meta learning example |
CN111466930A (en) * | 2020-04-16 | 2020-07-31 | 江西科技学院 | Audio-visual evoked emotion recognition method and system based on electroencephalogram signals |
CN112084935A (en) * | 2020-09-08 | 2020-12-15 | 南京邮电大学 | Emotion recognition method based on expansion of high-quality electroencephalogram sample |
CN112084935B (en) * | 2020-09-08 | 2022-07-26 | 南京邮电大学 | Emotion recognition method based on expansion of high-quality electroencephalogram sample |
CN113208633A (en) * | 2021-04-07 | 2021-08-06 | 北京脑陆科技有限公司 | Emotion recognition method and system based on EEG brain waves |
CN113869289A (en) * | 2021-12-02 | 2021-12-31 | 西北工业大学深圳研究院 | Multi-channel ship radiation noise feature extraction method based on entropy |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109330613A (en) | Human body Emotion identification method based on real-time brain electricity | |
CN111134666B (en) | Emotion recognition method of multi-channel electroencephalogram data and electronic device | |
Wang et al. | Phase-locking value based graph convolutional neural networks for emotion recognition | |
Wang et al. | A prototype-based SPD matrix network for domain adaptation EEG emotion recognition | |
CN106886792B (en) | Electroencephalogram emotion recognition method for constructing multi-classifier fusion model based on layering mechanism | |
George et al. | Recognition of emotional states using EEG signals based on time-frequency analysis and SVM classifier. | |
CN103631941B (en) | Target image searching system based on brain electricity | |
CN112861604B (en) | Myoelectric action recognition and control method irrelevant to user | |
CN108491077A (en) | A kind of surface electromyogram signal gesture identification method for convolutional neural networks of being divided and ruled based on multithread | |
CN112244873A (en) | Electroencephalogram time-space feature learning and emotion classification method based on hybrid neural network | |
CN110367976B (en) | Brain wave signal detection method, related device and storage medium | |
Wan et al. | EEG fading data classification based on improved manifold learning with adaptive neighborhood selection | |
CN113010013A (en) | Wasserstein distance-based motor imagery electroencephalogram migration learning method | |
Li et al. | EEG signal classification method based on feature priority analysis and CNN | |
CN109359610A (en) | Construct method and system, the data characteristics classification method of CNN-GB model | |
Jinliang et al. | EEG emotion recognition based on granger causality and capsnet neural network | |
US20230101539A1 (en) | Physiological electric signal classification processing method and apparatus, computer device and storage medium | |
Moin et al. | Emotion recognition framework using multiple modalities for an effective human–computer interaction | |
CN114145744B (en) | Cross-equipment forehead electroencephalogram emotion recognition based method and system | |
Tang et al. | A hybrid SAE and CNN classifier for motor imagery EEG classification | |
CN114145745B (en) | Graph-based multitasking self-supervision emotion recognition method | |
Long et al. | Semi-supervised joint spatio-temporal feature selection for P300-based BCI speller | |
CN114652328A (en) | Bidirectional closed-loop brain-computer interaction electro-acupuncture parameter intelligent matching system for insomnia | |
CN110889335A (en) | Human skeleton double-person interaction behavior recognition method based on multi-channel space-time fusion network | |
CN107045624A (en) | Electroencephalogram signal preprocessing and classifying method based on maximum weighted cluster |
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 | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20190215 |