CN110826527A - Electroencephalogram negative emotion recognition method and system based on aggressive behavior prediction - Google Patents
Electroencephalogram negative emotion recognition method and system based on aggressive behavior prediction Download PDFInfo
- Publication number
- CN110826527A CN110826527A CN201911141842.8A CN201911141842A CN110826527A CN 110826527 A CN110826527 A CN 110826527A CN 201911141842 A CN201911141842 A CN 201911141842A CN 110826527 A CN110826527 A CN 110826527A
- Authority
- CN
- China
- Prior art keywords
- feature vector
- emotion
- negative
- negative emotion
- initial
- 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.)
- Withdrawn
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
- G06F2218/02—Preprocessing
- G06F2218/04—Denoising
-
- 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
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
- G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
- G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
-
- 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
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Physics & Mathematics (AREA)
- Public Health (AREA)
- Data Mining & Analysis (AREA)
- Biomedical Technology (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Artificial Intelligence (AREA)
- Primary Health Care (AREA)
- Biophysics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Epidemiology (AREA)
- Databases & Information Systems (AREA)
- Pathology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Signal Processing (AREA)
- Computational Linguistics (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Measurement And Recording Of Electrical Phenomena And Electrical Characteristics Of The Living Body (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The invention discloses an electroencephalogram negative emotion recognition method and system based on aggressive behavior prediction, which comprises the steps of processing and feature extraction on obtained sample data to obtain an initial emotion sample feature vector; the sample data comprises electroencephalogram signals generated by respectively stimulating a healthy subject by adopting a plurality of negative emotion stimulation modes and a negative emotion stimulation mode corresponding to each electroencephalogram signal; training a deep neural network based on the initial emotion sample feature vector, and determining the intermediate layer feature of the trained deep neural network model as an optimized sample feature vector; training a classifier according to the optimized sample feature vector and the initial emotion sample feature vector, and determining a negative emotion recognition classification model; and processing the electroencephalogram signal of the testee, and identifying the negative emotion of the testee according to the electroencephalogram signal and the negative emotion identification classification model after the electroencephalogram signal is processed by the testee. The method can improve the emotion electroencephalogram classification recognition rate, and further avoid and prevent aggressive behaviors.
Description
Technical Field
The invention relates to the technical field of electroencephalogram negative emotion recognition, in particular to an electroencephalogram negative emotion recognition method and system based on aggressive behavior prediction.
Background
The college students, as a group with young impulsivity, are relatively easily affected by negative emotions to make aggressive behaviors such as attack, opposition, destruction and the like. The aggressive behavior is closely related to negative emotions. The strong negative emotions at higher arousals are highly disruptive and aggressive behavior can be considered as an outward manifestation of this emotional state.
Deep learning is of particular interest as an important branch of the field of machine learning. Deep learning, which may also be referred to as deep structure learning or hierarchical learning, theoretically combines features at lower levels and transforms them into a more abstract feature space at higher levels, thereby making classification and identification easier. The method is essentially characterized in that on the basis of constructing a multi-hidden-layer machine learning model, mass data are used for training to achieve the purpose of learning more useful characteristics, and the performance of a classification recognition system is conveniently improved. Therefore, the deep learning model is applied to feature extraction and classification recognition, so that the emotion electroencephalogram classification recognition rate can be improved, and the deep learning model is worthy of further discussion and research.
Disclosure of Invention
The invention aims to provide an electroencephalogram negative emotion recognition method and system based on aggressive behavior prediction, which can improve emotion electroencephalogram classification recognition rate and further avoid and prevent aggressive behaviors.
In order to achieve the purpose, the invention provides the following scheme:
an electroencephalogram negative emotion recognition method based on aggressive behavior prediction comprises the following steps:
acquiring sample data; the sample data comprises a plurality of data sets; each data set comprises electroencephalogram signals generated by respectively stimulating healthy subjects of the same gender and the same age by adopting a plurality of negative emotion stimulation modes, and the negative emotion stimulation mode corresponding to each electroencephalogram signal, and the ages or the genders of the healthy subjects corresponding to different data sets are different;
preprocessing the electroencephalogram signals in the sample data;
performing feature extraction on the preprocessed sample data to obtain an initial emotion sample feature vector;
training a deep neural network based on the initial emotion sample feature vector and a negative emotion stimulation mode corresponding to the initial emotion sample feature to obtain a trained deep neural network model, and determining the middle layer feature of the trained deep neural network model as an optimized sample feature vector;
training a classifier according to the optimized sample feature vector and the initial emotion sample feature vector, and determining a negative emotion recognition classification model;
acquiring and processing an electroencephalogram signal of a subject;
and identifying the negative emotion of the testee according to the electroencephalogram signals processed by the testee and the negative emotion identification classification model.
Optionally, the preprocessing the electroencephalogram signal in the sample data specifically includes:
and denoising the electroencephalogram signal by adopting an independent component analysis algorithm and the waveform characteristics of the original data to obtain preprocessed sample data.
Optionally, the performing feature extraction on the preprocessed sample data to obtain an initial emotion sample feature vector specifically includes:
for each data group, sequentially extracting preprocessed electroencephalogram signals in each negative emotional stimulation mode in the data group according to the time point of stimulation of the first negative emotional stimulation mode, and forming an epoch set;
locking the power spectrum of the epoch set in the same negative emotional stimulation mode, and calculating according to frequency to obtain a two-dimensional time-frequency spectrum of event-related disturbance;
performing characteristic analysis on the two-dimensional time-frequency spectrum, and judging by combining a one-factor variance analysis algorithm to obtain a frequency range and brain area distribution;
and extracting PSD values according to the frequency range and the brain area distribution, and forming high-dimensional feature vectors by the PSD values to be used as initial emotion sample feature vectors.
Optionally, the nodes of the network input layer and the last hidden layer of the trained deep neural network model are Gaussian nodes, the nodes of other hidden layers are Bernoulli nodes, and the node of the output layer is a Softmax node.
Optionally, the training a classifier according to the feature vector of the optimized sample and the feature vector of the initial emotion sample, and determining a negative emotion recognition classification model specifically includes:
training a classifier according to the optimized sample feature vector and the initial emotion sample feature vector, and determining a plurality of initial negative emotion recognition classification models;
and screening the initial negative emotion recognition classification model by using a cross validation algorithm and taking the minimum cross validation error rate as a standard to determine a final negative emotion recognition classification model.
An electroencephalogram negative emotion recognition system based on aggressive behavior prediction, comprising:
the sample data acquisition module is used for acquiring sample data; the sample data comprises a plurality of data sets; each data set comprises electroencephalogram signals generated by respectively stimulating healthy testees of the same gender and the same age by adopting a plurality of negative emotion stimulation modes, and the negative emotion stimulation mode corresponding to each electroencephalogram signal, and the ages or the genders of the healthy testees corresponding to different data sets are different;
the preprocessing module is used for preprocessing the electroencephalogram signals in the sample data;
the initial emotion sample feature vector acquisition module is used for extracting features of the preprocessed sample data to acquire an initial emotion sample feature vector;
the optimized sample feature vector determining module is used for training a deep neural network based on the initial emotion sample feature vector and a negative emotion stimulation mode corresponding to the initial emotion sample feature to obtain a trained deep neural network model, and determining the middle layer feature of the trained deep neural network model as an optimized sample feature vector;
the negative emotion recognition classification model determining module is used for training a classifier according to the optimized sample feature vector and the initial emotion sample feature vector and determining a negative emotion recognition classification model;
the electroencephalogram signal acquisition processing module is used for acquiring and processing an electroencephalogram signal of a subject;
and the negative emotion recognition module is used for recognizing the negative emotion of the testee according to the electroencephalogram signals processed by the testee and the negative emotion recognition classification model.
Optionally, the preprocessing module specifically includes:
and the preprocessing unit is used for denoising the electroencephalogram signal by adopting an independent component analysis algorithm and the waveform characteristics of the original data to obtain preprocessed sample data.
Optionally, the initial emotion sample feature vector obtaining module specifically includes:
the epoch set combination unit is used for sequentially extracting the preprocessed electroencephalogram signals in each negative emotional stimulation mode in each data group according to the stimulation time point of the first negative emotional stimulation mode for each data group, and forming an epoch set;
the two-dimensional time-frequency map determining unit is used for locking the power spectrum of the epoch set in the same negative emotional stimulation mode, and calculating according to frequency to obtain a two-dimensional time-frequency map of event-related disturbance;
the frequency range and brain area distribution determining unit is used for performing characteristic analysis on the two-dimensional time-frequency spectrum and judging to obtain a frequency range and brain area distribution by combining a one-factor variance analysis algorithm;
and the initial emotion sample feature vector acquisition unit is used for extracting a PSD value according to the frequency range and the brain area distribution, and forming a high-dimensional feature vector by using the PSD value as an initial emotion sample feature vector.
Optionally, the negative emotion recognition classification model determining module specifically includes:
the initial negative emotion recognition classification model determining unit is used for training a classifier according to the optimized sample feature vector and the initial emotion sample feature vector and determining a plurality of initial negative emotion recognition classification models;
and the negative emotion recognition classification model determining unit is used for screening the initial negative emotion recognition classification model by using a cross validation algorithm and taking the minimum cross validation error rate as a standard to determine a final negative emotion recognition classification model.
According to the specific embodiment provided by the invention, the invention discloses the following technical effects:
the invention relates to an electroencephalogram negative emotion recognition method and system based on aggressive behavior prediction, which comprises the steps of obtaining sample data; the sample data comprises a plurality of data sets; each data set comprises electroencephalogram signals generated by respectively stimulating healthy testees of the same gender and the same age by adopting a plurality of negative emotion stimulation modes, and the negative emotion stimulation mode corresponding to each electroencephalogram signal, and the ages or the genders of the healthy testees corresponding to different data sets are different; preprocessing the electroencephalogram signals in the sample data; performing feature extraction on the preprocessed sample data to obtain an initial emotion sample feature vector; training a deep neural network based on the initial emotion sample feature vector and a negative emotion stimulation mode corresponding to the initial emotion sample feature to obtain a trained deep neural network model, and determining the middle layer feature of the trained deep neural network model as an optimized sample feature vector; training a classifier according to the optimized sample feature vector and the initial emotion sample feature vector, and determining a negative emotion recognition classification model; acquiring and processing an electroencephalogram signal of a subject; and identifying the negative emotion of the testee according to the electroencephalogram signals processed by the testee and the negative emotion identification classification model. The method can improve the emotion brain electricity classification recognition rate, and further avoid and prevent the occurrence of aggressive behaviors.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art that other drawings can be obtained according to the drawings without inventive labor.
FIG. 1 is a flow chart of an electroencephalogram negative emotion recognition method based on aggressive behavior prediction according to an embodiment of the present invention;
FIG. 2 is a flow chart of a negative emotion recognition classification model determination according to an embodiment of the present invention;
FIG. 3 is a frame diagram of the real-time deep neural network-based emotional feature extraction of the present invention;
fig. 4 is a structural schematic diagram of an electroencephalogram negative emotion recognition system based on aggressive behavior prediction in the embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be described clearly and completely with reference to the accompanying drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the 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.
Aggressive behavior is closely related to negative emotions. The knowledge of a high arousal negative emotion can be used to some extent as a basis for predictive assessment of the development of an offensive behavioral crime. The invention provides an electroencephalogram negative emotion recognition method and system based on aggressive behavior prediction. The negative emotion with higher arousal degree is subjected to cognitive analysis and classification prediction, the psychological state of the testee is judged and evaluated, and the corresponding emotion regulation and psychological treatment measures are combined for the testee which is predicted to generate high negative emotion easily, so that the occurrence of aggressive behaviors is avoided and prevented.
In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.
Fig. 1 is a flow diagram of an electroencephalogram negative emotion recognition method based on aggressive behavior prediction according to an embodiment of the present invention, fig. 2 is a flow diagram of a negative emotion recognition classification model determination according to an embodiment of the present invention, referring to fig. 1 and 2, and the method for recognizing electroencephalogram negative emotion based on aggressive behavior prediction according to the present invention includes the following steps.
Step 101: acquiring sample data; the sample data comprises a plurality of data sets; each data set comprises electroencephalogram signals generated by respectively stimulating healthy testees of the same gender and the same age by adopting a plurality of negative emotion stimulation modes, and the negative emotion stimulation modes corresponding to the electroencephalogram signals, and the data sets are different in age or gender of the corresponding healthy testees.
Step 101 specifically comprises: the age and the sex of the healthy subjects are determined, and then grouping experiments are carried out on the healthy subjects according to the difference of the sex and the age.
And adopting multiple negative emotional stimulation modes to respectively stimulate the grouped healthy testees, then acquiring electroencephalograms of the healthy testees, and then forming sample data by the electroencephalograms and the negative emotional stimulation modes corresponding to all the electroencephalogram signals. The invention adopts a multi-pilot signal acquisition mode.
Step 102: and preprocessing the electroencephalogram signals in the sample data.
Because the electroencephalogram signal is very weak, the electroencephalogram signal is easily interfered by other noise signals in the acquisition process. The preprocessing of the electroencephalogram signals mainly means that all mixed artifacts in the acquired electroencephalogram signals are removed. The artifacts mainly comprise electro-oculogram, myoelectricity, electrocardio, power frequency interference, electromagnetic interference, task-unrelated electroencephalogram signals and the like. The electroencephalogram signals after the artifacts are removed are subjected to feature extraction for different analysis researches. In order to reduce the influence of artifacts on the analysis of the electroencephalogram signals, the electroencephalogram signals are required to be filtered and denoised.
The method adopts an Independent Component Analysis (ICA) algorithm to combine the waveform characteristics of original data to denoise the electroencephalogram signal (sample data) to obtain the preprocessed sample data. The original data waveform characteristic signal generally refers to the characteristic waveform characteristics of an eye movement signal, an electrocardiosignal and the like in an electroencephalogram signal. And finding out the corresponding IC signal waveform in the sample data signal calculated by the ICA algorithm according to the waveform characteristics of the eye movement signal, the heartbeat signal and the like, and removing the IC signal waveform to fulfill the aim of removing the noise signal.
Step 103: and performing feature extraction on the preprocessed sample data to obtain an initial emotion sample feature vector.
According to the time point of the first negative image stimulation, EEG (electroencephalogram) data of each negative emotional stimulation stage are sequentially extracted to form an epoch set, the power spectrum of the epoch set of the whole EEG data is locked in the same negative emotional stimulation mode, the average power change of the epoch set is calculated according to the frequency, and a two-dimensional time-frequency map of event-related disturbance (ERSP) is obtained. The method comprises the steps of performing feature analysis on a two-dimensional time-frequency map of event-related disturbance (ERSP), judging and obtaining a frequency range and brain area distribution with high recognition rate according to a one-way ANOVA algorithm, and extracting corresponding PSD values to form a high-dimensional feature vector as an initial emotion sample feature vector according to the frequency range and the brain area distribution with the high recognition rate.
Step 104: training a deep neural network based on the initial emotion sample feature vector and a negative emotion stimulation mode corresponding to the initial emotion sample feature to obtain a trained deep neural network model, and determining the middle layer feature of the trained deep neural network model as an optimized sample feature vector.
And (4) performing depth feature extraction by adopting a depth neural network model. The deep neural network can directly relate the features and the categories to which the features belong, namely, the multi-layer network structure of the deep neural network is utilized to realize the mapping from the input-layer features to the categories to which the features belong. According to the idea of deep learning, that is, the output of each layer of nodes can be regarded as the same projection in different feature spaces, it is obvious that the intermediate layer features intercepted from the middle of the network will also have certain distinctiveness.
In a Deep Neural Network (DNN), an original feature space is projected into a new feature space formed by nodes of a hidden layer, a feature representation form of the original feature space in a first hidden layer is obtained, and finally the original feature space is mapped into a state space through a Softmax network in an output layer and is associated with a class to which a feature belongs, as shown in fig. 3. And finally selecting a certain hidden layer in the middle by utilizing the original input of the network through the nonlinear mapping of a plurality of hidden layers, and intercepting the node output of the hidden layer as a new characteristic. The key of the feature extraction algorithm is how to train a deep learning model by using the PSD features of the original emotion.
Selecting Gaussian nodes for modeling in a network input layer and a last hidden layer (corresponding to an intermediate characteristic output layer); selecting Bernoulli type nodes from other hidden layers; the output layer selects nodes of the Softmax type.
The input and output of the Softmax type node satisfy:
wherein x isiAnd yiRepresenting the input and output of the node, respectively.
Training algorithm of the deep neural network:
firstly, a method of pre-training a Deep Belief Network (DBN) layer by layer is adopted to initialize network parameters. In the network parameter tuning stage, an emotion state id corresponding to each frame of training data is obtained through forced alignment and is used as a training label; training data is used as excitation, the training data reaches an output layer after being mapped by each hidden layer, and a node with the maximum output value is taken as a prediction label. And adjusting parameters by using an error back propagation algorithm by taking the minimum cross entropy as an objective function. The objective function is noted as:
wherein θ represents a network parameter; n represents the total number of different emotion labels; y (i) represents the output value of the i node of the network output layer under the condition of given training data, namely the predicted occurrence probability of the emotional state i; y' (i) represents the actual probability of occurrence of emotional state i.
After the deep neural network training is completed, the original input is mapped by using the network parameters to obtain the new characteristics of the emotion, namely
y＝f(x，θDNN)；
Where x represents the original acoustic feature input, θDNNAnd the parameters of the deep neural network are represented, and y represents the intermediate layer characteristics extracted after the deep neural network is mapped.
Step 105: training a classifier according to the optimized sample feature vector and the initial emotion sample feature vector, and determining a negative emotion recognition classification model.
The method comprises the following steps of respectively introducing common and popular classifier models through optimized characteristic parameters extracted based on a deep learning model and PSD characteristic parameters obtained based on ERSP analysis: such as algorithms of a support vector machine, K nearest neighbor, linear discriminant analysis, naive Bayes, random forest and the like, and the negative emotion recognition accuracy is compared and analyzed to determine a negative emotion recognition classification model suitable for high arousal degree. In the above steps, the classifier algorithm selection is not performed singly, and the invention considers selecting different algorithm combination (feature extraction model + classification model) modes to reach the classification model of the optimal collocation scheme. The specific operation steps are as follows: training a classifier according to the optimized sample feature vector and the initial emotion sample feature vector, and determining a plurality of initial negative emotion recognition classification models; and screening the initial negative emotion recognition classification model by using a cross validation algorithm and taking the minimum cross validation error rate as a standard to determine a final negative emotion recognition classification model.
Step 106: acquiring and processing the EEG signal of the testee. The processing procedure is the same as steps 102 to 104.
Step 107: and identifying the negative emotion of the testee according to the electroencephalogram signals processed by the testee and the negative emotion identification classification model.
In order to achieve the above object, the present invention further provides an electroencephalogram negative emotion recognition system based on aggressive behavior prediction, as shown in fig. 4, including:
a sample data obtaining module 201, configured to obtain sample data; the sample data comprises a plurality of data sets; each data set comprises electroencephalogram signals generated by stimulating healthy subjects of the same gender and the same age by adopting a plurality of negative emotional stimulation modes, and the negative emotional stimulation modes corresponding to the electroencephalogram signals, and the data sets are different corresponding to the ages or the sexes of the healthy subjects.
And the preprocessing module 202 is configured to preprocess the electroencephalogram signals in the sample data.
An initial emotion sample feature vector obtaining module 203, configured to perform feature extraction on the preprocessed sample data, and obtain an initial emotion sample feature vector.
And the optimized sample feature vector determining module 204 is configured to train a deep neural network based on the initial emotion sample feature vector and a negative emotion stimulation mode corresponding to the initial emotion sample feature to obtain a trained deep neural network model, and determine an intermediate layer feature of the trained deep neural network model as an optimized sample feature vector.
And the negative emotion recognition classification model determining module 205 is configured to train a classifier according to the optimized sample feature vector and the initial emotion sample feature vector, and determine a negative emotion recognition classification model.
And the electroencephalogram signal acquisition processing module 206 is used for acquiring and processing an electroencephalogram signal of the subject.
And the negative emotion recognition module 207 is used for recognizing the negative emotion of the testee according to the electroencephalogram signals processed by the testee and the negative emotion recognition classification model.
The preprocessing module 202 specifically includes:
and the preprocessing unit is used for denoising the electroencephalogram signal by adopting an independent component analysis algorithm and the waveform characteristics of the original data to obtain preprocessed sample data.
The initial emotion sample feature vector obtaining module 203 specifically includes:
and the epoch set combination unit is used for sequentially extracting the preprocessed electroencephalogram signals in each negative emotional stimulation mode in each data group according to the time point stimulated by the first negative emotional stimulation mode for each data group, and forming an epoch set.
And the two-dimensional time-frequency map determining unit is used for locking the power spectrum of the epoch set in the same negative emotional stimulation mode, and calculating according to frequency to obtain the two-dimensional time-frequency map of the event-related disturbance.
And the frequency range and brain area distribution determining unit is used for performing characteristic analysis on the two-dimensional time-frequency spectrum and judging to obtain the frequency range and brain area distribution by combining a one-factor variance analysis algorithm.
And the initial emotion sample feature vector acquisition unit is used for extracting a PSD value according to the frequency range and the brain area distribution, and forming a high-dimensional feature vector by using the PSD value as an initial emotion sample feature vector.
The negative emotion recognition classification model determining module 205 specifically includes:
and the initial negative emotion recognition classification model determining unit is used for training a classifier according to the optimized sample feature vector and the initial emotion sample feature vector and determining a plurality of initial negative emotion recognition classification models.
And the negative emotion recognition classification model determining unit is used for screening the initial negative emotion recognition classification model by using a cross validation algorithm and taking the minimum cross validation error rate as a standard to determine a final negative emotion recognition classification model.
In order to prevent offensive crimes, particularly aggressive behaviors of college student groups, the invention provides a method for establishing an electroencephalogram-based aggressive behavior prediction model by classifying and identifying by using a deep learning model in the research of discussing negative emotion cognition and aggressive behavior prediction. The psychological state of the testee is judged and evaluated through cognitive analysis and classification prediction of negative emotions with higher arousal degree, and corresponding emotion regulation and psychological treatment measures are combined for the tested college students which are predicted to generate high negative emotions easily, so that the occurrence of aggressive behaviors is avoided and prevented. The research content and achievement can further enrich the cognition of the electroencephalogram-based aggressive behavior tendency prediction mode, deepen the understanding of a human brain information processing mechanism in the negative emotion induction process, increase the flexibility, the high efficiency and the practicability of the current negative emotion recognition mode, and greatly improve the accuracy, so that the process of the electroencephalogram-based negative emotion aggressive behavior tendency prediction model towards practical application can be accelerated. In addition, the analysis and summary of the electroencephalogram signals generated under the negative emotional stimulation also provide reference basis for the field of cognitive physiology.
The embodiments in the present description are described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments are referred to each other. For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the method part for description.
The principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.
Claims (9)
1. An electroencephalogram negative emotion recognition method based on aggressive behavior prediction is characterized by comprising the following steps:
acquiring sample data; the sample data comprises a plurality of data sets; each data set comprises electroencephalogram signals generated by respectively stimulating healthy testees of the same gender and the same age by adopting a plurality of negative emotion stimulation modes, and the negative emotion stimulation mode corresponding to each electroencephalogram signal, and the ages or the genders of the healthy testees corresponding to different data sets are different;
preprocessing the electroencephalogram signals in the sample data;
performing feature extraction on the preprocessed sample data to obtain an initial emotion sample feature vector;
training a deep neural network based on the initial emotion sample feature vector and a negative emotion stimulation mode corresponding to the initial emotion sample feature to obtain a trained deep neural network model, and determining the intermediate layer feature of the trained deep neural network model as an optimized sample feature vector;
training a classifier according to the optimized sample feature vector and the initial emotion sample feature vector, and determining a negative emotion recognition classification model;
acquiring and processing an electroencephalogram signal of a subject;
and identifying the negative emotion of the testee according to the electroencephalogram signals processed by the testee and the negative emotion identification classification model.
2. The method for recognizing negative emotion of brain electricity based on aggressive behavior prediction as claimed in claim 1, wherein said preprocessing the brain electricity signal in said sample data specifically comprises:
and denoising the electroencephalogram signal by adopting an independent component analysis algorithm and the waveform characteristics of the original data to obtain preprocessed sample data.
3. The method for recognizing the electroencephalogram negative emotion based on the aggressive behavior prediction as recited in claim 1, wherein the step of performing feature extraction on the preprocessed sample data to obtain an initial emotion sample feature vector specifically comprises the steps of:
for each data group, sequentially extracting the preprocessed electroencephalogram signals in each negative emotional stimulation mode in the data group according to the time point of stimulation of the first negative emotional stimulation mode, and forming an epoch set;
locking the power spectrum of the epoch set in the same negative emotional stimulation mode, and calculating according to frequency to obtain a two-dimensional time-frequency spectrum of event-related disturbance;
performing characteristic analysis on the two-dimensional time-frequency spectrum, and judging by combining a one-factor variance analysis algorithm to obtain a frequency range and brain area distribution;
and extracting a PSD value according to the frequency range and the brain area distribution, and forming a high-dimensional feature vector by using the PSD value as an initial emotion sample feature vector.
4. The method for recognizing the negative emotion of the brain electricity based on aggressive behavior prediction as recited in claim 1, wherein the nodes of the network input layer and the last hidden layer of the trained deep neural network model are Gaussian-type nodes, the nodes of other hidden layers are Bernoulli-type nodes, and the node of the output layer is Softmax-type node.
5. The electroencephalogram negative emotion recognition method based on aggressive behavior prediction as recited in claim 1, wherein the training of a classifier according to the optimized sample feature vector and the initial emotion sample feature vector to determine a negative emotion recognition classification model specifically comprises:
training a classifier according to the optimized sample feature vector and the initial emotion sample feature vector, and determining a plurality of initial negative emotion recognition classification models;
and screening the initial negative emotion recognition classification model by using a cross validation algorithm and taking the minimum cross validation error rate as a standard to determine a final negative emotion recognition classification model.
6. An electroencephalogram negative emotion recognition system based on aggressive behavior prediction is characterized by comprising:
the sample data acquisition module is used for acquiring sample data; the sample data comprises a plurality of data sets; each data set comprises electroencephalogram signals generated by respectively stimulating healthy testees of the same gender and the same age by adopting a plurality of negative emotion stimulation modes, and the negative emotion stimulation mode corresponding to each electroencephalogram signal, and the ages or the genders of the healthy testees corresponding to different data sets are different;
the preprocessing module is used for preprocessing the electroencephalogram signals in the sample data;
the initial emotion sample feature vector acquisition module is used for extracting features of the preprocessed sample data to acquire an initial emotion sample feature vector;
the optimized sample feature vector determining module is used for training a deep neural network based on the initial emotion sample feature vector and a negative emotion stimulation mode corresponding to the initial emotion sample feature to obtain a trained deep neural network model, and determining the middle layer feature of the trained deep neural network model as an optimized sample feature vector;
the negative emotion recognition classification model determining module is used for training a classifier according to the optimized sample feature vector and the initial emotion sample feature vector and determining a negative emotion recognition classification model;
the electroencephalogram signal acquisition processing module is used for acquiring and processing an electroencephalogram signal of a subject;
and the negative emotion recognition module is used for recognizing the negative emotion of the testee according to the electroencephalogram signals processed by the testee and the negative emotion recognition classification model.
7. The system for recognizing brain negative emotion based on aggressive behavior prediction as claimed in claim 6, wherein said preprocessing module specifically comprises:
and the preprocessing unit is used for denoising the electroencephalogram signal by adopting an independent component analysis algorithm and the waveform characteristics of the original data to obtain preprocessed sample data.
8. The system for recognizing the negative emotion of the brain electricity based on aggressive behavior prediction as claimed in claim 6, wherein the initial emotion sample feature vector obtaining module specifically comprises:
the epoch set combination unit is used for sequentially extracting the preprocessed electroencephalogram signals in each negative emotional stimulation mode in each data group according to the stimulation time point of the first negative emotional stimulation mode for each data group, and forming an epoch set;
the two-dimensional time-frequency map determining unit is used for locking the power spectrum of the epoch set in the same negative emotional stimulation mode, and calculating according to frequency to obtain a two-dimensional time-frequency map of event-related disturbance;
the frequency range and brain area distribution determining unit is used for performing characteristic analysis on the two-dimensional time-frequency spectrum and judging to obtain a frequency range and brain area distribution by combining a one-factor variance analysis algorithm;
and the initial emotion sample feature vector acquisition unit is used for extracting a PSD value according to the frequency range and the brain area distribution, and forming a high-dimensional feature vector by using the PSD value as an initial emotion sample feature vector.
9. The system for recognizing negative emotion of brain electricity based on aggressive behavior prediction as claimed in claim 6, wherein said negative emotion recognition classification model determining module specifically comprises:
the initial negative emotion recognition classification model determining unit is used for training a classifier according to the optimized sample feature vector and the initial emotion sample feature vector and determining a plurality of initial negative emotion recognition classification models;
and the negative emotion recognition classification model determining unit is used for screening the initial negative emotion recognition classification model by using a cross validation algorithm and taking the minimum cross validation error rate as a standard to determine a final negative emotion recognition classification model.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911141842.8A CN110826527A (en) | 2019-11-20 | 2019-11-20 | Electroencephalogram negative emotion recognition method and system based on aggressive behavior prediction |
| AU2020100027A AU2020100027A4 (en) | 2019-11-20 | 2020-01-08 | Electroencephalogram-based negative emotion recognition method and system for aggressive behavior prediction |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911141842.8A CN110826527A (en) | 2019-11-20 | 2019-11-20 | Electroencephalogram negative emotion recognition method and system based on aggressive behavior prediction |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110826527A true CN110826527A (en) | 2020-02-21 |
Family
ID=69526426
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911141842.8A Withdrawn CN110826527A (en) | 2019-11-20 | 2019-11-20 | Electroencephalogram negative emotion recognition method and system based on aggressive behavior prediction |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN110826527A (en) |
| AU (1) | AU2020100027A4 (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111557660A (en) * | 2020-06-08 | 2020-08-21 | 东北大学 | Arrhythmia identification method under sub-population deep learning framework |
| CN111579939A (en) * | 2020-04-23 | 2020-08-25 | 天津大学 | Method for detecting partial discharge phenomenon of high-voltage power cable based on deep learning |
| CN112163518A (en) * | 2020-09-28 | 2021-01-01 | 华南理工大学 | Emotion modeling method for emotion monitoring and adjusting system |
| CN113128353A (en) * | 2021-03-26 | 2021-07-16 | 安徽大学 | Emotion sensing method and system for natural human-computer interaction |
| CN113180669A (en) * | 2021-05-12 | 2021-07-30 | 中国人民解放军中部战区总医院 | Emotional regulation training system and method based on nerve feedback technology |
| CN113191225A (en) * | 2021-04-19 | 2021-07-30 | 华南师范大学 | Emotional electroencephalogram recognition method and system based on graph attention network |
| CN113317802A (en) * | 2021-05-21 | 2021-08-31 | 上海六河创意设计有限公司 | Information early warning system and method based on brain waves |
| CN113486752A (en) * | 2021-06-29 | 2021-10-08 | 吉林大学 | Emotion identification method and system based on electrocardiosignals |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111860463B (en) * | 2020-08-07 | 2024-02-02 | 北京师范大学 | Emotion recognition method based on joint norm |
| CN113723365B (en) * | 2021-09-29 | 2024-05-14 | 西安电子科技大学 | Millimeter wave Lei Dadian cloud data-based target feature extraction and classification method |
| CN114662541A (en) * | 2022-03-24 | 2022-06-24 | 西人马(深圳)科技有限责任公司 | Fault diagnosis model construction method and device and electronic equipment |
-
2019
- 2019-11-20 CN CN201911141842.8A patent/CN110826527A/en not_active Withdrawn
-
2020
- 2020-01-08 AU AU2020100027A patent/AU2020100027A4/en not_active Ceased
Cited By (13)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111579939A (en) * | 2020-04-23 | 2020-08-25 | 天津大学 | Method for detecting partial discharge phenomenon of high-voltage power cable based on deep learning |
| CN111557660A (en) * | 2020-06-08 | 2020-08-21 | 东北大学 | Arrhythmia identification method under sub-population deep learning framework |
| CN112163518A (en) * | 2020-09-28 | 2021-01-01 | 华南理工大学 | Emotion modeling method for emotion monitoring and adjusting system |
| CN112163518B (en) * | 2020-09-28 | 2023-07-18 | 华南理工大学 | Emotion modeling method for emotion monitoring and adjusting system |
| CN113128353A (en) * | 2021-03-26 | 2021-07-16 | 安徽大学 | Emotion sensing method and system for natural human-computer interaction |
| CN113128353B (en) * | 2021-03-26 | 2023-10-24 | 安徽大学 | Emotion perception method and system oriented to natural man-machine interaction |
| CN113191225B (en) * | 2021-04-19 | 2023-07-04 | 华南师范大学 | Emotion electroencephalogram recognition method and system based on graph attention network |
| CN113191225A (en) * | 2021-04-19 | 2021-07-30 | 华南师范大学 | Emotional electroencephalogram recognition method and system based on graph attention network |
| CN113180669A (en) * | 2021-05-12 | 2021-07-30 | 中国人民解放军中部战区总医院 | Emotional regulation training system and method based on nerve feedback technology |
| CN113180669B (en) * | 2021-05-12 | 2024-04-26 | 中国人民解放军中部战区总医院 | Emotion adjustment training system and method based on nerve feedback technology |
| CN113317802A (en) * | 2021-05-21 | 2021-08-31 | 上海六河创意设计有限公司 | Information early warning system and method based on brain waves |
| CN113486752A (en) * | 2021-06-29 | 2021-10-08 | 吉林大学 | Emotion identification method and system based on electrocardiosignals |
| CN113486752B (en) * | 2021-06-29 | 2023-06-16 | 吉林大学 | Emotion recognition method and system based on electrocardiosignal |
Also Published As
| Publication number | Publication date |
|---|---|
| AU2020100027A4 (en) | 2020-02-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110826527A (en) | Electroencephalogram negative emotion recognition method and system based on aggressive behavior prediction | |
| Cheng et al. | Emotion recognition from multi-channel EEG via deep forest | |
| CN106886792B (en) | Electroencephalogram emotion recognition method for constructing multi-classifier fusion model based on layering mechanism | |
| Özseven | Investigation of the effect of spectrogram images and different texture analysis methods on speech emotion recognition | |
| Zheng | Multichannel EEG-based emotion recognition via group sparse canonical correlation analysis | |
| CN111329474B (en) | Electroencephalogram identity recognition method and system based on deep learning and information updating method | |
| CN114052735B (en) | Deep field self-adaption-based electroencephalogram emotion recognition method and system | |
| Ari et al. | Wavelet ELM-AE based data augmentation and deep learning for efficient emotion recognition using EEG recordings | |
| Zhou et al. | Deception detecting from speech signal using relevance vector machine and non-linear dynamics features | |
| Pfister et al. | Robustifying independent component analysis by adjusting for group-wise stationary noise | |
| Kulasingham et al. | Deep belief networks and stacked autoencoders for the p300 guilty knowledge test | |
| Lai et al. | Arrangements of resting state electroencephalography as the input to convolutional neural network for biometric identification | |
| CN115770044B (en) | Emotion recognition method and device based on electroencephalogram phase amplitude coupling network | |
| Hussain et al. | A radial base neural network approach for emotion recognition in human speech | |
| Wahengbam et al. | A group decision optimization analogy-based deep learning architecture for multiclass pathology classification in a voice signal | |
| CN115414051A (en) | Emotion classification and recognition method of electroencephalogram signal self-adaptive window | |
| Qi et al. | Robust deep network with maximum correntropy criterion for seizure detection | |
| Xu et al. | EEG emotion classification based on baseline strategy | |
| Stuchi et al. | Frequency learning for image classification | |
| Tang et al. | Deep neural network for emotion recognition based on meta-transfer learning | |
| Feather et al. | Model metamers illuminate divergences between biological and artificial neural networks | |
| Wijayanto et al. | Biometric identification based on EEG signal with photo stimuli using Hjorth descriptor | |
| Buza | Asterics: Projection-based classification of eeg with asymmetric loss linear regression and genetic algorithm | |
| Shashanka | Latent variable framework for modeling and separating single-channel acoustic sources | |
| Guodong et al. | Multi feature fusion EEG emotion recognition |
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 | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20200221 |
|
| WW01 | Invention patent application withdrawn after publication |