Method and device for classifying brain wave based on brain-computer interface

Abstract

A brainwave classification method is provided that can improve the accuracy of brainwave classification based on correlation coefficients between brainwaves measured on different days. The brainwave classification method extracts brainwaves from past brainwaves that have the most similar pattern to the currently measured brainwaves based on the correlation between the currently measured brainwaves and past brainwaves, and can accurately classify the currently measured brainwaves by using one of multiple classifiers previously learned from past brainwaves based on the extracted brainwaves.

Description

Technology Field

[0001] The present invention relates to a brainwave classification method and apparatus capable of improving the accuracy of brainwave classification based on correlation coefficients between brainwaves measured on different days. Background Technology

[0002] Brainwaves can be obtained from a subject through methods such as electroencephalography (EEG), which uses electrical signals resulting from brain activity; magnetoencephalography (MEG), which uses magnetic signals induced along with electrical signals; functional magnetic resonance imaging (FMRI), which uses changes in blood oxygen saturation; or near-infrared spectroscopy (NIRS). EEG signals, which offer excellent portability and temporal resolution, are widely used.

[0003] Brain-computer interface (BCI) devices capable of controlling external devices using these brainwaves are being developed. BCI devices can control the operation of the device by utilizing the intention (or imagination) of movement of specific body parts of the subject, spatiotemporal changes in brainwave patterns induced by specific stimuli, or changes in brainwaves in response to visual stimuli. Such BCI devices can be used as rehabilitation and daily living aids for individuals with limited physical mobility, such as patients with ALS (Lou Gehrig's disease), or as training tools for entertainment like games or for improving concentration.

[0004] A typical BCI device collects brainwaves to train a classifier by repeatedly performing pre-set tasks prior to the device's use. Specifically, the brain-computer interface device extracts feature values ​​capable of distinguishing task types from arbitrary brainwaves based on the collected brainwaves, and trains the classifier using the extracted feature values.

[0005] However, brainwaves have the characteristic of changing patterns depending on changes in the subject's state or the measurement date. Accordingly, conventional BCI devices perform learning on the classifier every time the device is used to prevent a decrease in classification performance due to changes in brainwave patterns. This has resulted in a problem of increased fatigue for users of conventional BCI devices. Prior art literature

[0006] Korean Registered Patent No. 10-1939363 (January 16, 2019) The problem to be solved

[0007] The present invention aims to provide a brainwave classification method and device capable of improving the accuracy of brainwave classification based on correlation coefficients between brainwaves measured on different days. means of solving the problem

[0008] A brainwave classification method according to one embodiment of the present invention comprises: a step of extracting a plurality of second coefficients corresponding to each of the plurality of parameters from a second brainwave based on a plurality of dictionary coefficients corresponding to each of the plurality of parameters extracted from a first brainwave and a plurality of first coefficients; a step of calculating a correlation coefficient between the plurality of first coefficients and the plurality of second coefficients and selecting one of the plurality of parameters based on the calculated correlation coefficient; and a step of selecting one classifier corresponding to the selected parameter from among a plurality of classifiers that have been learned corresponding to each of the plurality of parameters, and classifying the second brainwave through the selected classifier.

[0009] Here, the first brainwave and the second brainwave may each be signals measured on different days.

[0010] The step of extracting the plurality of second coefficients is a step of extracting the plurality of second coefficients corresponding to each of the plurality of parameters using a lasso regression algorithm.

[0011] The step of calculating the correlation coefficient comprises: a step of calculating a first average value for the plurality of first coefficients; a step of calculating a second average value for the plurality of second coefficients; a step of calculating a correlation coefficient between the first average value and the second average value; a step of calculating a similarity value corresponding to each of the plurality of parameters based on the correlation coefficient; and a step of selecting one parameter corresponding to the maximum similarity value among the plurality of parameters.

[0012] Here, each of the plurality of dictionary coefficients and the plurality of first coefficients is extracted from the first brainwave using a sparse dictionary learning algorithm.

[0013] Additionally, each of the plurality of classifiers is trained to classify each of the plurality of first coefficients into target brainwaves and non-target brainwaves when receiving each of the plurality of first coefficients and a correct classification answer for each of the plurality of first coefficients as label data. Here, the target brainwave may be a P300 brainwave.

[0014] The brainwave classification method of the present embodiment further includes the step of extracting a third coefficient from the measured brainwave through Lasso regression based on the selected parameters. At this time, the selected classifier classifies the third coefficient into one of a target brainwave and a non-target brainwave.

[0015] Each of the above plurality of dictionary coefficients includes a pattern for the first brainwave, each of the above plurality of first coefficients includes a numerical value for the first brainwave, and each of the above plurality of second coefficients includes a numerical value for the second brainwave.

[0016] A brainwave classification device according to an embodiment of the present invention includes: a memory storing a brainwave classification program for classifying brainwaves; and a processor that executes the brainwave classification program to extract a plurality of second coefficients corresponding to each of the plurality of parameters from a second brainwave based on a plurality of dictionary coefficients corresponding to each of the plurality of parameters extracted from a first brainwave and a plurality of first coefficients, calculates a correlation coefficient between the plurality of first coefficients and the plurality of second coefficients, selects one of the plurality of parameters based on the calculated correlation coefficient, selects one classifier corresponding to the selected parameter from a plurality of classifiers that have been learned corresponding to each of the plurality of parameters, and classifies the second brainwave through the selected classifier. Here, the first brainwave and the second brainwave may each be signals measured on different days.

[0017] Each of the above plurality of classifiers is trained to classify each of the plurality of first coefficients into target brainwaves and non-target brainwaves when receiving each of the plurality of first coefficients and a correct classification answer for each of the plurality of first coefficients as label data. At this time, the target brainwave may be a P300 brainwave.

[0018] The processor calculates a similarity value corresponding to each of the plurality of parameters based on the correlation coefficient between the average value of the plurality of first coefficients and the average value of the plurality of second coefficients, and selects one parameter corresponding to the maximum similarity value among the plurality of parameters.

[0019] In addition, the processor extracts a plurality of dictionary coefficients and a first coefficient corresponding to each of the plurality of parameters from the first brainwave through a previously learned sparse dictionary learning algorithm, and extracts a plurality of second coefficients corresponding to each of the plurality of parameters from the second brainwave through a Lasso regression algorithm. Effects of the invention

[0020] The present invention can extract brainwaves having a pattern most similar to the currently measured brainwaves among the previously measured brainwaves based on the correlation between the currently measured brainwaves and the previously measured brainwaves.

[0021] Accordingly, the present invention can accurately classify a specific brainwave among currently measured brainwaves, for example, a P300 brainwave used in a brain-computer interface device, by using one of a plurality of classifiers that have been learned from past measured brainwaves based on extracted brainwaves.

[0022] Therefore, the present invention can improve the classification accuracy of P300 brainwaves for brainwaves that change according to the measurement date, even without training the classifier every time a brainwave is measured. Brief explanation of the drawing

[0023] FIG. 1 is a block diagram showing a brainwave classification device according to an embodiment of the present invention. Figure 2 is a block diagram conceptually illustrating the functions of the brainwave classification program of Figure 1. Figure 3 is a diagram showing the coefficient extraction unit of Figure 2. Figure 4 is a diagram showing a method for training the brainwave classification unit of Figure 2. FIG. 5 is a flowchart illustrating a brainwave classification method according to an embodiment of the present invention. Specific details for implementing the invention

[0024] The advantages and features of the present invention and the methods for achieving them will become clear by referring to the embodiments described below in detail together with the accompanying drawings. However, the present invention is not limited to the embodiments disclosed below but can be implemented in various different forms. These embodiments are provided merely to ensure that the disclosure of the present invention is complete and to fully inform those skilled in the art of the scope of the invention, and the present invention is defined only by the scope of the claims.

[0025] In describing the embodiments of the present invention, specific descriptions of known functions or configurations will be omitted if it is determined that such detailed descriptions could unnecessarily obscure the essence of the invention. Furthermore, the terms described below are defined in consideration of their functions in the embodiments of the present invention, and these definitions may vary depending on the intentions or practices of the user or operator. Therefore, such definitions should be based on the content throughout this specification.

[0026] Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the attached drawings.

[0027] FIG. 1 is a block diagram showing a brainwave classification device according to an embodiment of the present invention.

[0028] Referring to FIG. 1, the brainwave classification device (100) of the present embodiment may include an input / output unit (110), a processor (120), and a memory (130).

[0029] The processor (120) can control the overall operation of the brainwave classification device (100). The processor (120) receives brainwaves measured from a subject through the input / output unit (110) and classifies the measured brainwaves using a brainwave classification program (140) stored in the memory (130) to be described later.

[0030] Here, the measured brainwaves can be classified into target brainwaves and non-target brainwaves, and the target brainwaves may be P300 brainwaves among various brainwaves.

[0031] The memory (130) can store information necessary for the execution of the brainwave classification program (140) and the same. The brainwave classification program (140) of the present embodiment may be software that includes a plurality of commands programmed to classify the corresponding brainwave into target brainwaves and non-target brainwaves based on the pattern of the input brainwave, such as the numerical value of the brainwave.

[0032] Accordingly, the processor (120) loads and executes a brainwave classification program (140) from memory (130), and can use it to classify and output brainwaves input from the outside.

[0033] Figure 2 is a block diagram conceptually illustrating the functions of the brainwave classification program of Figure 1.

[0034] Referring to FIGS. 1 and 2, the brainwave classification program (140) of the present embodiment may include a coefficient extraction unit (210), a correlation analysis unit (220), and a brainwave classification unit (230).

[0035] The coefficient extraction unit (210), correlation analysis unit (220), and brainwave classification unit (230) illustrated in FIG. 2 are conceptually separated to easily explain the functions of the brainwave classification program (140), but the present invention is not limited thereto.

[0036] For example, according to an embodiment of the present invention, the coefficient extraction unit (210), the correlation analysis unit (220), and the brainwave classification unit (230) may have their functions merged or separated, and may be implemented as a series of instructions included in a single program.

[0037] The coefficient extraction unit (210) can extract coefficients corresponding to each of a plurality of preset parameters from brainwaves input through the input / output unit (110).

[0038] Here, the coefficient extraction unit (210) can extract multiple second coefficients corresponding to each of the multiple parameters from the second brainwave measured on the current date, for example, the second day after the first day, based on multiple dictionary coefficients corresponding to each of the multiple parameters extracted from the first brainwave measured on the previous date, for example, the first day, and the first coefficient.

[0039] Figure 3 is a diagram showing the coefficient extraction unit of Figure 2.

[0040] Referring to FIG. 3, the coefficient extraction unit (210) may include a first coefficient extraction unit (211) and a second coefficient extraction unit (215).

[0041] The first coefficient extraction unit (211) can extract a plurality of dictionary coefficients (D[λ1:λn]) and a plurality of first coefficients (C1[λ1:λn]) corresponding to a plurality of parameters (λ1 to λn) from the first brainwave (BS1).

[0042] Here, a plurality of parameters (λ1 to λn) may include a first parameter (λ1) to an nth (n is a natural number) parameter (λn). Accordingly, the first coefficient extraction unit (211) can extract dictionary coefficients (D[λ1:λn]) and first coefficients (C1[λ1:λn]) corresponding to each parameter (λ1 to λn).

[0043] This first coefficient extraction unit (211) can extract dictionary coefficients (D[λ1:λn]) and first coefficients (C1[λ1:λn]) for each parameter (λ1 to λn) from the first brainwave (BS1) using a sparse dictionary learning algorithm.

[0044] At this time, the dictionary coefficient (D[λ1:λn]) for each parameter (λ1 to λn) includes one or more patterns for the first brainwave (BS1), and the first coefficient (C1[λ1:λn]) for each parameter (λ1 to λn) may include numerical values ​​for the first brainwave (BS1).

[0045] Additionally, the first brainwave (BS1) may be a signal composed of brainwaves measured multiple times on the first day. Accordingly, the first coefficient extraction unit (211) can extract multiple dictionary coefficients (D[λ1:λn]) and multiple first coefficients (C1[λ1:λn]) for each parameter (λ1 to λn) according to the number of measurements of the first brainwave (BS1).

[0046] The first coefficient extraction unit (211) can output a plurality of dictionary coefficients (D[λ1:λn]) to the second coefficient extraction unit (215) and output a plurality of first coefficients (C1[λ1:λn]) to the brainwave classification unit (230) to be described later.

[0047] The second coefficient extraction unit (215) can extract a plurality of second coefficients (C2[λ1:λn]) corresponding to a plurality of parameters (λ1 to λn) from the second brainwave (BS2) based on a plurality of dictionary coefficients (D[λ1:λn]) provided by the first coefficient extraction unit (211).

[0048] As described above, a plurality of parameters (λ1 to λn) may include a first parameter (λ1) to an nth parameter (λn) (n is a natural number). Accordingly, the second coefficient extraction unit (215) can extract a second coefficient (C2[λ1:λn]) corresponding to each parameter (λ1 to λn).

[0049] This second coefficient extraction unit (215) can extract a plurality of second coefficients (C2[λ1:λn]) corresponding to a plurality of parameters (λ1 to λn) from the second brainwave (BS2) using a lasso regression algorithm. At this time, the second coefficients (C2[λ1:λn]) for each parameter (λ1 to λn) may include numerical values ​​for the second brainwave (BS2).

[0050] Here, the second brainwave (BS2) may be a brainwave measured once on the second day. Accordingly, the second coefficient extraction unit (215) can extract a plurality of second coefficients (C2[λ1:λn]) for each parameter (λ1 to λn) from the second brainwave (BS2) provided by a single measurement.

[0051] The correlation analysis unit (220) calculates a correlation coefficient, such as a Pearson correlation coefficient, between a plurality of first coefficients (C1[λ1:λn]) extracted from the coefficient extraction unit (210) and a plurality of second coefficients (C2[λ1:λn]), and can select one parameter from a plurality of parameters (λ1 to λn) based on the calculated correlation coefficient.

[0052] For example, the correlation analysis unit (220) can calculate an average value for a plurality of first coefficients (C1[λ1:λn]), such as a first average value. Here, since the plurality of first coefficients (C1[λ1:λn]) are extracted as many times as the first brainwave (BS1) is measured, the first average value can also be calculated as many times as the first brainwave (BS1) is measured.

[0053] For example, if the first brainwave (BS1) is measured 10 times and there are 10 pre-set parameters (λ1 to λn), there may be 10 first coefficients (C1[λ1:λn]) for each measurement, and the total number may be the product of the number of measurements and the number of parameters (λ1 to λn). Accordingly, the correlation analysis unit (220) can calculate the average value of the multiple first coefficients (C1[λ1:λn]) for each measurement and calculate 10 first average values.

[0054] Additionally, the correlation analysis unit (220) can calculate an average value for a plurality of second coefficients (C2[λ1:λn]), such as a second average value. Here, since the plurality of second coefficients (C2[λ1:λn]) are extracted from a second brainwave (BS2) measured once, there may be only one second average value.

[0055] The correlation analysis unit (220) can calculate the correlation coefficient between the first average value and the second average value. Here, the correlation analysis unit (220) can calculate the correlation coefficient between the two average values ​​according to the following [Mathematical Formula 1]. At this time, as previously mentioned, since there are multiple first average values ​​and one second average value, the correlation analysis unit (220) can calculate a number of correlation coefficients corresponding to each of the multiple first average values.

[0056] [Mathematical Formula 1]

[0057]

[0058] Here, mλ,n is one of a plurality of first mean values, mλ,1 is a second mean value, cov() is the covariance of the first mean value and the second mean value, and σ represents the variance of the first mean value and the second mean value.

[0059] The correlation analysis unit (220) can determine the calculated multiple correlation coefficients as similarity values ​​corresponding to each of the multiple parameters (λ1 to λn). Then, one value, such as a maximum similarity value, can be extracted from the similarity values ​​corresponding to each of the multiple parameters (λ1 to λn), and one parameter corresponding to the extracted maximum similarity value can be selected from the multiple parameters.

[0060] The brainwave classification unit (230) can classify the second brainwave (BS2) into one of the target brainwave and the non-target brainwave. As illustrated in FIG. 4, this brainwave classification unit (230) may include a plurality of classifiers (241, 245), and can classify the second brainwave (BS2) using one of the plurality of classifiers (241, 245) according to a pre-selected parameter.

[0061] Multiple classifiers (241, 245) of the brainwave classification unit (230) can be learned based on multiple first coefficients (C1[λ1:λn]) extracted from the aforementioned coefficient extraction unit (210).

[0062] Figure 4 is a diagram showing a method for training the brainwave classification unit of Figure 2.

[0063] Referring to FIG. 4, the brainwave classification unit (230) may include a plurality of classifiers (241, 245) corresponding to each of a plurality of first coefficients (C1[λ1:λn]).

[0064] Each of the multiple classifiers (241, 245) can be trained to receive the correct answer for brainwave classification for the first coefficient (C1[λ1:λn]) as label data along with the corresponding first coefficient (C1[λ1:λn]), and to classify the first coefficient (C1[λ1:λn]) into target brainwaves and non-target brainwaves and output them.

[0065] Each of the multiple classifiers (241, 245) can compare the correct classification result input as label data with the actual output classification result and generate a classification loss from the comparison result. Accordingly, each of the multiple classifiers (241, 245) can repeat the aforementioned learning, that is, the learning of classifying the first coefficient (C1[λ1:λn]) into one of the target brainwave and the non-target brainwave, so that the classification loss is minimized.

[0066] Additionally, each of the plurality of first coefficients (C1[λ1:λn]) can correspond to each of the plurality of parameters (λ1 to λn). Accordingly, each of the plurality of classifiers (241, 245) can also correspond to each of the plurality of parameters (λ1 to λn).

[0067] Accordingly, the brainwave classification unit (230) can select one classifier corresponding to one parameter selected by the correlation analysis unit (220) from among a plurality of classifiers (241, 245) learned based on a plurality of first coefficients (C1[λ1:λn]). Then, the brainwave classification unit (230) can classify the second brainwave (BS2) into one of the target brainwave and the non-target brainwave using the selected classifier.

[0068] Meanwhile, the aforementioned coefficient extraction unit (210) can extract a third coefficient from the second brainwave (BS2) based on a single parameter selected by the correlation analysis unit (220). The coefficient extraction unit (210) provides the third coefficient to the brainwave classification unit (230), and the brainwave classification unit (230) can output a classification result for the second brainwave (BS2) from the third coefficient using a selected classifier. Here, the coefficient extraction unit (210) can extract a third coefficient from the second brainwave (BS2) through a Lasso regression algorithm based on a single parameter.

[0069] FIG. 5 is a flowchart illustrating a brainwave classification method according to an embodiment of the present invention.

[0070] Hereinafter, for convenience of explanation, a brainwave classification method of the brainwave classification device (100) according to the present embodiment will be described with reference to FIG. 1 to FIG. 4 together with FIG. 5.

[0071] First, the brainwave classification device (100) may receive the first brainwave (BS1) of a subject measured on a past measurement date, for example, the first day. The processor (120) of the brainwave classification device (100) executes a brainwave classification program (140) stored in memory (130), and through the coefficient extraction unit (210) thereof, may extract a plurality of dictionary coefficients (D[λ1:λn]) corresponding to a plurality of preset parameters (λ1 to λn) and a plurality of first coefficients (C1[λ1:λn]) respectively from the first brainwave (BS1).

[0072] For example, the first coefficient extraction unit (211) of the coefficient extraction unit (210) can extract a plurality of dictionary coefficients (D[λ1:λn]) and a plurality of first coefficients (C1[λ1:λn]) corresponding to a plurality of parameters (λ1 to λn) using a sparse dictionary learning algorithm.

[0073] Here, a plurality of parameters (λ1 to λn) may include a first parameter (λ1) to an nth (n is a natural number) parameter (λn). Additionally, a dictionary coefficient (D[λ1:λn]) for each parameter (λ1 to λn) may include one or more patterns for the first brainwave (BS1), and a first coefficient (C1[λ1:λn]) for each parameter (λ1 to λn) may include a numerical value for the first brainwave (BS1).

[0074] Additionally, the first brainwave (BS1) may be a signal composed of brainwaves measured multiple times on the first day. Accordingly, the first coefficient extraction unit (211) can extract multiple dictionary coefficients (D[λ1:λn]) and multiple first coefficients (C1[λ1:λn]) for each parameter (λ1 to λn) according to the number of measurements of the first brainwave (BS1).

[0075] For example, if the first brainwave (BS1) is measured 10 times and there are 10 pre-set parameters (λ1 to λn), then the multiple first coefficients (C1[λ1:λn]) can be extracted as 10 coefficients for each measurement.

[0076] A plurality of first coefficients (C1[λ1:λn]) extracted from the first coefficient extraction unit (211) are provided to the brainwave classification unit (230) to train a plurality of classifiers (241, 245) of the brainwave classification unit (230). For example, each of the plurality of classifiers (241, 245) of the brainwave classification unit (230) can be trained to receive each of the plurality of first coefficients (C1[λ1:λn]) and the corresponding classification answer, and to output a classification result for each of the plurality of first coefficients (C1[λ1:λn]), such as a target brainwave or a non-target brainwave.

[0077] After the classification learning of the brainwave classification unit (230) is completed, the brainwave classification device (100) may receive the second brainwave (BS2) of the subject measured on the current measurement date, for example, the second day. The processor (120) of the brainwave classification device (100) executes the brainwave classification program (140) stored in the memory (130) and can extract a plurality of second coefficients (C2[λ1:λn]) corresponding to a plurality of preset parameters (λ1 to λn) from the second brainwave (BS2) through the coefficient extraction unit (210) thereof (S10).

[0078] For example, the second coefficient extraction unit (215) of the coefficient extraction unit (210) can extract a plurality of second coefficients (C2[λ1:λn]) corresponding to a plurality of parameters (λ1 to λn) from the second brainwave (BS2) based on a plurality of dictionary coefficients (D[λ1:λn]) for the first brainwave (BS1). At this time, the second coefficient extraction unit (215) can extract a plurality of second coefficients (C2[λ1:λn]) from the second brainwave (BS2) using a lasso regression algorithm.

[0079] Next, the correlation analysis unit (220) calculates a correlation coefficient, such as a Pearson correlation coefficient, between a plurality of first coefficients (C1[λ1:λn]) and a plurality of second coefficients (C2[λ1:λn]) extracted from the coefficient extraction unit (210), and can determine the calculated correlation coefficient as a similarity value corresponding to each of the plurality of parameters (λ1 to λn) (S20).

[0080] First, the correlation analysis unit (220) can calculate a first average value for a plurality of first coefficients (C1[λ1:λn]). Here, if a plurality of first coefficients (C1[λ1:λn]) are calculated for each of the multiple measurements of the first brainwave (BS1), the correlation analysis unit (220) can calculate a first average value for a plurality of first coefficients (C1[λ1:λn]) for each of the multiple measurements.

[0081] Additionally, the correlation analysis unit (220) can calculate a second average value for a plurality of second coefficients (C2[λ1:λn]). Here, since the plurality of second coefficients (C2[λ1:λn]) are extracted from a second brainwave (BS2) measured once, there may be only one second average value.

[0082] Next, the correlation analysis unit (220) can calculate the correlation coefficient between the first average value and the second average value. At this time, since there are multiple first average values ​​and one second average value, the correlation analysis unit (220) can calculate the number of correlation coefficients corresponding to each of the multiple first average values.

[0083] And, the correlation analysis unit (220) can determine the calculated multiple correlation coefficients as similarity values ​​corresponding to each of the multiple parameters (λ1 to λn).

[0084] Next, the correlation analysis unit (220) can extract the maximum similarity value among the calculated multiple similarity values ​​and select one parameter corresponding to the extracted maximum similarity value among the multiple parameters (S30).

[0085] Accordingly, the brainwave classification unit (230) can select one classifier corresponding to one parameter selected by the correlation analysis unit (220) among a plurality of classifiers (241, 245) (S40). At this time, each of the plurality of classifiers (241, 245) of the brainwave classification unit (230) may be in a learned state using a plurality of first coefficients (C1[λ1:λn]) previously extracted from the subject's first brainwave (BS1).

[0086] Furthermore, the coefficient extraction unit (210) can extract a corresponding third coefficient from the second brainwave (BS2) based on one parameter selected by the correlation analysis unit (220) (S50).

[0087] Next, the brainwave classification unit (230) can classify the third coefficient extracted from the second brainwave (BS2) into one of the target brainwave and the non-target brainwave using a classifier selected by one parameter (S60).

[0088] As described above, the brainwave classification method of the present embodiment can train each of the multiple classifiers of the brainwave classification unit using multiple first coefficients for each parameter extracted from the brainwave measured on the subject's past measurement date. Then, by extracting multiple second coefficients for each parameter from the brainwave measured on the subject's current measurement date, analyzing the correlation with the multiple first coefficients for the brainwave of the past measurement date, and selecting one of the previously trained multiple classifiers accordingly, the brainwave of the current measurement date can be classified into either a target brainwave or a non-target brainwave.

[0089] Accordingly, the present invention extracts a brainwave having the most similar pattern to the currently measured brainwave among the previously measured brainwaves based on the correlation between the currently measured brainwave and the previously measured brainwave, and accordingly, accurately classifies the currently measured brainwave by using one of a plurality of classifiers previously learned from the previously measured brainwaves based on the corresponding brainwave. Therefore, the present invention can improve the classification accuracy of brainwaves that change depending on the measurement date.

[0090] The combinations of each block of the block diagram of the present invention and each step of the flowchart described above may be executed by computer program instructions. Since these computer program instructions may be loaded into an encoding processor of a general-purpose computer, a special-purpose computer, or other programmable data processing equipment, the instructions executed through the encoding processor of the computer or other programmable data processing equipment create means for performing the functions described in each block of the block diagram or each step of the flowchart. Since these computer program instructions may also be stored in computer-available or computer-readable memory (or computer-readable recording medium) that can be directed toward a computer or other programmable data processing equipment to implement the function in a specific way, the instructions stored in the computer-available or computer-readable memory (or computer-readable recording medium) may also produce a manufactured item containing instruction means for performing the function described in each block of the block diagram or each step of the flowchart. Since computer program instructions can be loaded onto a computer or other programmable data processing equipment, instructions that execute a computer or other programmable data processing equipment by performing a series of operation steps on the computer or other programmable data processing equipment to create a process executed by the computer may also provide steps for executing the functions described in each block of the block diagram and each step of the flowchart.

[0091] Additionally, each block or each step may represent a module, segment, or part of code containing one or more executable instructions for executing a specified logical function(s). Also, it should be noted that in some alternative embodiments, the functions mentioned in the blocks or steps may occur out of order. For example, two blocks or steps described in succession may actually be performed substantially simultaneously, or the blocks or steps may sometimes be performed in reverse order according to the corresponding function.

[0092] The above description is merely an illustrative explanation of the technical concept of the present invention, and those skilled in the art to which the present invention pertains will be able to make various modifications and variations within the scope of the essential quality of the present invention. Accordingly, the embodiments disclosed in the present invention are intended to explain, not limit, the technical concept of the present invention, and the scope of the technical concept of the present invention is not limited by such embodiments. The scope of protection of the present invention shall be interpreted by the claims below, and all technical concepts within the equivalent scope shall be interpreted as being included within the scope of rights of the present invention. Explanation of the symbols

[0093] 100: Brainwave classification device 110: Input / output unit 120: Processor 130: Memory 140: Brainwave Classification Program

Claims

Claim 1 A brainwave classification method performed by a brainwave classification device, comprising: a step of extracting a plurality of dictionary coefficients and a plurality of first coefficients corresponding to each of a plurality of parameters extracted from a first brainwave using a sparse dictionary learning algorithm; a step of extracting a plurality of second coefficients corresponding to each of the plurality of parameters from a second brainwave using the plurality of dictionary coefficients and a lasso regression algorithm; a step of calculating a correlation coefficient between the plurality of first coefficients and the plurality of second coefficients and selecting one parameter among the plurality of parameters that has the maximum calculated correlation coefficient; and a step of selecting one classifier corresponding to the selected parameter from among a plurality of classifiers that have been learned corresponding to each of the plurality of parameters, and classifying the second brainwave into one of a target brainwave and a non-target brainwave using the selected classifier. Claim 2 A brainwave classification method according to claim 1, wherein each of the first brainwave and the second brainwave is a signal measured on a different day. Claim 3 delete Claim 4 delete Claim 5 delete Claim 6 A brainwave classification method according to claim 1, wherein each of the plurality of classifiers is trained to classify each of the plurality of first coefficients into target brainwaves and non-target brainwaves when receiving each of the plurality of first coefficients and a classification answer for each of the plurality of first coefficients as label data. Claim 7 In paragraph 6, the above target brainwave is a brainwave classification method in which the target brainwave is a P300 brainwave. Claim 8 delete Claim 9 A brainwave classification method according to claim 1, wherein each of the plurality of dictionary coefficients includes a pattern for the first brainwave, each of the plurality of first coefficients includes a numerical value for the first brainwave, and each of the plurality of second coefficients includes a numerical value for the second brainwave. Claim 10 A brainwave classification device comprising: a memory storing a brainwave classification program for classifying brainwaves; and a processor that executes the brainwave classification program to extract a plurality of dictionary coefficients and a plurality of first coefficients corresponding to each of a plurality of parameters extracted from a first brainwave using a sparse dictionary learning algorithm, extracts a plurality of second coefficients corresponding to each of the plurality of parameters from a second brainwave using the plurality of dictionary coefficients and a Lasso regression algorithm, calculates a correlation coefficient between the plurality of first coefficients and the plurality of second coefficients, selects one parameter among the plurality of parameters that has the maximum calculated correlation coefficient, selects one classifier corresponding to the selected parameter from among a plurality of classifiers that have been learned corresponding to each of the plurality of parameters, and classifies the second brainwave into one of a target brainwave and a non-target brainwave using the selected classifier, wherein the first brainwave and the second brainwave are signals measured on different days. Claim 11 In claim 10, each of the plurality of classifiers is trained to classify each of the plurality of first coefficients into target brainwaves and non-target brainwaves when receiving each of the plurality of first coefficients and a classification answer for each of the plurality of first coefficients as label data, and the target brainwave is a P300 brainwave, in a brainwave classification device. Claim 12 delete Claim 13 delete Claim 14 A brainwave classification device according to claim 10, wherein each of the plurality of dictionary coefficients includes a pattern for the first brainwave, each of the plurality of first coefficients includes a numerical value for the first brainwave, and each of the plurality of second coefficients includes a numerical value for the second brainwave. Claim 15 A computer-readable recording medium storing a computer program, wherein the computer program comprises instructions for performing a brainwave classification method, the steps of: extracting a plurality of dictionary coefficients and a plurality of first coefficients corresponding to each of a plurality of parameters extracted from a first brainwave using a sparse dictionary learning algorithm; extracting a plurality of second coefficients corresponding to each of the plurality of parameters from a second brainwave using the plurality of dictionary coefficients and a Lasso regression algorithm; calculating a correlation coefficient between the plurality of first coefficients and the plurality of second coefficients and selecting one parameter among the plurality of parameters having the maximum calculated correlation coefficient; and selecting one classifier corresponding to the selected parameter from among a plurality of classifiers that have been learned corresponding to each of the plurality of parameters, and classifying the second brainwave into one of a target brainwave and a non-target brainwave using the selected classifier. Claim 16 A computer program stored on a computer-readable recording medium, wherein the computer program comprises instructions for performing a brainwave classification method, the computer program comprising: a step of extracting a plurality of dictionary coefficients and a plurality of first coefficients corresponding to each of a plurality of parameters extracted from a first brainwave using a sparse dictionary learning algorithm; a step of extracting a plurality of second coefficients corresponding to each of the plurality of parameters from a second brainwave using the plurality of dictionary coefficients and a Lasso regression algorithm; a step of calculating a correlation coefficient between the plurality of first coefficients and the plurality of second coefficients and selecting one parameter among the plurality of parameters having the maximum calculated correlation coefficient; and a step of selecting one classifier corresponding to the selected parameter from among a plurality of classifiers that have been learned corresponding to each of the plurality of parameters, and classifying the second brainwave into one of a target brainwave and a non-target brainwave using the selected classifier.

Images