CN112433616B - Multi-user multi-dimensional recognition model collaborative decision making system and method - Google Patents

Multi-user multi-dimensional recognition model collaborative decision making system and method Download PDF

Info

Publication number
CN112433616B
CN112433616B CN202011447933.7A CN202011447933A CN112433616B CN 112433616 B CN112433616 B CN 112433616B CN 202011447933 A CN202011447933 A CN 202011447933A CN 112433616 B CN112433616 B CN 112433616B
Authority
CN
China
Prior art keywords
dimensional
tasks
task
identified
decision
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN202011447933.7A
Other languages
Chinese (zh)
Other versions
CN112433616A (en
Inventor
曾令李
叶泽祺
于扬
周宗潭
胡德文
李月昊
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
National University of Defense Technology
Original Assignee
National University of Defense Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by National University of Defense Technology filed Critical National University of Defense Technology
Priority to CN202011447933.7A priority Critical patent/CN112433616B/en
Publication of CN112433616A publication Critical patent/CN112433616A/en
Application granted granted Critical
Publication of CN112433616B publication Critical patent/CN112433616B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F3/00—Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
    • G06F3/01—Input arrangements or combined input and output arrangements for interaction between user and computer
    • G06F3/011—Arrangements for interaction with the human body, e.g. for user immersion in virtual reality
    • G06F3/015—Input arrangements based on nervous system activity detection, e.g. brain waves [EEG] detection, electromyograms [EMG] detection, electrodermal response detection
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/901—Indexing; Data structures therefor; Storage structures
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/44—Arrangements for executing specific programs
    • G06F9/451—Execution arrangements for user interfaces
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F2218/00—Aspects of pattern recognition specially adapted for signal processing

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • General Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • General Physics & Mathematics (AREA)
  • Physics & Mathematics (AREA)
  • Human Computer Interaction (AREA)
  • Databases & Information Systems (AREA)
  • Neurosurgery (AREA)
  • Health & Medical Sciences (AREA)
  • Neurology (AREA)
  • General Health & Medical Sciences (AREA)
  • Dermatology (AREA)
  • Biomedical Technology (AREA)
  • Data Mining & Analysis (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention provides a multi-user multi-dimensional recognition model collaborative decision-making system and a method. The multi-person multi-dimensional identification model collaborative decision making system comprises a multi-dimensional identification model and a calculation processing module, wherein the multi-dimensional identification model is a model obtained by combining MDAbstract layout of candidate tasks into a containing MDIn the D-dimensional structure space of each node, K decision makers identify tasks to be identified in the D-dimensional structure space, and each decision maker is responsible for M tasks including MD‑1The D-1 dimensional subspace block of each alternative task identifies the D-1 dimensional subspace block in which the task to be identified is positioned, and the calculation processing module obtains the multidimensional coordinate (C) in which the task to be identified is positioned by calculating the identification results of all decision makers1,C2,…CD‑1,CD) To obtain a recognition result or a recognition area, and when K is D, to obtain all values of the multi-dimensional coordinates where the task to be recognized is located. The invention utilizes the brain-computer interface for multi-person cooperative decision making, can realize the identification of a plurality of tasks through the coordinates of a plurality of dimensions, has high identification speed and breaks through the limitation of the traditional paradigm on the number of identifiable tasks.

Description

Multi-user multi-dimensional recognition model collaborative decision making system and method
Technical Field
The invention relates to the technical field of brain science and cognitive science, in particular to a multi-user multi-dimensional recognition model collaborative decision system and a method.
Background
Most of the existing identification mechanisms are identification of a simple plane, and tasks to be identified which are set on the simple plane corresponding to coordinate nodes can be obtained by identifying whether a target is in row/column coordinates on the plane. The recognition mechanism is simple in structure, can only adapt to recognition of limited candidates in a limited space by limited number of people, cannot be applied to scenes of more recognition objects in multi-dimension by multiple people, and is limited in recognition decision quantity, low in recognition decision efficiency and poor in recognition decision effect under the condition of more recognition tasks or multi-decision objects.
In fact, in the field of brain-computer interfaces, the multitasking problem is quite common. In the past, brain-computer interface technology was used primarily in the medical rehabilitation field, with operator first-entry being limited to one person, i.e., a particular patient. Therefore, the solution to the multi-tasking problem is also limited to a single person framework. However, in recent years, the field of artificial intelligence has a voice to indicate that 'true artificial intelligence can be realized only by human-computer fusion', which arouses the wide attention of experts and scholars at sea and abroad. Brain-controlled unmanned vehicles, brain-controlled unmanned aerial vehicles, and even brain-controlled multi-robot formations appear in succession. With the development of these platforms, the complexity of the brain-computer interface control instructions is more and more demanding. This also leaves the brain-computer interface technology completely free of speed and instruction count deficiencies. In the meantime, the brain-computer interface is greatly stepped towards the man-computer hybrid intelligence direction, and a method for realizing multi-person cooperation and multi-instruction parallel is realized by a single person or even multiple persons and multiple computers, so that the method is a good scheme for realizing higher-level man-computer hybrid intelligence.
In fact, the machine intelligence itself originates from, and serves, humans. Therefore, talents are the core of human-computer hybrid intelligence. The human brain as the core processor is very similar to the computer to the human-computer hybrid intelligent system as the CPU. In 1971, intel introduced the first global microprocessor 4004, which was composed of 2300 transistors. The larger the number of transistors integrated on one chip, the higher the dominant frequency, the faster the operation speed. In 2004, Pentium 4570 single-core processor released by Intel has a dominant frequency of 3.8GHz, i.e., 38 hundred million operations per second, which is about 5135 times of 74KHz in the year. However, at higher frequencies, higher voltages are required to increase the rate at which the capacitor charges and discharges. Scientists have come to appreciate that the frequency cannot be increased indefinitely under the constraints of temperature, power consumption, etc. When the frequency cannot be increased significantly, scientists have attempted to increase the performance of single-core CPUs to some extent by reducing the clock cycles, instruction pipelines, etc. required to execute a single instruction. However, in essence, a single-core CPU can only achieve concurrency through a serial algorithm, and cannot really achieve synchronous execution of multiple tasks. Until 2005, intel released Pentium Extreme Edition 840 dual-core CPUs first, and after several weeks, AMD also released the Athlon 64X 2 series, thus formally pulling the preface of multi-core CPUs open. The multiple operation cores of the multi-core CPU can simultaneously execute multiple tasks on different threads in the same time slice, so that the simultaneous execution of multiple tasks is realized in the true sense. In 2006, intel released core 2 dual-core CPUs, which improved performance by 40% compared to the best processors before intel corporation, but reduced power consumption by 40% instead.
Since, multi-core CPUs can bring historical changes to computer performance. Then, do increasing the number of brains that also serve as processing cores bring a breakthrough to the performance of the brain-computer interface system? The answer is considered positive. Currently, computer-dependent brain-computer interface technologies are temporarily unable to mimic the same communication patterns as brain nerves. The induced brain-computer interface technology is limited by the physiological sense resolution of human body, such as vision, hearing, etc., and the space for single person to increase speed is very limited. When the stimulation frequency is too fast to be resolved by a person, it is considered to be co-occuring, and the stimulation paradigm is disabled. Even the spontaneous brain-computer interface technology is limited by the inherent property that digital signals in a computer are discrete, and the single brain-computer interface technology cannot always make the brain control external equipment as well as control the body of the user. Although, considerable research into single-man brain-computer interface technology is currently being pursued by researchers. However, similar to a single core CPU, a single brain-computer interface faces the multitasking problem and is still solved by serial or concurrent in nature. Moreover, contradictions between the recognition speed, the recognition accuracy and the recognition quantity always exist, and the improvement of one index always causes the reduction of the other index, which also causes the overall performance of the system to be incapable of being greatly improved. In this mode, the performances of the man and the machine are highly mismatched, and the man and the machine just like stepping on an accelerator and a brake at the same time, the real power of man-machine mixed intelligence cannot be exerted at all. Therefore, the invention provides a novel multi-person BCI framework, and provides a mechanism for decomposing an instruction into a plurality of tasks and completing the tasks by a plurality of persons in parallel.
Disclosure of Invention
The invention provides a multi-user multi-dimensional recognition model collaborative decision-making system and a multi-user multi-dimensional recognition model collaborative decision-making method, and aims to solve the technical problems of limited number of recognition decisions, low efficiency and poor effect in multi-task recognition or multi-decision-making objects in the background art.
In order to achieve the above object, the multi-user multi-dimensional recognition model collaborative decision-making system provided by the embodiment of the present invention includes a multi-dimensional recognition model and a calculation processing module, wherein the multi-dimensional recognition model is a model obtained by combining MDAbstract layout of candidate tasks to an inclusion MDIn the D-dimensional structure space of each node, K decision makers identify tasks to be identified in the D-dimensional structure space, and each decision maker is responsible for M tasks including MD-1The D-1 dimensional subspace block of each alternative task identifies the D-1 dimensional subspace block in which the task to be identified is positioned, and the calculation processing module obtains the multidimensional coordinate (C) in which the task to be identified is positioned by calculating the identification results of all decision makers1,C2,...CD-1,CD) All or a part of the values of (a) to obtain a recognition result or a recognition area.
Preferably, the identifying in which D-1 dimensional subspace block the task to be identified is specifically: and M normal form interfaces are displayed for each decision maker, the normal form interfaces are two-dimensional display screens containing MD-1 alternative tasks, and when the task to be identified is in the displayed two-dimensional display screens, the decision maker feeds back the identification result to the calculation processing module.
Preferably, the obtained identification result or the identification area is MDThe alternative tasks are continuously arranged according to D dimensions, and the coordinate (C) corresponding to the coordinate of each task to be identified1,C2,...CD-1,CD) The index value of (d) may be expressed as:
Figure BDA0002831196920000031
the calculation processing module can obtain the corresponding coordinate (C) through the index value1,C2,...CD-1,CD) And the corresponding alternative task is the task to be identified.
Preferably, the two-dimensional display screen flashing mechanism is specifically: each stimulus of the paradigm is M contained in a random subspaceD-1The individual targets are highlighted by crosses; when the number of participants is the same as the dimension of the abstract space, the stimulation can be simplified to the fact that a cross appears at the center of each subspace block, and after each object is tested to find the target to be recognized, only the center of the subspace block where the target is located needs to be observed.
Preferably, the two-dimensional display screen flashing time is specifically: the order of stimulation flashes appeared as a list of 1 × L, L being a code length of each tested stimulation paradigm, and can be expressed as the equation:
Figure BDA0002831196920000032
the elements in the list are called code words and are non-repeated pseudo-random codes consisting of all integers in the range of [0, L-1 ]; the highlight time and the extinguishing time of each flicker are collectively called as stimulus generation asynchronism and are represented by S, L times of stimulus flicker in a list are defined as a trial, the number of the trial required by each recognition is represented by R, the number of times of each task which needs to be accepted by each test is selected as L R, an interval time is given between two task selections for the test rest, result processing and task prompt, the interval time is defined as I, and the time consumption T of each recognition is the time of the flicker times multiplied by the time consumption of each flicker and added with the interval between tasks, which is represented by the equation:
T＝L*R*S+I。
the invention provides a method for applying a multi-user multi-dimensional recognition model collaborative decision-making system, which comprises the following steps:
step S1, decision making preparation work by multi-computer interface voting: multiple testees wear electrodes, alternative tasks are identified in front of M paradigm interface screens, and the paradigm interface comprises MD-1A two-dimensional display screen of the alternative tasks;
step S2, acquiring the task to be identified, and generating a task list: according to MDGenerating tasks to be identified by the alternative tasks and the content needing decision, wherein the tasks to be identified form a task list;
step S3, decision making by voting through a brain-computer interface: a plurality of testees wear electrodes, tasks to be identified are displayed in M normal mode interface screens, the plurality of testees watch target characters according to the prompts of the tasks to be identified, and the multiple testees respectively acquire a D-1 dimensional subspace block in which the tasks to be identified are located;
step S4, obtaining a final result by calculation: and obtaining all values or partial values of the multi-dimensional coordinates where the tasks to be identified are located by calculating the identification results of all decision makers so as to obtain the identification results or the identification areas.
The technical effects which can be achieved by adopting the invention are as follows:
the multi-person collaborative decision-making technical scheme provided by the invention is used for identifying multiple tasks or making decisions on multiple objects, and has the advantages of high identification decision quantity, high efficiency and good effect.
Drawings
FIG. 1 is a schematic structural diagram of a D-dimensional structure space of a collaborative decision system for a multi-user multi-dimensional recognition model according to an embodiment of the present invention;
FIG. 2 is a diagram showing the probability P of the occurrence of adjacent target stimuli in the multi-person multi-dimensional recognition model collaborative decision system according to the present inventionneiA relation diagram of d and a normal form interface number M;
fig. 3 is a schematic diagram of a variable relationship between the task arrangement number N and the number K of decision makers and the number M of canonical form interfaces in the multi-user multi-dimensional recognition model collaborative decision making system of the present invention.
Detailed Description
In order to make the technical problems, technical solutions and advantages of the present invention more apparent, the following detailed description is given with reference to the accompanying drawings and specific embodiments.
Aiming at the existing problems, the invention provides a multi-user multi-dimensional recognition model collaborative decision-making system which comprises a multi-dimensional recognition model and a calculation processing module, wherein the multi-dimensional recognition model is a model obtained by combining MDAbstract layout of candidate tasks into a containing MDIn the D-dimensional structure space of the nodes, 6 are arranged as shown in FIG. 13Abstract layout of alternative tasks to one containing 63In the 3-dimensional structure space of each node, K decision makers identify tasks to be identified in the D-dimensional structure space, and each decision maker is responsible for M tasks including MD-1The D-1 dimensional subspace block of each alternative task identifies the D-1 dimensional subspace block in which the task to be identified is positioned, and the calculation processing module obtains the multidimensional coordinate (C) in which the task to be identified is positioned by calculating the identification results of all decision makers1,C2,...CD-1,CD) All or a part of the values of (a) to obtain a recognition result or a recognition area. When K is D, all values of the multi-dimensional coordinates where the task to be identified is located can be obtained.
The identification of the D-1 dimensional subspace block in which the task to be identified is specifically as follows: and M normal form interfaces are displayed for each decision maker, the normal form interfaces are two-dimensional display screens containing MD-1 alternative tasks, and when the task to be identified is in the displayed two-dimensional display screens, the decision maker feeds back the identification result to the calculation processing module.
The obtained identification result or identification region is specifically the MDThe alternative tasks are continuously arranged according to D dimensions, and the coordinate (C) corresponding to the coordinate of each task to be identified1,C2,...CD-1,CD) The index value of (d) may be expressed as:
Figure BDA0002831196920000051
the calculation processing module can obtain the corresponding coordinate (C) through the index value1,C2,...CD-1,CD) And the corresponding alternative task is the task to be identified.
If the two-dimensional situation is adopted, D is 2, and C1 is 3 if the target coordinate obtained by the first decision maker is 3; the second decision maker obtains the target coordinate of 2, and C2 is 2. The index value of the task identified by the decision maker is 6 ° × 3+6 × 2 ═ 15. The task is 15 tasks from left to right and from top to bottom. For example, the first row is 0-5, the second row is 6-11; the third row is 12-17:
then, in the process that the decision maker obtains the identification result or the identification area by identifying the flicker screen, the number of rows and columns of the identification target can be obtained by the sequence number of the flicker screen, so that the corresponding index value can be obtained according to the number of rows and columns, and the target to be identified is determined.
In the high-dimensional theory, the tasks are numbered according to the index number sequence of the formula during layout. And then, each decision maker can directly calculate an index value through the formula according to the sequence number of the flicker screen after obtaining the result, and can know the target task to be identified corresponding to the index value.
The positioning of the coordinates by reducing the subspace position of the dimension specifically comprises: when a decision maker sees the target stimulation, a peak appears in the electroencephalogram signal, and by detecting the peak, the decision maker can know which time of flicker induces the peak, namely the sequence number of a flicker screen.
For an N-dimensional space, the coordinates in the Nth dimension can be known by looking at the several N-1-dimensional subspace evoked. For the subspace with the dimension of N-1, when the screen is displayed, all elements in the subspace with the dimension of N-1 are pulled into a line, or are spread into a plane, or are arranged into a cube for displaying. Each stimulus is given only one over the entire N-1 dimensional subspace. Thus, the target stimulation induced by the N-1 dimensional subspace can be known, and the coordinate of the Nth dimension can be known.
The invention provides a method for applying a multi-user multi-dimensional recognition model collaborative decision-making system, which comprises the following steps:
step S1, decision making preparation work by multi-computer interface voting: multiple testees wear electrodes to identify alternative tasks in front of M range type interface screensThe canonical interface is MD-1A two-dimensional display screen of each alternative task;
step S2, acquiring the task to be identified, and generating a task list: according to MDGenerating tasks to be identified by the alternative tasks and the content needing decision, wherein the tasks to be identified form a task list;
step S3, decision making by voting through a brain-computer interface: a plurality of testees wear electrodes, tasks to be identified are displayed in M normal mode interface screens, the plurality of testees watch target characters according to the prompts of the tasks to be identified, and the multiple testees respectively acquire a D-1 dimensional subspace block in which the tasks to be identified are located;
step S4, obtaining a final result by calculation: and obtaining all values or partial values of the multi-dimensional coordinates where the tasks to be identified are located by calculating the identification results of all decision makers so as to obtain the identification results or the identification areas.
The two-dimensional display screen flickering mechanism specifically comprises the following steps: each stimulus of the paradigm is M contained in a random subspaceD -1Each target is highlighted by a cross. When the number of participants is the same as the dimension of the abstract space, the stimulation can be simplified to the fact that a cross appears at the center of each subspace block, and after each object is tested to find the target to be recognized, only the center of the subspace block where the target is located needs to be observed.
The two-dimensional display screen flashing time specifically comprises the following steps: the order of stimulation flashes appeared as a list of 1 × L, L being a code length of each tested stimulation paradigm, and can be expressed as the equation:
Figure BDA0002831196920000071
the elements in the list are called code words and are non-repeated pseudo-random codes consisting of all integers in the range of [0, L-1 ]; the highlight time and the extinguishing time of each flicker are collectively called as stimulus generation asynchronism and are represented by S, L times of stimulus flicker in a list are defined as a trial, the number of the trial required by each recognition is represented by R, the number of times of each task which needs to be accepted by each test is selected as L R, an interval time is given between two task selections for the test rest, result processing and task prompt, the interval time is defined as I, and the time consumption T of each recognition is the time of the flicker times multiplied by the time consumption of each flicker and added with the interval between tasks, which is represented by the equation:
T＝L*R*S+I。
the multi-user multi-dimensional recognition model collaborative decision-making system provided by the invention is used for recognizing multiple tasks or making decisions on multiple objects, and has the advantages of high recognition decision-making quantity, high efficiency and good effect. Can identify multiple tasks, and identify MDAnd multiple persons make decisions for the decision alternative tasks to be identified, each decision maker decides to identify the dimensional subspace block where the task to be identified is located, the index value corresponding to the task to be identified can be calculated according to the logic, and the corresponding task to be identified can be found according to the index value. Therefore, the identification decision can be carried out by multiple persons from a plurality of alternative tasks, the identification speed of the identification decision can be increased, and the identification accuracy is high.
The method is evaluated from three aspects of identification speed, identification accuracy and identification quantity.
First, identify the speed
When K is 1, the system is identified by one person, and the code length L of the system is1Is composed of
L1＝M*D
In the same way contain MDUnder the abstract layout of D-dimensional space of each node, the time consumption T of each identification1Can be expressed as: t is1＝M*D*R*S+I
The efficiency improvement factor KT of a multi-person system compared to a single-person system can be expressed as
Figure BDA0002831196920000072
The analytical formula may rewrite the expression of the efficiency improvement factor KT as:
Figure BDA0002831196920000081
as shown in the above formula, the efficiency improvement factor KT is necessarily greater than 1 as long as K is greater than 1. Then it can be concluded that: on the premise of the same layout of tasks, the recognition efficiency of the multi-person system is higher than that of the single-person system. If the identification interval I is not considered, the multi-person system in which the K persons participate can improve the identification efficiency to K times that of a single-person system.
Second, recognition accuracy
First, from overlapping interference that temporally adjacent target stimuli may cause ERP (event-related potential).
In the M possible coordinate values in each dimension, one coordinate value is the target coordinate value in the dimension. When the adjacent code words in time are coordinate values of the targets in different dimensions, that is, when two target stimuli occur continuously, the ERP signals overlap, thereby affecting the classification result.
In the flash of a three, there is a probability P that the target stimulus appears continuouslyneiCan be achieved by the probability that all target stimuli are not continuously presented
Figure BDA0002831196920000082
Calculated by the indirect formula
Figure BDA0002831196920000083
For events where all target stimuli do not occur continuously, the calculation can be performed by interpolation. For a multi-person system with K participants, the number of dimensions d that each decision maker needs to be responsible for can be expressed as:
Figure BDA0002831196920000084
this means that there are d target stimuli in the L code words for each decision maker. The rest L-d non-target stimuli are arranged in a whole way
Figure BDA0002831196920000085
Seed events, producing L-d +1 vacancies. Insertion of d target stimuli into L-d +1 vacancies can generate
Figure BDA0002831196920000086
And (4) sowing the events. The probability that all target stimuli are not continuous can be determined by
Figure BDA0002831196920000087
The probability P of continuous occurrence of the target stimulus is obtained by combining the formulasneiCan be expressed as:
Figure BDA0002831196920000091
reduced to two independent variables
Figure BDA0002831196920000092
When d and M both belong to [2,10 ]]When the number is an integer of (1). Taking independent variables as M and d and dependent variable as PneiDrawing a three-dimensional graph is shown in fig. 2.
As can be observed from FIG. 2, when d and M both belong to [2,10 ]]When an integer of (1) is PneiIncreasing with increasing M and decreasing with increasing d. And d is inversely proportional to the number of people K, that is, in this case, the probability of the time error rate increases as the number of people increases. However, when the number of people K increases to be equal to the spatial dimension D, i.e., D is 1, since each decision maker is responsible for processing only one target stimulus, there is no adjacent problem, and the probability of the time error rate is suddenly reduced to 0. It can thus be concluded that: when the number of participating people of the multi-people cooperative system is equal to the dimension D of the task layout space, the system is optimal, and the probability of the time error rate is 0. Otherwise there will be a time error rate, and the error rate increases with the number of people.
(II) analysis from spatially adjacent stimuli that can lead to false positives and interference with VEP (visual evoked potential) components
Since the case where M is less than 3 is not fully described, it is analyzed as M.gtoreq.3.
For a two-dimensional space:
tasks falling on the top corner are respectively 1 on two lines connected with the tasks, namely 2 adjacent tasks are easy to cause wrong division;
the tasks falling on the sideline comprise 3 adjacent tasks;
the task that falls on the midplane has 4 adjacent tasks.
For a three-dimensional space:
the tasks falling on the top corner are respectively 1 on three lines connected with the tasks, namely 3 adjacent tasks are easy to be wrongly divided;
the tasks falling on the sideline comprise 4 adjacent tasks;
the task on the surface has 5 adjacent tasks;
there are 6 adjacent tasks for the task that falls inside the three-dimensional space.
Deductive reasoning to the general case, for each decision maker that needs to identify d dimensions:
tasks falling on top corners (which can be understood as 0 dimension), only one adjacent task exists for each dimension, namely d adjacent tasks are easy to cause wrong division;
a task falling on a line (which can be understood as a 1-dimensional task) has d +1 adjacent tasks;
a task that falls on a surface (which can be understood as 2-dimensional), there are d +2 adjacent tasks;
……
tasks that fall in the center of the d-dimensional space have a maximum of 2 x d adjacent tasks.
In summary, if a single person wants to realize high-dimensional encoding, K is 1 and D is D. There are at most 2D neighboring tasks that tend to cause spatial misclassification. If D is 1 when the number of people K is equal to the dimension D of the abstract space, the number of adjacent tasks that are prone to spatial misinterpretation is at most 2. The more people that can be analyzed, the fewer dimensions d each person needs to be responsible for, and the fewer number of adjacent tasks 2 x d that are prone to spatial misclassification. It can thus be concluded that: the multi-person cooperation BCI system can reduce the spatial error rate, thereby improving the accuracy of the system.
Thirdly, identifying the number
Compared with the traditional two-dimensional recognition mechanism, the traditional two-dimensional interface can arrange the number of tasks N on the premise of the same number of people KoldCan be expressed as
Nold＝M2·K
Comparing the number of participating people K of the multi-person collaborative BCI system with the abstract space dimension D, the maximum number N of arrangement can be realized by high-dimensional codingnewCan be expressed as
Nnew＝MK
A graph is shown in FIG. 3 with K and M as independent variables and the number of task assignments as dependent variables.
The latticed surface is a task quantity curved surface of the two-dimensional code, and the pure color surface is a task quantity curved surface of the high-dimensional code. It can be seen from the figure that when the number K is between 2 and 3, the solid curved surface exceeds the lattice-shaped surface. And, only when the number of people K increases to 4, the arrangement number of the red curved surfaces exceeds 6000, and the lattice surfaces are not changed much compared with before.
The high-dimensional coding improves the original order of magnitude of the number multiplication operation to the order of magnitude of the power, and greatly improves the quantity of programmable tasks.
While the foregoing is directed to the preferred embodiment of the present invention, it will be understood by those skilled in the art that various changes and modifications may be made without departing from the spirit and scope of the invention as defined in the appended claims.

Claims (5)

1. The multi-user multi-dimensional recognition model collaborative decision-making system is characterized by comprising a multi-dimensional recognition model and a calculation processing module, wherein the multi-dimensional recognition model is a model obtained by combining MDAbstract layout of candidate tasks to an inclusion MDIn the D-dimensional structure space of each node, K decision makers identify tasks to be identified in the D-dimensional structure space, and each decision maker is responsible for M tasks including MD-1The D-1 dimensional subspace block of each alternative task identifies the D-1 dimensional subspace block in which the task to be identified is positioned, and the calculation processing module obtains the multidimensional coordinate (C) in which the task to be identified is positioned by calculating the identification results of all decision makers1,C2,...CD-1,CD) To obtain a recognition result or a recognition area, where the recognition of the task to be recognized in which D-1 dimensional subspace block is specifically: displayed to each decision maker are M paradigm interfaces which are M-inclusiveD-1When the task to be identified is in the displayed two-dimensional display screen, the decision maker feeds back the identification result to the calculation processing module, wherein D and M both represent natural numbers larger than 1, and CiCoordinate values representing each dimension, i ═ 1,2, … D.
2. The multi-user multi-dimensional recognition model decision-making system as claimed in claim 1, wherein the obtained recognition result or the recognition region is MDThe alternative tasks are continuously arranged according to D dimensions, and the coordinate (C) corresponding to the coordinate of each task to be identified1,C2,...CD-1,CD) The index value of (d) may be expressed as:
Figure FDA0003639161750000011
the calculation processing module can obtain the corresponding coordinate (C) through the index value1,C2,...CD-1,CD) And the corresponding alternative task is the task to be identified.
3. The multi-person multi-dimensional recognition model collaborative decision making system according to claim 2, wherein the two-dimensional display screen flashing mechanism is specifically: each stimulus of the paradigm is M contained in a random subspaceD-1The individual targets are highlighted by crosses; when the number of participants is the same as the dimension of the abstract space, the stimulation can be simplified to the fact that a cross appears at the center of each subspace block, and after each object is tested to find the target to be recognized, only the center of the subspace block where the target is located needs to be observed.
4. The multi-person multi-dimensional recognition model collaborative decision making system according to claim 3, wherein the two-dimensional display screen blinking time is specifically: the order of stimulation flashes appears as a list of 1 × L, L being a code length of the stimulation paradigm for each decision maker, and can be expressed as the equation:
Figure FDA0003639161750000021
k represents the number of decision makers, elements in the list are called code words and are non-repeated pseudo-random codes consisting of all integers in the range of [0, L-1 ]; the highlight time and the extinguishing time of each flashing are called as stimulus generation asynchronism and are represented by S, L times of stimulus flashing in a list is defined as a trial, the number of the trials required by each recognition is represented by R, the number of times required to be accepted by each decision maker is selected by each task to be L x R, a period of time is given between two task selections for the decision maker to have a rest, result processing and task prompting, the period of time is defined as I, and the time consumed for each recognition is represented by the time spent on flashing multiplied by the time spent on each flashing and added with the interval between tasks, and is represented as an equation:
T＝L*R*S+I。
5. a method for applying the multi-person multi-dimensional recognition model collaborative decision-making system according to any one of claims 1 to 4, comprising the steps of:
step S1, decision making preparation work by multi-computer interface voting: multiple testees wear electrodes, and alternative tasks are identified in front of M paradigm interface screens, wherein the paradigm interface comprises MD-1Two-dimensional display screen of alternative tasks, wherein D and MAll represent natural numbers greater than 1;
step S2, acquiring the task to be identified, and generating a task list: according to MDGenerating tasks to be identified by the alternative tasks and the content needing decision, and forming a task list by the tasks to be identified;
step S3, decision making by voting through a brain-computer interface: a plurality of testees wear electrodes, tasks to be identified are displayed in M normal mode interface screens, the plurality of testees watch target characters according to the prompts of the tasks to be identified, and the multiple testees respectively acquire a D-1 dimensional subspace block in which the tasks to be identified are located;
step S4, obtaining a final result by calculation: and obtaining all values or partial values of the multi-dimensional coordinates where the tasks to be identified are located by calculating the identification results of all decision makers so as to obtain the identification results or the identification areas.
CN202011447933.7A 2020-12-11 2020-12-11 Multi-user multi-dimensional recognition model collaborative decision making system and method Active CN112433616B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202011447933.7A CN112433616B (en) 2020-12-11 2020-12-11 Multi-user multi-dimensional recognition model collaborative decision making system and method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202011447933.7A CN112433616B (en) 2020-12-11 2020-12-11 Multi-user multi-dimensional recognition model collaborative decision making system and method

Publications (2)

Publication Number Publication Date
CN112433616A CN112433616A (en) 2021-03-02
CN112433616B true CN112433616B (en) 2022-06-14

Family

ID=74691532

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202011447933.7A Active CN112433616B (en) 2020-12-11 2020-12-11 Multi-user multi-dimensional recognition model collaborative decision making system and method

Country Status (1)

Country Link
CN (1) CN112433616B (en)

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101219048A (en) * 2008-01-25 2008-07-16 北京工业大学 Method for extracting brain electrical character of imagine movement of single side podosoma
CN106650793A (en) * 2016-11-25 2017-05-10 天津津航计算技术研究所 Method of controlling PWM (pulse-width modulation) duty cycle based on brain-computer interface
CN106774847A (en) * 2016-11-24 2017-05-31 天津大学 3D vision P300 Speller systems based on virtual reality technology
CN108960299A (en) * 2018-06-15 2018-12-07 东华大学 A kind of recognition methods of multiclass Mental imagery EEG signals

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US10945864B2 (en) * 2016-08-17 2021-03-16 Teledyne Scientific & Imaging, Llc System and method for noninvasive identification of cognitive and behavioral goals

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101219048A (en) * 2008-01-25 2008-07-16 北京工业大学 Method for extracting brain electrical character of imagine movement of single side podosoma
CN106774847A (en) * 2016-11-24 2017-05-31 天津大学 3D vision P300 Speller systems based on virtual reality technology
CN106650793A (en) * 2016-11-25 2017-05-10 天津津航计算技术研究所 Method of controlling PWM (pulse-width modulation) duty cycle based on brain-computer interface
CN108960299A (en) * 2018-06-15 2018-12-07 东华大学 A kind of recognition methods of multiclass Mental imagery EEG signals

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
基于脑电和生理信号的多模态情感分类方法研究;鞠思航;《中国优秀硕士学位论文全文数据库（电子期刊）》;20200115(第1期);I136-214 *

Also Published As

Publication number Publication date
CN112433616A (en) 2021-03-02

Similar Documents

Publication Publication Date Title
Bang et al. Spatio-spectral feature representation for motor imagery classification using convolutional neural networks
Song et al. Graph-embedded convolutional neural network for image-based EEG emotion recognition
Ai et al. Feature extraction of four-class motor imagery EEG signals based on functional brain network
Esfahani et al. Classification of primitive shapes using brain–computer interfaces
CN107126193A (en) Based on the adaptively selected multivariable Causality Analysis Approach of lag order
CN109684959A (en) The recognition methods of video gesture based on Face Detection and deep learning and device
CN108021232B (en) A kind of decoded method and apparatus of cerebral cortex electric signal
CN110377049B (en) Brain-computer interface-based unmanned aerial vehicle cluster formation reconfiguration control method
CN110263606A (en) Scalp brain electrical feature based on end-to-end convolutional neural networks extracts classification method
Tian et al. A hierarchical classification method for automatic sleep scoring using multiscale entropy features and proportion information of sleep architecture
CN105094324A (en) Brain state recognition method based on electroencephalogram generated from left and right hand motor imagery
Wei et al. Image feature extraction and object recognition based on vision neural mechanism
CN102222231B (en) Visual attention information computing device based on guidance of dorsal pathway and processing method thereof
Gu et al. Frame-level teacher-student learning with data privacy for eeg emotion recognition
Kauppi et al. Decoding magnetoencephalographic rhythmic activity using spectrospatial information
CN112433616B (en) Multi-user multi-dimensional recognition model collaborative decision making system and method
CN114145745A (en) Multi-task self-supervision emotion recognition method based on graph
Shen et al. Exemplar-based age progression prediction in children faces
Lian et al. A novel time-incremental end-to-end shared neural network with attention-based feature fusion for multiclass motor imagery recognition
Fu et al. Gesture based fear recognition using nonperformance dataset from VR horror games
Xu et al. Decode brain system: A dynamic adaptive convolutional quorum voting approach for variable-length EEG data
Li et al. Key band image sequences and a hybrid deep neural network for recognition of motor imagery EEG
Wang et al. The quantitative application of channel importance in movement intention decoding
Tsotsos et al. Different binding strategies for the different stages of visual recognition
Zhang et al. Face Forgery Detection via Multi-Feature Fusion and Local Enhancement

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant