CN103530505B - Human brain language cognition modeling method - Google Patents

Abstract

The invention discloses a human brain language cognition modeling method. The human brain language cognition modeling method comprises the following steps of initialization of a cognitive state example, mapping of probability distribution between activation characteristics and observation data, definition of a brain tacit cognition model and parameter analysis of the tacit cognition model. In the cognition modeling process, input stimulation, the observation result and the tacit cognition state are defined as triple time sequences related to a dynamic event, namely a cognition stimulation task time sequence, an observation characteristic time sequence and a tacit cognition state time sequence, the triple time sequences are related to one another through a set of probability distribution, and not all collected brain data are treated as static information for statistics. Therefore, the human brain language cognition modeling method does not need to meet the basic assumption based on statistics, is established under a small sample data condition, and guarantees the correctness of the cognition analysis result, thereby achieving cognition modeling under the small sample data condition. The human brain language cognition modeling method improves the accuracy of cognition modeling, and provides an effective approach for complex cognition analysis.

Description

A kind of human brain language cognition modeling method

Technical field

The present invention relates to a kind of human brain language cognition technology, particularly a kind of human brain language cognition modeling method.

Background technology

It is brain area activation or inhibition level that cognition the most directly reflects in human brain, from the angle of nuroinformatics, to language The key of speech Cognitive Study is the process of research human brain language cognition process Midbrain Area state and these state transitions, and it is the heaviest Want the requirement contrived experiment that one of means are according to nuroinformatics, stimulate recognizing of the experimenter volunteer of special group (mostly) Know function, and feature gathered data is carried out to it, then using certain modeling technique, these data are analyzed.In recent years The brain imaging technique (positron emission computerized tomography, function nuclear magnetic resonance, NMR) growing up is that people directly observe and analysis human brain Functional activity provides means.The Brian Imaging analytical technology that present cognitive institute relies on mainly passes through Statistical Parametric Mapping system System, functional neurosurgery Image analysis system etc. are based on statistical modeling tool generation corresponding t inspection mind map and are found which brain Area controls a certain cognitive function.These cognitive modeling methods based on counting are to sample data based on a basic assumption Carry out statistical analysiss, this supposition is thought：Given certain cognitive stimulation, must the cognition of this stimulation in human brain cognition State.And it is true that the cognitive activation feature observed has event correlation, cognitive state has additivity data sample Less feature, does not possess strict statistical significance, is unsatisfactory for counting the premise of cognitive modeling method yet, thus is easily caused mistake Cognitive result by mistake.

Content of the invention

For solving the problems referred to above that prior art exists, the present invention will design one kind and be suitable to small sample, has dynamic time The human brain language cognition modeling method of feature.

To achieve these goals, technical scheme is as follows：A kind of human brain language cognition modeling method, including with Lower step：

A, the initialization of cognitive state example

Identify the cognitive state in human brain with h, because h can not be directly observed with instrument and equipment, thus h is referred to as Recessive cognitive state, recessive cognitive state h is expressed as four-tuple<W,d,Θ,Ω>, wherein：D represents h continuing in time Scalar；W is the reflection square that Spatial dimensionality is d × V, the brain area dimension that wherein V is related to for h, and its Spatial Dimension component is v；Θ is h Parameter space, Ω be parameter space codomain；In cognitive observation process, an input stimulus Δ, the activation meeting of brain cognitive state There is certain time lag；To the cognitive state being in activated state, it is cognitive state example, with ξ labelling it is considered to recessive cognition shape The time factor of state h, cognitive state example is a tlv triple<h,λ,o>, wherein：H is recessive cognitive state accordingly, and λ is The input time point that experiment stimulates, o is the time migration of cognitive state, and the initial time of cognitive state example ξ is λ+o, and it is held Continuous time d is identical with corresponding recessiveness cognitive state h duration, all uses d (h) to represent；

Probability distribution between B, mapping activation feature and observation data

It is assumed that in cognitive experimentation, the t period activates feature b by the brain that experimenter observed by imaging device_tIt is by L The coefficient result of cognitive state example, uses ξ₁,ξ₂,…,ξ_LIndicate this L cognitive state example, then note b_t={ ξ₁, ξ₂,…,ξ_L}；b_tComponent s with period t, Spatial Dimension component v in observation time sequence S_tvNormal Distribution, that is,：

s_tv～N (μ_tv(b_t),σ_v) (1)

Wherein σ_vIt is standard variance, it reflects the noise profile related to observation time sequence, has time independence；μ_tv (b_t) reflection all related cognitive state of corresponding period Overlay, that is, meet following formula：

${\mathrm{\&mu;}}_{tv}\left(b_t\right)=\underset{\mathrm{\&xi;}\&Element;b_t}{\mathrm{\&Sigma;}}\stackrel{d\left(h\right(\mathrm{\&xi;}\left)\right)}{\mathrm{\&Sigma;}}\mathrm{\&delta;}\left(\mathrm{\&lambda;}\right(\mathrm{\&xi;})+o(\mathrm{\&xi;})=t-\mathrm{\&tau;})w_{\mathrm{\&tau;}v}\left(h\right(\mathrm{\&xi;}\left)\right)---\left(2\right)$

Wherein δ (.) is indicator function, if independent variable is true, this function result is 1, otherwise for 0；H (ξ) mark is cognitive Stateful example ξ corresponding recessiveness cognitive state；w_τv(h (ξ)) is then that on the τ time step of mark period t, recessive cognitive state is anti- Reflect the v component of square W (h (ξ)).

C, definition brain low-profile cognitive model

Define brain low-profile cognitive model and be one hexa-atomic group<H,Φ,I,S,Ξ,Γ>, use HCM labelling, wherein：H is recessive Cognitive state set；Φ is the parameter vector of low-profile cognitive model, and they rely on experimental design parameter, that is,：The type, defeated stimulating Enter the time point of stimulation；S is observation time sequence；The stimulus sequence that Ξ is made up of input stimulus Δ；I is cognitive state example Time serieses, because the activationary time point of cognitive state example is corresponding with observation time sequence, thus by the element b in I_t It is abbreviated as b；Γ is<σ₁,σ₂,…,σ_v>Variance collection, wherein σ_vIt is the standard variance in formula (1), the noise of reflecting time sequence Distribution；HCM defines I and S, Ξ constitute triple time serieses of probability correlation,Stochastic variable defined in I, then observe The probability distribution that activates between feature and recessive cognitive state of brain be：

$P\left(S\right|HCM,\mathrm{\&Delta;})=\underset{b\&Element;I}{\mathrm{\&Sigma;}}P\left(S\right|HCM,\hat{I}=b)P(\hat{I}=b|HCM,\mathrm{\&Delta;})---\left(3\right)$

Wherein,

$P(\hat{I}=b|HCM,\mathrm{\&Delta;})=\frac{\underset{\mathrm{\&xi;}\&Element;b}{\mathrm{\&Pi;}}P\left(h\right(\mathrm{\&xi;}\left)\right|HCM,\mathrm{\&Delta;})P\left(o\right(\mathrm{\&xi;}\left)\right|h\left(\mathrm{\&xi;}\right),HCM,\mathrm{\&Delta;})}{\underset{b^{,}\&Element;I}{\mathrm{\&Sigma;}}\underset{{\mathrm{\&xi;}}^{,}\&Element;b^{,}}{\mathrm{\&Pi;}}P\left(h\right({\mathrm{\&xi;}}^{,}\left)\right|HCM,\mathrm{\&Delta;})P\left(o\right({\mathrm{\&xi;}}^{,}\left)\right|h\left({\mathrm{\&xi;}}^{,}\right),HCM,\mathrm{\&Delta;})}---\left(4\right)$

In formula, and P (h (ξ) | HCM, Δ) it is that Chinese Cognition experimentation conditional stimulates as the recessive cognitive state h of Δ activation And the conditional probability of corresponding cognitive state example ξ, and P (o (ξ) | h (ξ), HCM, Δ) it is corresponding time lag conditional probability；Low-profile is recognized The modeling process of perception model needs to determine its conditional probability for specific cognitive process, and low-profile cognitive model is also predefined simultaneously Parameter vector, maximal possibility estimation is carried out to HCM parameters；For HCM, had according to Bayes theorem：

$P(\hat{I}=b|S,HCM,\mathrm{\&Delta;})=\frac{P\left(S\right|\hat{I}=b,HCM)P(\hat{I}=b|HCM,\mathrm{\&Delta;})}{\underset{b^{\&prime;}\&Element;I}{\mathrm{\&Sigma;}}P\left(S\right|\hat{I}=b^{\&prime;},HCM)P(\hat{I}=b^{\&prime;}|HCM,\mathrm{\&Delta;})}---\left(5\right)$

D, low-profile cognitive model Parameter analysis of electrochemical

Low-profile cognitive model by learning training determine the low-profile cognitive state probability that matches with observation time sequence and Its parameter, that is, minimize object function：

$f\left(\mathrm{\&Phi;}\right)=\underset{v=1}{\overset{V}{\mathrm{\&Sigma;}}}\underset{t=1}{\overset{T}{\mathrm{\&Sigma;}}}\underset{b\&Element;I}{\mathrm{\&Sigma;}}\frac{P(\hat{I}=b|S,\mathrm{\&Phi;})}{2{\mathrm{\&sigma;}}_v^2}{(s_{tv}-{\mathrm{\&mu;}}_{tv}(b\left)\right)}^2---\left(6\right)$

In formula, Φ represents that in low-profile cognitive model, cognitive stimulation task time sequence, observational characteristic time serieses, recessiveness are recognized Know the mapping probabilities parameter between the triple time serieses of state for time sequence, T is the total time of low-profile cognitive model.

D1, initialization search group：

Carry out parallel search using the population that n Search of Individual is formed in space determined by Φ, make Search of Individual Maximal rate v_max=r；During time step t=0, random initializtion is carried out to n Search of Individual, i.e. i-th Search of Individual The position p of jth dimension_ij=Rand (- r, speed v of the jth dimension of r) He i-th Search of Individual_ij=Rand (- v_max,v_max)；R is Domain of definition, t is time step；

If D2 meets predetermined maximum iteration time or 10 iteration result no improve, output result p* and f (p*) simultaneously tie Bundle calculates；Otherwise, go to step D3；

In formula, best individual state in the group that p* is formed by Search of Individual, f (p*) is the group that Search of Individual is formed In best adaptive value determined by individual state.

D3, the adaptive value of calculating Search of Individual

Calculate the adaptive value of Search of Individual according to formula (6)；

D4, optimal save strategy

Make t=t+1, implement optimum maintaining strategy, that is,：

$p_i^{\#}\left(t\right)=\arg \underset{1\&le;i\&le;n}{min}\left(f\right(p_i^{\#}\left(t-1\right)),f(p_i\left(t\right)\left)\right)$

$p^{*}\left(t\right)=\arg \underset{1\&le;i\&le;n}{min}\left(f\right(p^{*}\left(t-1\right)),f(p_1\left(t\right)),...,f(p_n\left(t\right)\left)\right)$

In formula, best individual state, p in the group that p* is formed by Search of Individual_i ^#It is that i-th Search of Individual is opened from t=0 Begin to iterate to currently best state, f (p*) is to adapt to determined by best individual state in the group that Search of Individual is formed Value；

D5, state transfer joint operation

Introduce the community superiority that formed of Search of Individual dynamically to search for, for each Search of Individual every dimension according to Formula (7) and (8) execution state transfer joint operation：

$v_{ij}\left(t\right)={\mathrm{wv}}_{ij}(t-1)+c_1{r}_1\left(p_{ij}^{\#}\right(t-1)-p_{ij}(t-1\left)\right)+c_2{r}_2\left(p_{ij}^{*}\right(t-1)-p_{ij}(t-1\left)\right)---\left(7\right)$

x_ij(t)=v_ij(t)+x_ij(t-1) (8)

Go to step D2.

Compared with prior art, the invention has the advantages that：

1st, the present invention, during cognitive modeling, input stimulus, observed result, low-profile cognitive state is defined as dynamic thing The related triple time serieses of part, that is,：When cognitive stimulation task time sequence, observational characteristic time serieses, recessive cognitive state Between sequence, and this triple time series is associated by one group of probability distribution, is not that the brain being gathered data is considered as static state Information is counted, thus the cognitive modeling method of the present invention need not meet the basic assumption based on statistics, in Small Sample Database Under the conditions of still set up it is ensured that the correctness of cognitive analysis result, thus the cognitive modeling solving under Small Sample Database is asked Topic.

2nd, the time series brain being observed in cognitive experiment activation data is corresponded to Chinese Cognition state in fact by the present invention The reflection square set of example, wherein each stateful example works according to probability in certain time interval, every in the period that it acts on Individual stateful example affects the activation feature being observed during data acquisition.If multiple stateful example almost work simultaneously, The activation feature so observing is the effect superposition that all correlation behavior examples produce, and is that cognitive state subdivision provides base Plinth, and then improve the precision of cognitive modeling, also provide effective way for complicated cognitive analysis.

3rd, brain low-profile cognitive model proposed by the invention introduces the overall T of model cognitive state, model cognitive state is seen Survey period t and its time step τ, the input time point λ of experiment stimulation, the time migration o time quantum of cognitive state, so that this mould Type has cognitive state time effect, is that a kind of time serieses of the event correlation stimulating series as foundation with cognitive experiment are dynamic Observation procedure, the cognitive activation characteristic sequence that will be observed that during cognition detection is mapped to one or more low-profile cognitions and exists The impact that the same period is activated to brain, and observation sequence and recessive cognitive state sequence are not to correspond, but with thorn Swash related and activate different recessive cognitive states according to certain probability, provide the foundation for cognitive dynamic analysis.

Brief description

The present invention has 2, accompanying drawing, wherein：

Fig. 1 is the flow chart of the present invention.

Fig. 2 is triple time serieses map example figures of the present invention.

Specific embodiment

Below in conjunction with the accompanying drawings the present invention is further described through.As shown in Figure 1-2, the model method of the present invention comprises Triple time serieses, that is,：Upper strata stimulation time sequence Ξ, middle level low-profile cognition time sequence I, observation time sequence S, in observation There is the cognitive superposition of low-profile, that is, middle level low-profile cognition sequence I has multiple arrows sensing phases when being mapped to observation sequence S in sequence Same map section, but there are basic mapping relations in this triple time serieses, and also this relation is determined by probability parameter Φ, Probability mapping relations determined by its probability parameter are parsed according to formula (6) by step D1～D5.Provide simple examples： Input stimulus time serieses Ξ are { Δ₁,Δ₂,Δ₃}；Observation time sequence S is { s₁,s₂,s₃}；Low-profile cognition example time sequence I is { b₁,b₂,b₃, wherein b₁={ ξ₁ ⁽¹⁾,ξ₂ ⁽¹⁾,ξ_L3 ⁽¹⁾},b₂={ ξ₁ ⁽²⁾,ξ₂ ⁽²⁾,ξ₃ ⁽²⁾,ξ₄ ⁽²⁾},b₃={ ξ₁ ⁽³⁾,ξ₂ ⁽³⁾, ξ₃ ⁽³⁾, ξ parenthesized numeral subscript identifies it and belongs to cognitive example accordingly；I corresponding low-profile cognitive state set H is { h₁, h₂,h₃,h₄}.Probability parameter Φ is determined according to formula (6) by step D1～D5, main result is：ξ₁ ⁽¹⁾→s₁(0.82), ξ₂ ⁽¹⁾→s₁(0.91),ξ₃ ⁽¹⁾→s₂(0.86)；

ξ₁ ⁽²⁾→s₂(0.90),ξ₂ ⁽²⁾→s₂(0.76),ξ₃ ⁽²⁾→s₃(0.84),ξ₄ ⁽²⁾→s₃(0.69)；

ξ₁ ⁽³⁾→s₃(0.78),ξ₂ ⁽³⁾→s₃(0.85),ξ₃ ⁽³⁾→s₃(0.80)；

ξ₁ ⁽¹⁾→h₁(0.90),ξ₂ ⁽¹⁾→h₂(0.74),ξ₃ ⁽¹⁾→h₃(0.92)；

ξ₁ ⁽²⁾→h₁(0.97),ξ₂ ⁽²⁾→h₂(0.88),ξ₃ ⁽²⁾→h₃(0.79),ξ₄ ⁽²⁾→h₄(0.85)；

ξ₁ ⁽³⁾→h₁(0.93),ξ₂ ⁽³⁾→h₃(0.87),ξ₃ ⁽³⁾→h₄(0.91).

In formula：There are mapping relations in " → " expression, the numeral in bracket represents corresponding mapping probabilities.

Claims (1)

1. a kind of human brain language cognition modeling method it is characterised in that：Comprise the following steps：

A, the initialization of cognitive state example

Identify the cognitive state in human brain with h, because h can not be directly observed with instrument and equipment, thus h is referred to as recessiveness Cognitive state, recessive cognitive state h is expressed as four-tuple<W,d,Θ,Ω>, wherein：D represents h lasting mark in time Amount；W is the reflection square that Spatial dimensionality is d × V, the brain area dimension that wherein V is related to for h, and its Spatial Dimension component is v；Θ is h's Parameter space, Ω is the codomain of parameter space；In cognitive observation process, an input stimulus Δ, the activation of brain cognitive state can be deposited In certain time lag；To the cognitive state being in activated state, it is cognitive state example, with ξ labelling it is considered to recessive cognitive state The time factor of h, cognitive state example is a tlv triple<h,λ,o>, wherein：H is real for recessive cognitive state accordingly, λ Test the input time point of stimulation, o is the time migration of cognitive state, the initial time of cognitive state example ξ is λ+o, and it continues Time d is identical with corresponding recessiveness cognitive state h duration, all uses d (h) to represent；

Probability distribution between B, mapping activation feature and observation data

It is assumed that in cognitive experimentation, the t period activates feature b by the brain that experimenter observed by imaging device_tIt is cognitive by L The coefficient result of stateful example, uses ξ₁,ξ₂,…,ξ_LIndicate this L cognitive state example, then note b_t={ ξ₁,ξ₂,…, ξ_L}；b_tComponent s with period t, Spatial Dimension component v in observation time sequence S_tvNormal Distribution, that is,：

s_tv～N (μ_tv(b_t),σ_v) (1)

Wherein σ_vIt is standard variance, it reflects the noise profile related to observation time sequence, has time independence；μ_tv(b_t) Reflect the Overlay of all related cognitive state of corresponding period, that is, meet following formula：

${\mathrm{\&mu;}}_{tv}\left(b_t\right)=\underset{\mathrm{\&xi;}\&Element;b_t}{\&Sigma;}\stackrel{d\left(h\right(\mathrm{\&xi;}\left)\right)}{\mathrm{\&Sigma;}}\mathrm{\&delta;}\left(\mathrm{\&lambda;}\right(\mathrm{\&xi;})+o(\mathrm{\&xi;})=t-\mathrm{\&tau;})w_{\mathrm{\&tau;}v}\left(h\right(\mathrm{\&xi;}\left)\right)---\left(2\right)$

Wherein δ (.) is indicator function, if independent variable is true, this function result is 1, otherwise for 0；H (ξ) identifies cognitive state Example ξ corresponding recessiveness cognitive state；w_τv(h (ξ)) is then recessive cognitive state reflection square W on the τ time step of mark period t The v component of (h (ξ))；

C, definition brain low-profile cognitive model

Define brain low-profile cognitive model and be one hexa-atomic group<H,Φ,I,S,Ξ,Γ>, use HCM labelling, wherein：H is recessive cognitive State set；Φ is the parameter vector of low-profile cognitive model, and they rely on experimental design parameter, that is,：The type stimulating, input are pierced Sharp time point；S is observation time sequence；The stimulus sequence that Ξ is made up of input stimulus Δ；When I is cognitive state example Between sequence, because the activationary time point of cognitive state example is corresponding with observation time sequence, thus by the element b in I_tWrite a Chinese character in simplified form For b；Γ is<σ₁,σ₂,…,σ_v>Variance collection, wherein σ_vIt is the standard variance in formula (1), the noise of reflecting time sequence divides Cloth；HCM defines I and S, Ξ constitute triple time serieses of probability correlation,Stochastic variable defined in I, then observe The probability distribution that brain activates between feature and recessive cognitive state is：

$P\left(S\right|HCM,\mathrm{\&Delta;})=\underset{b\&Element;I}{\&Sigma;}P\left(S\right|HCM,\hat{I}=b)P(\hat{I}=b|HCM,\mathrm{\&Delta;})---\left(3\right)$

Wherein,

$P(\hat{I}=b|HCM,\mathrm{\&Delta;})=\frac{\underset{\mathrm{\&xi;}\&Element;b}{\&Pi;}P\left(h\right(\mathrm{\&xi;}\left)\right|HCM,\mathrm{\&Delta;})P\left(o\right(\mathrm{\&xi;}\left)\right|h\left(\mathrm{\&xi;}\right),HCM,\mathrm{\&Delta;})}{\underset{b^{,}\&Element;I}{\&Sigma;}\underset{{\mathrm{\&xi;}}^{,}\&Element;b^{,}}{\&Pi;}P\left(h\right({\mathrm{\&xi;}}^{,}\left)\right|HCM,\mathrm{\&Delta;})P\left(o\right({\mathrm{\&xi;}}^{,}\left)\right|h\left({\mathrm{\&xi;}}^{,}\right),HCM,\mathrm{\&Delta;})}---\left(4\right)$

In formula, and P (h (ξ) | HCM, Δ) it is that Chinese Cognition experimentation conditional stimulates as the recessive cognitive state h of Δ activation and phase Answer the conditional probability of cognitive state example ξ, and P (o (ξ) | h (ξ), HCM, Δ) it is corresponding time lag conditional probability；Low-profile cognition mould The modeling process of type needs to determine its conditional probability for specific cognitive process, and the ginseng of low-profile cognitive model is also predefined simultaneously HCM parameters are carried out maximal possibility estimation by number vector；For HCM, had according to Bayes theorem：

$P(\hat{I}=b|S,HCM,\mathrm{\&Delta;})=\frac{P\left(S\right|\hat{I}=b,HCM)P(\hat{I}=b|HCM,\mathrm{\&Delta;})}{\underset{b^{\&prime;}\&Element;I}{\&Sigma;}P\left(S\right|\hat{I}=b^{\&prime;},HCM)P(\hat{I}=b^{\&prime;}|HCM,\mathrm{\&Delta;})}---\left(5\right)$

D, low-profile cognitive model Parameter analysis of electrochemical

Low-profile cognitive model determines the low-profile cognitive state probability and its ginseng matching with observation time sequence by learning training Number, that is, minimize object function：

$f\left(\mathrm{\&Phi;}\right)=\underset{v=1}{\overset{V}{\&Sigma;}}\underset{t=1}{\overset{T}{\&Sigma;}}\underset{b\&Element;I}{\&Sigma;}\frac{P(\hat{I}=b|S,\mathrm{\&Phi;})}{2{\mathrm{\&sigma;}}_v^2}{(s_{tv}-{\mathrm{\&mu;}}_{tv}(b\left)\right)}^2---\left(6\right)$

In formula, Φ represents cognitive stimulation task time sequence in low-profile cognitive model, observational characteristic time serieses, recessive cognition shape Mapping probabilities parameter between the triple time serieses of state time serieses, T is the total time of low-profile cognitive model；

D1, initialization search group：

Carry out parallel search using the population that n Search of Individual is formed in space determined by Φ, make Search of Individual Big speed v_max=r；During time step t=0, random initializtion is carried out to n Search of Individual, i.e. the jth of i-th Search of Individual The position p of dimension_ij=Rand (- r, speed v of the jth dimension of r) He i-th Search of Individual_ij=Rand (- v_max,v_max)；R is definition Domain, t is time step；

If D2 meets predetermined maximum iteration time or 10 iteration result no improve, output result p* and f (p*) simultaneously terminate to count Calculate；Otherwise, go to step D3；

In formula, best individual state in the group that p* is formed by Search of Individual, f (p*) be in the group that Search of Individual is formed Good adaptive value determined by individual state；

D3, the adaptive value of calculating Search of Individual

Calculate the adaptive value of Search of Individual according to formula (6)；

D4, optimal save strategy

Make t=t+1, implement optimum maintaining strategy, that is,：

$p_i^{\#}\left(t\right)=\arg \underset{1\&le;i\&le;n}{min}\left(f\right(p_i^{\#}\left(t-1\right)),f(p_i\left(t\right)\left)\right)$

$p^{*}\left(t\right)=\arg \underset{1\&le;i\&le;n}{min}\left(f\right(p^{*}\left(t-1\right)),f(p_1\left(t\right)),...,f(p_n\left(t\right)\left)\right)$

In formula, best individual state, p in the group that p* is formed by Search of Individual_i ^#It is that i-th Search of Individual starts to change from t=0 In generation, f (p*) was adaptive value determined by best individual state in the group that Search of Individual is formed to currently best state；

D5, state transfer joint operation

Introduce the community superiority that formed of Search of Individual dynamically to search for, for each Search of Individual every dimension according to formula (7) execute state and shift joint operation with (8)：

$v_{ij}\left(t\right)={\mathrm{wv}}_{ij}\left(t-1\right)+c_1{r}_1\left(p_{ij}^{\#}\left(t-1\right)-p_{ij}\left(t-1\right)\right)+c_2{r}_2\left(p_{ij}^{*}\left(t-1\right)-p_{ij}\left(t-1\right)\right)---\left(7\right)$

x_ij(t)=v_ij(t)+x_ij(t-1) (8)

Go to step D2.