CN106709569A - Parameter estimation method for FitzHugh-Nagumo neuron system - Google Patents
Parameter estimation method for FitzHugh-Nagumo neuron system Download PDFInfo
- Publication number
- CN106709569A CN106709569A CN201510777014.9A CN201510777014A CN106709569A CN 106709569 A CN106709569 A CN 106709569A CN 201510777014 A CN201510777014 A CN 201510777014A CN 106709569 A CN106709569 A CN 106709569A
- Authority
- CN
- China
- Prior art keywords
- fitzhugh
- nagumo
- parameter
- innovation
- parameter estimation
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 210000002569 neuron Anatomy 0.000 title claims abstract description 20
- 238000000034 method Methods 0.000 title claims abstract description 12
- 238000004364 calculation method Methods 0.000 claims abstract description 10
- 239000011159 matrix material Substances 0.000 claims abstract description 9
- 230000009466 transformation Effects 0.000 claims abstract description 4
- 230000001537 neural effect Effects 0.000 claims 1
- 238000012417 linear regression Methods 0.000 abstract 1
- 230000000739 chaotic effect Effects 0.000 description 2
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/088—Non-supervised learning, e.g. competitive learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
- Complex Calculations (AREA)
Abstract
The invention discloses a parameter estimation method for a FitzHugh-Nagumo neuron system, which comprises the steps of performing model transformation on the FitzHugh-Nagumo neuron system to acquire a linear regression model, acquiring input-output data, building an output matrix and an information matrix, further acquiring an innovation vector, and finally estimating system parameters under zero-mean white noise interference by using an innovation projection algorithm. The parameter estimation method has the advantages of high convergence rate, small calculation amount, no need of calculating a covariance matrix and the like.
Description
Background
In recent years, inspired by biological research results, biological neuron models are becoming one of the research hotspots. The biological neuron model has abundant calculation characteristics, and the characteristics bring wide application prospects for the biological neuron model. Research shows that the FitzHugh-Nagumo neuron system can show a chaotic state when parameters are properly selected, but system parameters are not necessarily known in practice, and although some parameter estimation methods of the chaotic biological neuron system exist at present, such as an adaptive synchronization algorithm and the like, the calculation amount is large, the realization is complex, and certain limitations exist.
The innovation projection algorithm is a generalization of the projection algorithm. The projection algorithm only uses the current data and the information to correct the parameter estimation value, and the information projection algorithm not only uses the current data and the information, but also fully uses the past data and the information, so the convergence speed is higher compared with the projection algorithm, and the parameter estimation accuracy is higher. The invention provides a novel parameter estimation method aiming at a FitzHugh-Nagumo neuron system by combining with an innovation projection algorithm.
Disclosure of Invention
The invention aims to estimate parameters of a FitzHugh-Nagumo neuron system by using an innovation projection algorithm under zero-mean white noise interference. The FitzHugh-Nagumo neuron system is obtained by simplifying a classical Hodgkin-Huxley neuron model, and the mathematical equation is as follows:
wherein x is1,x2For system state variables, a and f are the amplitude and frequency, ω, of the external stimulus current, respectively1(t) and ω2(t) is zero mean, variance is σ2R and b are the system parameters to be estimated.
The technical scheme adopted by the invention comprises the following steps:
step 1: and (5) model transformation. The FitzHugh-Nagumo neuron model was transformed into the following estimated standard form:
wherein,
θ=[b r]T。
step 2: and (6) parameter estimation. To estimate the parameter vector θ, the following recursive innovation projection algorithm is used:
(a) and initializing parameters. Setting the parameter estimation value asTaking the maximum iteration number as tmaxT is 0, the innovation length is taken as p, and the initial parameter is estimated asWherein 1 ═ 11 … 1]T。
(b) Collecting data, and constructing output matrix Y (p, t) and information matrix
Y(p,t)=[YT(t) YT(t-1) … YT(t-p+1)]T
(c) The innovation vector E (p, t) is calculated as follows:
(d) updating parameter estimates
(e) Let t: t +1, if t < tmaxTurning to the step (b), and continuing to perform recursive calculation;otherwise, go to step 3.
And step 3: stopping calculation to obtain the final estimated parameter value
Compared with the existing estimation algorithm, the method has the following advantages: the innovation projection algorithm finds a compromise between convergence speed and computational effort. Compared with a projection algorithm, the convergence speed of the innovation projection algorithm is higher, and the convergence speed can be improved by changing the innovation length of the innovation projection algorithm. Compared with the least square method, the innovation projection algorithm has small calculation amount and does not need to calculate a covariance matrix.
Drawings
Fig. 1 is a flow chart of parameter estimation based on the scheme of the present invention.
Fig. 2 is a graph of an iterative process of parameter estimation based on the scheme of the present invention in an example.
Detailed Description
In order to better understand the technical solution of the present invention, the following embodiments are further described in detail.
Example (b): consider a specific FitzHugh-Nagumo neuron system model:
wherein x is1,x2For the system state variables, a-0.1 and f-0.129 are the amplitude and frequency, respectively, of the external stimulus current, ω1(t) and ω2(t) is zero mean, variance is σ20.01 ═ 0.01White noise, r and b are system parameters to be estimated. When the system parameter is r-10 and b-1, the neuron system will exhibit chaos.
The working flow of the method of the invention is shown in fig. 1, and the specific implementation mode can be divided into the following steps:
step 1: and (5) model transformation. The FitzHugh-Nagumo neuron model was transformed into the following estimated standard form:
wherein,
θ=[b r]T。
step 2: and (6) parameter estimation. To estimate the parameter vector θ, the following recursive innovation projection algorithm is used:
(a) and initializing parameters. Setting the parameter estimation value asTaking the maximum iteration number as tmax1500, t is 0, the innovation length is p is 5, and the initial parameter is estimated asWherein 1 ═ 11 … 1]T。
(b) Collecting data (obtained by solving a neuron system equation by adopting a fourth-order Runge-Kutta method), and constructing an output matrixY (p, t) and information matrix
Y(p,t)=[YT(t) YT(t-1) … YT(t-p+1)]T
(c) The innovation vector E (p, t) is calculated as follows:
(d) updating parameter estimates
(e) Let t: t +1, if t < tmaxTurning to the step (b), and continuing to perform recursive calculation; otherwise, go to step 3.
And step 3: stopping calculation to obtain the final estimated parameter value
Fig. 2 shows a parameter estimation iteration process curve. As can be seen, all parameter estimates converge to true values approximately after 1000 iterations.
Claims (1)
1. A FitzHugh-Nagumo neuron system parameter estimation method is characterized by comprising the following steps:
step 1: and (5) model transformation. Consider the FitzHugh-Nagumo neuronal system:
wherein x is1,x2For system state variables, a and f are the amplitude and frequency, ω, of the external stimulus current, respectively1(t) and ω2(t) is zero mean, variance is σ2R and b are the system parameters to be estimated.
The FitzHugh-Nagumo neuron model was transformed into the following estimated standard form:
wherein,
θ=[b r]T。
step 2: and (6) parameter estimation. To estimate the parameter vector θ, the following recursive innovation projection algorithm is used:
(a) and initializing parameters. Setting the parameter estimation value asTaking the maximum iteration number as tmaxT is 0, the innovation length is taken as p, and the initial parameter is estimated asWherein 1 ═ 11 … 1]T。
(b) Collecting data, and constructing output matrix Y (p, t) and information matrix
Y(p,t)=[YT(t) YT(t-1) … YT(t-p+1)]T
(c) The innovation vector E (p, t) is calculated as follows:
(d) updating parameter estimates
(e) Let t: t +1, if t < tmaxTurning to the step (b), and continuing to perform recursive calculation; otherwise, go to step 3.
And step 3: stopping calculation to obtain the final estimated parameter value
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510777014.9A CN106709569A (en) | 2015-11-12 | 2015-11-12 | Parameter estimation method for FitzHugh-Nagumo neuron system |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510777014.9A CN106709569A (en) | 2015-11-12 | 2015-11-12 | Parameter estimation method for FitzHugh-Nagumo neuron system |
Publications (1)
Publication Number | Publication Date |
---|---|
CN106709569A true CN106709569A (en) | 2017-05-24 |
Family
ID=58931811
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510777014.9A Pending CN106709569A (en) | 2015-11-12 | 2015-11-12 | Parameter estimation method for FitzHugh-Nagumo neuron system |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106709569A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109886345A (en) * | 2019-02-27 | 2019-06-14 | 清华大学 | Self-supervisory learning model training method and device based on relation inference |
CN113361688A (en) * | 2021-06-04 | 2021-09-07 | 浙江工业大学 | Modeling method for action potential in giant axon of cuttlefish |
-
2015
- 2015-11-12 CN CN201510777014.9A patent/CN106709569A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109886345A (en) * | 2019-02-27 | 2019-06-14 | 清华大学 | Self-supervisory learning model training method and device based on relation inference |
CN113361688A (en) * | 2021-06-04 | 2021-09-07 | 浙江工业大学 | Modeling method for action potential in giant axon of cuttlefish |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Zarzoso et al. | Robust independent component analysis by iterative maximization of the kurtosis contrast with algebraic optimal step size | |
CN106487358B (en) | A kind of maneuvering target turning tracking | |
CN103326903B (en) | Based on the Internet network latency prediction method of Hidden Markov | |
CN109163720A (en) | Kalman filter tracking method based on fading memory exponent | |
CN108153259B (en) | Multi-controller optimal state estimation control strategy design method based on Kalman filtering | |
CN103258533B (en) | Novel model domain compensation method in remote voice recognition | |
CN109410251B (en) | Target tracking method based on dense connection convolution network | |
CN107844627A (en) | It is a kind of only to export Time variable structure modal parameter Bayesian Estimation method | |
CN103473790B (en) | Based on the online method for tracking target of increment Bi-2DPCA study and rarefaction representation | |
CN103684349B (en) | A kind of kalman filter method based on recursion Estimates on Covariance Matrix | |
CN103500455B (en) | A kind of improvement maneuvering target tracking method based on without inclined to one side finite impulse response filter (UFIR) | |
CN109388778A (en) | A kind of iteration volume point Unscented kalman filtering method | |
Zhou et al. | Network traffic prediction method based on improved echo state network | |
CN111862167B (en) | Rapid robust target tracking method based on sparse compact correlation filter | |
CN109687844A (en) | A kind of intelligent maneuver method for tracking target | |
CN106709569A (en) | Parameter estimation method for FitzHugh-Nagumo neuron system | |
CN110378932B (en) | Correlation filtering visual tracking method based on spatial regularization correction | |
WO2015176502A1 (en) | Image feature estimation method and device | |
CN103208113A (en) | Image segmentation method based on non-subsmapled contourlet and multi-phase chan-vese (CV) models | |
CN104463214A (en) | SVR parameter optimization method based on DRVB-ASCKF | |
CN104868876B (en) | A kind of Kalman filter method being directed under process noise covariance matrix Q unknown situations | |
CN112636719B (en) | ILC system input signal filtering method under data loss and channel noise interference | |
CN104134091B (en) | Neural network training method | |
Nguyen et al. | Asymptotic inference for hidden process regression models | |
CN112819743A (en) | General video time domain alignment method based on neural network |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
WD01 | Invention patent application deemed withdrawn after publication | ||
WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20170524 |