Medical, Pharma, Engineering, Science, Technology and Business

**Tawfiq LNM ^{*} and Al-Abrahemee KMM**

Department of Mathematics, College of Education for Pure Science/Ibn Al-Haitham, University of Baghdad, Baghdad, Iraq

- *Corresponding Author:
- Tawfiq LNM

Department of Mathematics

College of Education for pure science/Ibn Al-Haitham

University of Baghdad, Baghdad, Iraq

**E-mail:**[email protected]

**Received Date:** February 04, 2014; **Accepted Date:** March 25, 2014; **Published Date:** April 14, 2014

**Citation:** Tawfiq LNM, Al-Abrahemee KMM (2014) Design Neural Network to Solve Singular Perturbation Problems. J Appl Computat Math 3:160 doi: 10.4172/2168-9679.1000160

**Copyright:** © 2014 Tawfiq LNM, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

**Visit for more related articles at** Journal of Applied & Computational Mathematics

The aim of this paper is to design neural network to present a method to solve Singular perturbation problems (SPP) by using network having one hidden layer with 5 hidden units (neurons) and one linear output unit, the sigmoid activation of each hidden units is tansigmoid. The neural network trained by the back propagation with different algorithms such as quasi-Newton, Levenberg-Marquardt, and Bayesian Regulation. Finally the results of numerical experiments are compared with the exact solution in illustrative examples to confirm the accuracy and efficiency of the presented scheme.

Singularly perturbed problems; Neural network; Quasi- Newton; Levenberg-Marquardt, Bayesian regulation

Singularly perturbed problems (SPP) are common in applied sciences and engineering. They often occur in, for example, fluid dynamics, quantum mechanics, chemical reactions, electrical networks, etc. A well known fact is that the solution of such problems has a multi scale character, i.e., there are thin transition layers where the solution varies very rapidly, while away from the layers the solution behaves regularly and varies slowly. For a detailed discussion on the analytical and numerical treatment of such problems one may refer to the Malley [1], Doolan et al. [2], Roos et al. [3], and Miller et al. [4]. Numerically, the presence of the perturbation parameter leads to difficulties when classical numerical techniques are used to solve such problems, this is due to the presence of the boundary layers in these problems, see for example [1,5]. Even in the case when only the approximate solution is required.

Many methods have been developed so far solving Singularly perturbed boundary value problems (SPBVP) , nowadays there is a new way of computing denominated artificial intelligence which through different methods is capable of managing the imprecision's and uncertainties that appear when trying to solve problems related to the real world, offering strong solution and of easy implementation. One of those techniques is known as Artificial Neural Networks (ANN). Inspired, in their origin, in the functioning of the human brain, and entitled with some intelligence. These are the combination of a great amount of elements of process–artificial neurons interconnected that operating in a parallel way get to solve problems related to aspects of classification. The construction of any given ANN we can identify, depending on the location in the network, three kind of computational neurons: input, output and hidden.

The term "perturbation problem" is generally used in mathematics when one deals with the following situation: There is a family of problems depending on a small parameter ϵ> 0,, which we denote by P_{ϵ}, when ϵ=0, we have the reduced problem P_{o}.

We want to study the relationship between the solution of P_{ϵ} and the solution of P_{o} under appropriate assumptions. The perturbation problem (PP), may consist of an ordinary differential equation, or a system of differential equations, dong with some given conditions which illustrate the problem. Thus, the general form of the 2^{nd} order singularly perturbed problems (SPP):

(1)

where f are n-dimensional vector functions, x is a scalar variable in a given interval.

A perturbation problem (1) is called SPP if ϵ→ 0, the solution y_{ϵ}(x) converges to y_{o}(x) only in some interval of x, but not throughout the entire interval, thus giving rise to an "boundary layers" phenomena at both end-P_{o}ints [6].

Artificial neural network is a simplified mathematical model of the human brain. It can be implemented by both electric elements and computer software. It is a parallel distributed processor with large numbers of connections; it is an information processing system that has certain performance characters in common with biological neural networks [7].

The arriving signals, called inputs, multiplied by the connection weights (adjusted) are first summed (combined) and then passed through a transfer function to produce the output for that neuron. The transfer (activation) function acts on the weighted sum of the neuron’s inputs and the most commonly used transfer function is the sigmoid function (tansig.) [8].

There are two main connection formulas (types): feedback (recurrent) and feed forward connections. Feedback is one type of connection where the output of one layer routes back to the input of a previous layer, or to the same layer. Feed forward neural network (FFNN) does not have a connection back from the output to the input neurons [9]. There are many different training algorithms, but the most often used training algorithm is the back propagation (BP) rule. A NN is trained to map a set of input data by iterative adjustment of the weights. Information from inputs is fed forward through the network to optimize the weights between neurons. Optimization of the weights is made by backward propagation of the error during training phase. The ANN reads the input and output values in the training data set and changes the value of the weighted links to reduce the difference between the predicted and target (observed) values. The error in prediction is minimized across many training cycles (iteration or eP_{o}ch) until network reaches specified level of accuracy. A complete round of forward backward passes and weight adjustments using all input output pairs in the data set is called an eP_{o}ch or iteration. In order to perform a supervised training we need a way of evaluating the ANN output error between the actual and the expected outputs. A P_{o}pular measure is the mean squared error (MSE) or root mean squared error (RMSE) [10].

In an ANN expressions structure, architecture or toP_{o}logy, express the way in which computational neurons are organized in the network [11].

Particularly, these terms are focused in the description of how the nodes are connected and in how the information is transmitted through the network. As it has been mentioned, the distribution of computational in the following:

**Number of levels or layers**

Neurons in the neural network is done forming levels or layers of a determined number of nodes each one. As there are input, output and hidden neurons, we can talk about an input layer, an output layer and single layer or multilayer hidden layers. By the peculiarity of the behavior of the input nodes some authors consider just two kinds of layers in the ANN, the hidden and the output.

**Connection patterns**

Depending on the links between the elements of the different layers. The ANN can be classified as: totally connected, when all the outputs from a level get to all and each one of the nodes in the following level, if some of the links in the network are lost, then we say that the network is partially connected.

**Information flow**

Another classification of the ANN is obtained by considering the direction of the flow of the through the layers, when any output of the neurons is input of neurons of the same level or preceding levels, the network is described as feed forward. In counter P_{o}sition if there is at least one connected exit as entrance of neurons of previous levels or of the same level, including themselves, the network is denominated of feedback.

This section illustrates how our approach can be used to the approximation solution of the singular perturbation problem of the form:

(2)

where x ϵ D, D ⊂ R denoted the domain and y(x) is the solution to be computed.

If yt (x, p) denoted a trial solution with adjustable parameters p, the problem is transformed to a discredited from:

(3)

In our proP_{o}sed approach, the trial solution y_{t} employs an ANN and the parameters p corresP_{o}nd to the weights and biases of the neural architecture.

We choose a form for the trial function (x) such that achieved by writing as a sum of two terms:

(4)

where *N(x, p)* is ANN with parameters *p* and *n* input units fed with the input vector x. The term (x) contains no adjustable parameters and satisfies the BCs. The second term G is constructed so as not to contribute to the BCs, since (x) satisfy them. This term can be formed by using ANN whose weights and biases are to be adjusted in order to deal with the minimization problem.

In this section, we describe solution of SPP using ANN. To illustrate the method, we will consider the 2^{nd} order SPP:

(5)

where x ∈ [a, b] and the BC: y(a)=A, y(b)=B; a trial solution can be written as:

where *N(x, p)* is the output of the ANN with one input unit for x and weights p.

The error quantity to be minimized is given by:

(7)

where the x_{i} ∈ [a, b]. Since

(8)

It is straight forward to compute the gradient of the error with respect to the parameters p.

In this section, we reP_{o}rt numerical result, using amulti-layer ANN having one hidden layer with 5 hidden units (neurons) and one linear output unit. The sigmoid activation of each hidden unit is tansig, the analytic solution (x) was known in advance. Therefore, we test the accuracy of obtained solutions computing the mean square error (MSE).

**Example 1**

Consider the following 2^{nd} order singular perturbation problem:

BC′s (Dirishlit case): y(0)=-1, y(1)=3.9995, and the analytic solution [12]:

where,

c_{1} = 2.9995

according to the equation (6 ) the trial neural form of the solution is taken to be:

The ANN trained using a grid of ten equidistant P_{o}ints in 0, 1. **Figure 1** displays the analytic and neural solutions with different training algorithm. The neural results with different types of training algorithm such as: Levenberg–Marquardt (trainlm), quasi–Newton (trainbfg), Bayesian Regulation (trainbr) introduced in **Table 1** and its errors gave in **Table 2**, **Table 3** gives the performance of the train with eP_{o}ch and time and **Table 4** gives the weight and bias of the designer network.

input | Analytic solution | Out of suggested ANN y_{t}(x) for different training algorithm |
Numerical solution | ||
---|---|---|---|---|---|

x | y_{a}(x) |
Trainlm | Trainbfg | Trainbr | |

0.0 | 3.09950000000000 | -0.999880745270845 | -0.999880745270845 | -0.999934351350428 | -1.0000000 |

0.1 | 3.09950000000000 | 3.09952352215727 | 3.09952352215727 | 3.09928640827629 | 3.1035900 |

0.2 | 3.19950000000000 | 3.19982552466162 | 3.19982552466162 | 3.19982150310036 | 3.2011864 |

0.3 | 3.29950000000000 | 3.29890300366976 | 3.29890300366976 | 3.29912397768203 | 3.3003857 |

0.4 | 3.39950000000000 | 3.40001713095587 | 3.40001713095587 | 3.39987747119349 | 3.3999858 |

0.5 | 3.49950000000000 | 3.49974092352976 | 3.49974092352976 | 3.47742525061300 | 3.4997456 |

0.6 | 3.59950000000000 | 3.59934730598352 | 3.59934730598352 | 3.59853213880555 | 3.5995858 |

0.7 | 3.69950000000000 | 3.69942849140595 | 3.69942849140595 | 3.70080106004170 | 3.6994715 |

0.8 | 3.79950000000000 | 3.79968216405646 | 3.79968216405646 | 3.79794587586114 | 3.7993855 |

0.9 | 3.89950000000000 | 3.89954946888650 | 3.89954946888650 | 3.89853640620021 | 3.8993189 |

1.0 | 3.99950000000000 | 3.99860986942680 | 3.99860986942680 | 3.99996079784071 | 3.9992657 |

**Table 1:** Analytic and Neural solution of Example 1.

The error E(x) = | y_{t}(x)-y_{a}(x) | where y_{t}(x) computed by the following training algorithm |
||
---|---|---|

Trainlm | Trainbfg | Trainbr |

6.43341191782776e-05 | 6.77710633278927e-07 | 6.56486495714814e-05 |

5.53198735442351e-05 | 1.59832806989613e-07 | 0.000213591723708273 |

0.000261518497015612 | 1.20135476944228e-06 | 0.000321503100361831 |

0.00110130535869502 | 0.000916274984845256 | 0.000376022317968516 |

0.00122930168483482 | 7.29152511702580e-08 | 0.000377471193488521 |

0.000969607591307931 | 5.38443689634960e-09 | 0.0220747493870004 |

0.00213983449812005 | 0.00190942600971189 | 0.000967861194448805 |

0.000263003413316199 | 0.00156884042056982 | 0.00130106004170516 |

0.00288916575896980 | 5.40250320035796e-08 | 0.00155412413885836 |

0.00368039653067687 | 0.000814108450935880 | 0.000963593799789830 |

0.00180864374061107 | 2.08825303715798e-08 | 0.000460797840708516 |

**Table 2:** Accuracy of solutions for Example 1.

Train Function | Performance of train | EP_{o}ch |
Time | Msereg. |
---|---|---|---|---|

Trainlm | 1.04e-30 | 18 | 0:00:01 | 1.508187974455366e-07 |

Trainbfg | 3.45e-21 | 251 | 0:00:04 | 6.917729369371148e-07 |

Trainbr | 4.68e-04 | 1386 | 0:00:16 | 4.490158138376238e-05 |

**Table 3:** The performance of the train with eP_{o}ch and time.

Weights and bias for trainlm |
||||

Net.IW{1,1} | Net.LW{2,1} | Net.B{1} | ||

0.3313 | 0.1347 | 0.5047 | ||

0.4316 | 0.1199 | 0.8945 | ||

0.7179 | 0.8935 | 0.3857 | ||

0.9162 | 0.6531 | 0.2921 | ||

0.8900 | 0.0403 | 0.2340 | ||

Weights and bias for trainbfg |
||||

Net.IW{1,1} | Net.LW{2,1} | Net.B{1} | ||

0.1999 | 0.3585 | 0.2499 | ||

0.1630 | 0.1343 | 0.3648 | ||

0.0369 | 0.9986 | 0.3991 | ||

0.2727 | 0.5135 | 0.9263 | ||

0.2301 | 0.3878 | 0.4955 | ||

Weights and bias for trainbr |
||||

Net.IW{1,1} | Net.LW{2,1} | Net.B{1} | ||

0.9193 | 0.3865 | 0.6633 | ||

0.9889 | 0.6030 | 0.6023 | ||

0.9326 | 0.5603 | 0.6565 | ||

0.4615 | 0.8458 | 0.3099 | ||

0.9049 | 0.2848 | 0.3316 |

**Table 4:** Initial weight and bias of the network for different training algorithm.

**Example 2**

Consider the following 2^{nd} order singular perturbed problem:

with BC: y(0)=0, y(1)=0 and x∈ (0, 1).

The analytic solution is [12]:

The ANN trained using a grid of ten equidistant P_{o}ints in 0, 1. **Figure 2** display the analytic and neural solutions with different training algorithm. The neural results with different types of training algorithm such as: Levenberg–Marquardt (trainlm), quasi–Newton (trainbfg), Bayesian Regulation (trainbr) introduced in **Table 5** and its errors gave in **Tables 6** and **7** gives the performance of the train with eP_{o}ch and time and **Table 8** gives the weight and bias of the designer network.

In-put | Analytic solution | Out of suggested FFNN y_{t}(x) for different training algorithm |
Numerical Patching method | ||
---|---|---|---|---|---|

x | y_{a}(x) |
Trainlm | Trainbfg | Trainbr | |

0.0 | 0 | 3.88590164923235e-06 | -0.000108152119378016 | 0.0000000 | 5.30985922569390e-16 |

0.1 | 0.597837000755620 | 0.592918052256386 | 0.572073288709426 | 0.5978370 | 0.597837000755621 |

0.2 | 0.510825623765991 | 0.510779902455636 | 0.502977459576147 | 0.5108256 | 0.510825623765991 |

0.3 | 0.430782916092454 | 0.430881601043291 | 0.433818154990611 | 0.4307829 | 0.428258098617260 |

0.4 | 0.356674943938732 | 0.356684767145891 | 0.364595340116877 | 0.3566749 | 0.356228190041481 |

0.5 | 0.287682072451781 | 0.287601746123872 | 0.295308980153126 | 0.2876821 | 0.287682072451781 |

0.6 | 0.223143551314210 | 0.223043377389487 | 0.225959040328804 | 0.2231435 | 0.223143551314210 |

0.7 | 0.162518929497775 | 0.162467868943705 | 0.156545485907880 | 0.1625189 | 0.162518929497775 |

0.8 | 0.105360515657826 | 0.105383361925040 | 0.105425301654285 | 0.1053605 | 0.105379554764756 |

0.9 | 0.0512932943875505 | 0.0513474403606819 | 0.0358844139613455 | 0.0512933 | 0.0512932943875507 |

1.0 | 0 | -3.48131331875563e-05 | -0.0337201923390429 | 1.0000000 | -6.99557875567738e-05 |

**Table 5:** Analytic and Neural solution of Example 2.

The error E(x) thit(x) -y,(x) | 1 where yt(x) computed by the follming training algorithm | |
---|---|---|

Trait | Trainbfg | Trainbr |

5.30985922569390e-16 | 3.88590164923235e-06 | 0.0001081)2119373016 |

5.55111512312578e-16 | 0.00491894349923447 | 0.0257637120461945 |

0 | 4.57213103546295e-05 | 0.00784816413984335 |

0.00252481747519467 | 9.86849503369248e-05 | 0.00303523889815655 |

0.000446753897250962 | 9.82320715375451e-06 | 0.00792039617814461 |

0.00762690770134561 | 3.03263279092747e-05 | 2.77555756156239e.16 |

0.00281548901459333 | 0.000100173924722308 | 1.94289029309402e-16 |

0.00597344358989510 | 5.10605540695319e-05 | 4.44039209850063e-16 |

6.47859964587327e-05 | 2.28462672137442e-05 | 1.90391069293111e-05 |

0.0154038304262050 | 5.41459731314))9e-05 | 1.73472347597631e-16 |

0.0337201923390429 | 3.48131331375563e-05 | 6.99557875567738e-05 |

**Table 6: **Accuracy of solutions for Example 2.

Msereg. | Time | EP_{o}ch |
Performance of train | Train Function |
---|---|---|---|---|

5.383267058914925e-07 | 0:00:03 | 253 | 1.66e-32 | Trainlm |

1.982598144414512e-06 | 0:00:07 | 479 | 3.04e-09 | Trainbfg |

1.860207937865030e-04 | 0:00:22 | 1558 | 2.30e-13 | Trainbr |

**Table 7:** The performance of the train with eP_{o}ch and time.

Weights and bias for trainhn |
||

Net.B[11 | Net.LW(2,1; | Net.IW[1,1; |

0.6981 | 0.9831 | 0.2691 |

0.6665 | 0.3015 | 0.4228 |

0,1781 | 0.7011 | 0.5479 |

0,1280 | 0,6663 | 0.9427 |

0.9991 | 0.5391 | 0.4177 |

Weights and bias for trainhn |
||

Net.B[11 | Net.LW(2,1; | Net.IW[1,1; |

0.6981 | 0.9831 | 0.2691 |

0.6665 | 0.3015 | 0.4228 |

0,1781 | 0.7011 | 0.5479 |

0,1280 | 0,6663 | 0.9427 |

0.9991 | 0.5391 | 0.4177 |

Weights and bias for trainhn |
||

Net.B[11 | Net.LW(2,1; | Net.IW[1,1; |

0.6981 | 0.9831 | 0.2691 |

0.6665 | 0.3015 | 0.4228 |

0,1781 | 0.7011 | 0.5479 |

0,1280 | 0,6663 | 0.9427 |

0.9991 | 0.5391 | 0.4177 |

**Table 8:** Initial weight and bias of the network for different training algorithm.

In this paper, we design neural network to solve singular perturbed problem. A fast and efficient algorithm (LM) for ANN with one hidden layer has been presented and tested on two examples. Through the comparison with exact solutions show that the ANN method has good accuracy and efficiency and results obtained using the ANN method is with low error.

- O’Malley RE (1991) Singular Perturbation Methods for Ordinary Differential Euations, Springer Verlag, New York.
- Doolan EP, Miller JJH, Schilders WHA (1980) Uniform Numerical Methods for Problems with Initial and Boundary Layers, Boole Press, Dublin, Ireland.
- Roos HG, Stynes M, Tobiska L (1996) Numerical Methods for Singularly Perturbed Differential Equations, Springer, Berlin.
- Miller JJH, O’Riordan E, Shishkin GI (1996) Fitted Numerical Methods for Singular Perturbation Problems, World Scientific, SingaP
_{o}re. - O’Malley RE (1974) Introduction to Singular Perturbations, Academic Press, New York.
- Jianzhong M (1997) Some Singular Singularly Perturbed Problems, Calgary, Alberta, September.
- Galushkin AI (2007) Neural Networks Theory, Springer, Berlin, Germany.
- Ali MH (2012) Design fast feed forward neural networks to solvetwo P
_{o}int boundary value problems.MScthesis, University of Baghdad, College of Education for science/Ibn Al-Haitham. - Mehrotra K, Mohan CK, and Ranka S (1996) Elements of Artificial Neural Networks, Springer, New York, USA.
- Ghaffari A, Abdollahi H, Khoshayand MR, Bozchalooi IS, Dadgar A, et al, (2006) Performance comparison of neural network training algorithms in modeling of bimodal drug delivery. International Journal of Pharmaceutics, 327:126-138.
- Haykin S (1993) Neural networks: A comprehensive foundation.
- Padmajaa P, ReddyYN (2013) A Numerical Patching Method for Solving Singular Perturbation Problems Via Padé Approximates. International Journal of Applied Science and Engineering 1: 51-67.

Select your language of interest to view the total content in your interested language

- Adomian Decomposition Method
- Algebraic Geometry
- Analytical Geometry
- Applied Mathematics
- Axioms
- Balance Law
- Behaviometrics
- Big Data Analytics
- Binary and Non-normal Continuous Data
- Binomial Regression
- Biometrics
- Biostatistics methods
- Clinical Trail
- Complex Analysis
- Computational Model
- Convection Diffusion Equations
- Cross-Covariance and Cross-Correlation
- Differential Equations
- Differential Transform Method
- Fourier Analysis
- Fuzzy Boundary Value
- Fuzzy Environments
- Fuzzy Quasi-Metric Space
- Genetic Linkage
- Hamilton Mechanics
- Hypothesis Testing
- Integrated Analysis
- Integration
- Large-scale Survey Data
- Matrix
- Microarray Studies
- Mixed Initial-boundary Value
- Molecular Modelling
- Multivariate-Normal Model
- Noether's theorem
- Non rigid Image Registration
- Nonlinear Differential Equations
- Number Theory
- Numerical Solutions
- Physical Mathematics
- Quantum Mechanics
- Quantum electrodynamics
- Quasilinear Hyperbolic Systems
- Regressions
- Relativity
- Riemannian Geometry
- Robust Method
- Semi Analytical-Solution
- Sensitivity Analysis
- Smooth Complexities
- Soft biometrics
- Spatial Gaussian Markov Random Fields
- Statistical Methods
- Theoretical Physics
- Theory of Mathematical Modeling
- Three Dimensional Steady State
- Topology
- mirror symmetry
- vector bundle

- Total views:
**11965** - [From(publication date):

May-2014 - Aug 18, 2018] - Breakdown by view type
- HTML page views :
**8160** - PDF downloads :
**3805**

Peer Reviewed Journals

International Conferences 2018-19