Approximation of Functions by Multivariable Hermite Basis: A Hybrid Method

Size: px
Start display at page:

Download "Approximation of Functions by Multivariable Hermite Basis: A Hybrid Method"

Transcription

1 Approximation of Functions by Multivariable Hermite Basis: A Hybrid Method Bartlomiej Beliczynski Warsaw University of Technology, Institute of Control and Industrial Electronics, ul. Koszykowa 75, -66 Warszawa, Poland Bartlomiej.Beliczynski@ee.pw.edu.pl Abstract. In this paper an approximation of multivariable functions by Hermite basis is presented and discussed. Considered here basis is constructed as a product of one-variable Hermite functions with adjustable scaling parameters. The approximation is calculated via hybrid method, the expansion coefficients by using an explicit, non-search formulae, and scaling parameters are determined via a search algorithm. A set of excessive number of Hermite functions is initially calculated. To constitute the approximation basis only those functions are taken which ensure the fastest error decrease down to a desired level. Working examples are presented, demonstrating a very good generalization property of this method. Keywords: Function approximation, Neural networks, Orthonormal basis. Introduction Thanks their elegance and usefulness, for many years Hermite polynomials and Hermite functions have been attractive in various fields of science and engineering. In quantum mechanics of harmonic oscillators, ultra high band telecommunication channels, ECG data compression and various sorts of approximation tasks they proved to be useful tools. A set of Hermite functions forming an orthonormal basis is naturally suitable for approximation, classification and data compression tasks. These basis functions are defined over the real numbers set R and they can be recursively calculated. The approximating function coefficients can be determined relatively easily to achieve the best approximation property. Since Hermite functions are eigenfunctions of the Fourier transform, time and frequency spectra are simultaneously approximated. Each subsequent basis function extends frequency bandwidth within a limited range of well concentrated energy; see for instance []. By introducing a scaling parameter we may control this bandwidth influencing at the same time the dynamic range of the input argument. As pointed out in [] the product of time and frequency bandwidths for Hermite functions, is the largest over set of continuous functions. Hermite functions display various geometrical shapes controlled by simple parameter(s). It was suggested to use Hermite functions as activation functions in A. Dobnikar, U. Lotrič, and B. Šter (Eds.): ICANNGA, Part I, LNCS 6593, pp. 3 39,. c Springer-Verlag Berlin Heidelberg

2 Approximation of Functions by Multivariable Hermite Basis 3 neural schemes. In [3], a so called constructive approximation scheme is used. It is a type of incremental approximation developed in [4], [5]. Every node in the hidden layer has a different activation function. Intuitively the most appropriate shape can be applied. However, in such approach the orthogonality of Hermite functions is not really exploited. If Hermite, one-variable functions are extended into two-variables, the approximation retains the same useful properties and it turns out to be very suitable for image compression tasks. For n-variables case, although main features are the same, the whole process become more complicated. The biggest advantage of approximation by Hermite basis is, that due to its orthonormality, the approximation does not involve search algorithms. However for an initial step of approximation, one has to consider the time and frequency bandwidths. For one-variable case, these two bandwidths could be controlled by a simple scaling parameter which could be selected to some extent arbitrarily. Much more difficult is to choose appropriate scaling parameters in a multivariable case. So we are suggesting to use a search algorithm for that, while the expansion coefficients are calculated explicitly via appropriate formulae. Because approximation by orthonormal basis is numerically very efficient, one can take advantage of that and calculate a larger number of basis functions, then select from them only those which contribute the most to the approximation error decrease. It seems, that this basis selection procedure is the main reason for a good generalization property of this method. This paper is organized as follows. In Section basic facts about approximation needed for later use are recalled. In Section 3 one-variable Hermite functions as basic components for multivariable case, are shortly described. Then we present our results in Section 4, describing multivariable Hermite basis construction, scaling parameters selection, final choice of basis functions and working examples. Finally in Section 5, conclusions are drawn. Approximation Framework Some selected facts on function approximation useful for this paper will be recalled. Let us consider the following function n f n+ = w i g i, () i= where g i G H,andH is a Hilbert space H =(H,. ), i=,..., n, and w i IR,i=,...,n. For any function f from a Hilbert space H and a closed (finite dimensional) subspace G Hwith basis {g,..., g n } there exists a unique best approximation of f by elements of G ([6]). Let us denote it by g b. Because the error of the best approximation is orthogonal to all elements of the approximation space f g b G, the coefficients w i may be calculated from the following set of linear equations g i,f g b =fori =,..., n () where.,. denotes inner product.

3 3 B. Beliczynski The formula () can also be written as g i,f n k= w kg k = g i,f n k= w k g i,g k =fori =,..., n or in the matrix form Γw= G f (3) where Γ =[ g i,g j ], i,j=,..., n, w =[w,..., w n ] T, G f =[ g,f,..., g n,f ] T and T denotes transposition. Because there exists a unique best approximation of f in a n + dimensional space G with basis {g,..., g n }, the matrix Γ is nonsingular and w b = Γ G f. For any basis {g,..., g n } one can find such orthonormal basis {e,..., e n }, e i,e j =wheni = j and e i,e j =wheni j that span{g,..., g n } = span{e,..., e n }. In such a case, Γ is a unit matrix and Finally () will take the form w b = [ e,f, e,f,..., e n,f ] T. (4) f n+ = n e i,f e i, i =,,..., n. (5) i= The squared error error n+ =<f f n,f f n > of the best approximation of a function f in the basis {e,..., e n } is thus expressible by error n+ = f n wi. (6) In a typically stated approximation problem, a basis of n + functions {e,e,..., e n } is given and we are looking for their expansion coefficients w i = e i,f,i=,,..., n. According to formula (6) those expansion coefficients are contributing directly to the error decrease, and they can be used to order the basis from the most to the least significant as far as error decrease is concerned. 3 One-Variable Hermite Functions Our multivariable basis for approximation will be composed from one-variable Hermite functions, so we will briefly describe these components. Let us consider aspacel + (, + ) with the inner product defined <x,y>= x(t)y(t)dt. In such space a sequence of orthonormal functions could be defined as follows (see for instance [6]): h (t),h (t),..., h n (t),... (7) where h n (t) =c n e t Hn (t); H n (t) =( ) n e t and H n (t) isapolynomial. i= dn dt n (e t ); c n = ( n n!. (8) π) /

4 Approximation of Functions by Multivariable Hermite Basis 33 The polynomials H n (t) are called Hermite polynomials and the functions h n (t) Hermite functions. According to (8) the first several Hermite functions could be calculated h (t) = t e π/4 ; h (t) = π /4 e t t; (9) h (t) = π /4 e t (4t ); h 3 (t) = 4 3π /4 e t (8t 3 t) () Plots of several Hermite functions are shown in Fig h h3 h Fig.. Hermite functions h, h, h 9 One can see that increasing of indices of Hermite functions cause enlarging bandwidths in time and frequency. So when approximating a function, it is reasonable to start from lower indices basis functions and gradually go for higher ones. If approximated function is located not in the range of a Hermite function as displayed in Fig., then one can modify the basis (7) by scaling t variable via σ (, ) as a parameter. So if one substitutes t := t σ into (8) and modifies c n to ensure orthonormality, then and h n (t, σ) =c n,σ e t σ H n ( t σ ) where c n,σ = (σ n n! () π) / h n (t, σ) = σ h n ( t σ )and h n (ω, σ) = σ h n (σω) ()

5 34 B. Beliczynski Note that h n as defined by () is the two arguments function whereas h n as defined by (8) has only one argument. These functions are related by (). Thus by introducing scaling parameter σ into () one may adjust both the dynamic range of the input argument h n (t, σ) and its frequency bandwidth t [ σ n +,σ n +]; ω [ n +, n + ] (3) σ σ Suppose that one-variable function f defined over the range of its argument t [ t max,t max ] has to be approximated by using Hermite expansions. Assume that the retained function angular frequency should at least be ω r, then according to (3), the following two conditions should be fulfilled σ n + t max and σ n + ωr (4) or σ [σ l,σ h ] where σ l = t max n + and σ h = (5) n + ω r One would expect that σ l σ h, what is equivalent to t max ω r n + (6) In order to preserve orthonormality of the set {h (t, σ),h (t, σ),..., h n (t, σ)}, σ must be chosen the same for all functions h i (t, σ), i=,..., n. Widely discussed on such occasion the lost of basis orthonormality due to basis truncation, in many practical cases is not crucial [7]. 4 Multivariable Function Approximation 4. Multivariable Hermite Basis Let function to be approximated f belongs to Hilbert space f H, H =(H,. ) and be function of n-variables. Let denote it explicitly as f(x,x,..., x n ). Let one-variable Hermite function be denoted as h i (x j,σ j ), where j {,..., m} and i {,,..., n} (7) and multivariable basis function h l (x,x,..., x m,σ,σ,..., σ m ) be the following h l (x,x,..., x m,σ,σ,..., σ m )=h i (x,σ )h i (x,σ )...h im (x m,σ m ) (8) where i,i,..., i m {,,..., n}. Clearly for each one out of m variables, there are n + indices of Hermite functions. This gives total (n +) m basis functions. They can be enumerated so l {,,..., (n +) m }. l = m i j (n +) j (9) j=

6 Approximation of Functions by Multivariable Hermite Basis 35 Naming now x =(x,x,..., x m )andσ = (σ,σ,..., σ m ), then instead of h l (x,x,..., x m,σ,σ,..., σ m ), we will write in short h l (x, σ) orh l. Finally the multivariable basis is the following { h, h,..., h (n+) m } () One can easily verify that the multivariable basis is orthonormal i.e. { } for i = j hi, h j = elsewhere The approximant f (n+) m of f will be expressed as where f (n+) m(x, σ) = (n+) m l= w l h l (x, σ) () w l = h l,f. () f (n+) m approaches function f if number of elements n goes to infinity f = f = l= w l h l. An interesting survey of math research on multivariables polynomials and Hermite interpolation one can find in [8]. 4. Scaling Parameters Hermite functions are well localized in frequency and time. If a scaling parameter is introduced, it influences both time and frequency ranges but in opposite ways (3). If it is chosen too small, then a fragment of function could poorly be approximated. If it is chosen too large only part of the approximated function spectrum is preserved. If only one-variable function is being approximated, the scaling parameter σ can even intuitively be chosen. If however several variables are involved, the best choice is more complicated and must be calculated. We suggest the following criterion σ =argmin σ f(n+) m(x, σ) f(x) Usually, in order to get σ, a number of iterations is needed. 4.3 Basis Selection If we approximate m-variables function and along each variable we use n + orthonormal components, then it will be (n +) m summation terms in (). For instance if we approximate a 3 variables function with 5 Hermitian components along each variable, then we have 3375 summation terms. One expects that a significant part of all components have a very small, practically negligible influence to the approximation. As clearly visible from formula (6), the components

7 36 B. Beliczynski associated with large wi (or w i ) are contributing the most to the error decrease. So taking advantage of efficiency of approximation by orthonormal basis, we initially calculate an excessive number of Hermite expansion terms and select only the most significant as far as error decrease is concerned. This basis selection can be interpreted as a simple pruning method, a classical neural technique improving generalization, see for instance [9]. 4.4 Examples Example. Let function to be approximated be the following f(x,x )=x e x x. (3) Its plot is presented in Fig.. Let us approximate the function in the range [ 3, 3]. We take 4 points along each axis obtaining totally 68 pairs of the (argument, function value) to be processed. Along each axis number of Hermite components was set to 3, so every one-variable Hermite function could have indices, or. We obtained 3 Hermite components. The expansion coefficients (weights) were calculated according to (). Two scaling factors σ and σ were determined via search-type procedure. Finally we found that σ = σ =.77. The Hermite expansion components () were ordered by squares of their coefficients w i. The first two components are written in (4). f 9 (x,x,σ,σ )=w h (x,σ )h (x,σ )+w 7 h (x,σ )h (x,σ )+... (4) and their expansion coefficients were w =.667 and w 7 =.563e 8. It is clear that to approximate this function it is sufficient to take only one node. Finally the result is the following f (x, σ) =.667h (x, σ), or f (x,x,.77,.77) =.667h (x,.77)h (x,.77) The h and h functions are calculated by using () and (9). Mean Squares Error (MSE) of the approximation is 5.6e, so the approximant is almost exactly the same as the origin. Performance of this approximation is an argument in favour of a good generalization property of this Hermite function based approximation. In fact one can write the following f(x,x )=x e x x π =( )( π 4 π =( )h (x, )h (x, e x σ x )( π 4 e x σ )= )=.667h (x,.77)h (x,.77) what means that generalization from numerical data is almost perfect. We have obtained the function formula which is suitable to be used anywhere, also outside the given region [ 3, 3].

8 Approximation of Functions by Multivariable Hermite Basis 37.5 z.5 4 y 4 4 x 4 Fig.. The original function More demanding generalization experiment is the following. For every function value, the noise signal is randomly generated in the range [.,.] and added to the function. The noised function is presented and Fig Fig. 3. Random noise added to the function values to be used as an input for approximation algorithm

9 38 B. Beliczynski As in the previous case there was only one expansion term sufficient. Because random feature of the experiment, we ran it 5 times, averaging obtained numbers. As the result w =.683, σ =.739, σ =.75 were calculated. Those parameters are very close to the originals. MSE between the original function and the approximation obtained from noisy function was.4e 5, what seems to be very good result of generalization. Example. In this example the function to be approximated is the following f(x,x,x 3 )=x e x x sin(x + x + x 3 ). (5) Let us use again the range [ 3, 3]. We take points along each axis obtaining totally 96 pairs of arguments and function values to be processed. Along each axis, the number of Hermite components was set again to 3, so every one-variable Hermite function could have indices, or. We obtained 3 3 Hermite components. Squares of the expansion coefficients (weights) ordered in nonincreasing order are plotted in Fig.4 It is clear from this plot that 4 out of 7 expansion Hermite terms is sufficient to approximate function (5). MSE between the original function and approximated function is on the level of 3.4e 4. If instead, one takes only out of 7, this ensures 99% of error reduction. When similarly to the previous example a noise generated randomly from the range [.,.] was added and noisy data were used to process function approximation, then again difference between the original function (5) and the approximant (MSE), was on similar level 3.6e 4. Again this is a good sign of generalization ability of this type Hermite based approximation l Fig. 4. Squares of w i () versus l from the most significant to the least

10 5 Conclusions Approximation of Functions by Multivariable Hermite Basis 39 We presented a hybrid method of multivariable function approximation by Hermite basis. The basis is composed from one-variable Hermite functions. Scaling parameters are determined via search algorithm, while expansion coefficients are calculated explicitly from appropriate formulae. Initially we take an excessive number of expansion terms and select only those which contribute the most to the error decrease. This procedure seems to be the reason for a very good generalization property of the method. References. Beliczynski, B.: Properties of the Hermite activation functions in a neural approximation scheme. In: Beliczynski, B., Dzielinski, A., Iwanowski, M., Ribeiro, B. (eds.) ICANNGA 7, Part II. LNCS, vol. 443, pp Springer, Heidelberg (7). Hlawatsch, F.: Time-Frequency Analysis and Synthesis of Linear Signal Spaces. Kluwer Academic Publishers, Dordrecht (998) 3. Ma, L., Khorasani, K.: Constructive feedforward neural networks using Hermite polynomial activation functions. IEEE Transactions on Neural Networks 6, (5) 4. Kwok, T., Yeung, D.: Constructive algorithms for structure learning in feedforward neural networks for regression problems. IEEE Trans. Neural Netw. 8(3), (997) 5. Kwok, T., Yeung, D.: Objective functions for training new hidden units in constructive neural networks. IEEE Trans. Neural Networks 8(5), 3 48 (997) 6. Kreyszig, E.: Introductory functional analysis with applications. J. Wiley, Chichester (978) 7. Beliczynski, B., Ribeiro, B.: Some enhanencement to approximation of one-variable functions by orthonormal basis. Neural Network World 9, 4 4 (9) 8. Lorentz, R.: Multivariate hermite interpolation by algebraic polynomials: A survey. Journal of Computational and Applied Mathematics, 67 () 9. Reed, R.: Pruning algorithms - a survey. IEEE Trans. on Neural Networks 4(5), (993)

On Riesz-Fischer sequences and lower frame bounds

On Riesz-Fischer sequences and lower frame bounds On Riesz-Fischer sequences and lower frame bounds P. Casazza, O. Christensen, S. Li, A. Lindner Abstract We investigate the consequences of the lower frame condition and the lower Riesz basis condition

More information

An Adaptively Constructing Multilayer Feedforward Neural Networks Using Hermite Polynomials

An Adaptively Constructing Multilayer Feedforward Neural Networks Using Hermite Polynomials An Adaptively Constructing Multilayer Feedforward Neural Networks Using Hermite Polynomials L. Ma 1 and K. Khorasani 2 1 Department of Applied Computer Science, Tokyo Polytechnic University, 1583 Iiyama,

More information

Approximating the Best Linear Unbiased Estimator of Non-Gaussian Signals with Gaussian Noise

Approximating the Best Linear Unbiased Estimator of Non-Gaussian Signals with Gaussian Noise IEICE Transactions on Information and Systems, vol.e91-d, no.5, pp.1577-1580, 2008. 1 Approximating the Best Linear Unbiased Estimator of Non-Gaussian Signals with Gaussian Noise Masashi Sugiyama (sugi@cs.titech.ac.jp)

More information

Classification of Ordinal Data Using Neural Networks

Classification of Ordinal Data Using Neural Networks Classification of Ordinal Data Using Neural Networks Joaquim Pinto da Costa and Jaime S. Cardoso 2 Faculdade Ciências Universidade Porto, Porto, Portugal jpcosta@fc.up.pt 2 Faculdade Engenharia Universidade

More information

Object Recognition Using Local Characterisation and Zernike Moments

Object Recognition Using Local Characterisation and Zernike Moments Object Recognition Using Local Characterisation and Zernike Moments A. Choksuriwong, H. Laurent, C. Rosenberger, and C. Maaoui Laboratoire Vision et Robotique - UPRES EA 2078, ENSI de Bourges - Université

More information

p(d θ ) l(θ ) 1.2 x x x

p(d θ ) l(θ ) 1.2 x x x p(d θ ).2 x 0-7 0.8 x 0-7 0.4 x 0-7 l(θ ) -20-40 -60-80 -00 2 3 4 5 6 7 θ ˆ 2 3 4 5 6 7 θ ˆ 2 3 4 5 6 7 θ θ x FIGURE 3.. The top graph shows several training points in one dimension, known or assumed to

More information

NEW BOUNDS FOR TRUNCATION-TYPE ERRORS ON REGULAR SAMPLING EXPANSIONS

NEW BOUNDS FOR TRUNCATION-TYPE ERRORS ON REGULAR SAMPLING EXPANSIONS NEW BOUNDS FOR TRUNCATION-TYPE ERRORS ON REGULAR SAMPLING EXPANSIONS Nikolaos D. Atreas Department of Mathematics, Aristotle University of Thessaloniki, 54006, Greece, e-mail:natreas@auth.gr Abstract We

More information

Separation of Variables in Linear PDE: One-Dimensional Problems

Separation of Variables in Linear PDE: One-Dimensional Problems Separation of Variables in Linear PDE: One-Dimensional Problems Now we apply the theory of Hilbert spaces to linear differential equations with partial derivatives (PDE). We start with a particular example,

More information

The Quantum Harmonic Oscillator

The Quantum Harmonic Oscillator The Classical Analysis Recall the mass-spring system where we first introduced unforced harmonic motion. The DE that describes the system is: where: Note that throughout this discussion the variables =

More information

PART II : Least-Squares Approximation

PART II : Least-Squares Approximation PART II : Least-Squares Approximation Basic theory Let U be an inner product space. Let V be a subspace of U. For any g U, we look for a least-squares approximation of g in the subspace V min f V f g 2,

More information

In the Name of God. Lectures 15&16: Radial Basis Function Networks

In the Name of God. Lectures 15&16: Radial Basis Function Networks 1 In the Name of God Lectures 15&16: Radial Basis Function Networks Some Historical Notes Learning is equivalent to finding a surface in a multidimensional space that provides a best fit to the training

More information

MATHEMATICAL METHODS AND APPLIED COMPUTING

MATHEMATICAL METHODS AND APPLIED COMPUTING Numerical Approximation to Multivariate Functions Using Fluctuationlessness Theorem with a Trigonometric Basis Function to Deal with Highly Oscillatory Functions N.A. BAYKARA Marmara University Department

More information

Nonlinear Filtering Revisited: a Spectral Approach, II

Nonlinear Filtering Revisited: a Spectral Approach, II Nonlinear Filtering Revisited: a Spectral Approach, II Sergey V. Lototsky Center for Applied Mathematical Sciences University of Southern California Los Angeles, CA 90089-3 sergey@cams-00.usc.edu Remijigus

More information

Quantum Mechanics-I Prof. Dr. S. Lakshmi Bala Department of Physics Indian Institute of Technology, Madras. Lecture - 21 Square-Integrable Functions

Quantum Mechanics-I Prof. Dr. S. Lakshmi Bala Department of Physics Indian Institute of Technology, Madras. Lecture - 21 Square-Integrable Functions Quantum Mechanics-I Prof. Dr. S. Lakshmi Bala Department of Physics Indian Institute of Technology, Madras Lecture - 21 Square-Integrable Functions (Refer Slide Time: 00:06) (Refer Slide Time: 00:14) We

More information

Angular Momentum in Quantum Mechanics

Angular Momentum in Quantum Mechanics Angular Momentum in Quantum Mechanics In classical mechanics the angular momentum L = r p of any particle moving in a central field of force is conserved. For the reduced two-body problem this is the content

More information

Discrete Orthogonal Harmonic Transforms

Discrete Orthogonal Harmonic Transforms Discrete Orthogonal Harmonic Transforms Speaker: Chun-Lin Liu, Advisor: Soo-Chang Pei Ph. D Image Processing Laboratory, EEII 530, Graduate Institute of Communication Engineering, National Taiwan University.

More information

Functional Preprocessing for Multilayer Perceptrons

Functional Preprocessing for Multilayer Perceptrons Functional Preprocessing for Multilayer Perceptrons Fabrice Rossi and Brieuc Conan-Guez Projet AxIS, INRIA, Domaine de Voluceau, Rocquencourt, B.P. 105 78153 Le Chesnay Cedex, France CEREMADE, UMR CNRS

More information

Neural Networks and the Back-propagation Algorithm

Neural Networks and the Back-propagation Algorithm Neural Networks and the Back-propagation Algorithm Francisco S. Melo In these notes, we provide a brief overview of the main concepts concerning neural networks and the back-propagation algorithm. We closely

More information

Bobby Hunt, Mariappan S. Nadar, Paul Keller, Eric VonColln, and Anupam Goyal III. ASSOCIATIVE RECALL BY A POLYNOMIAL MAPPING

Bobby Hunt, Mariappan S. Nadar, Paul Keller, Eric VonColln, and Anupam Goyal III. ASSOCIATIVE RECALL BY A POLYNOMIAL MAPPING Synthesis of a Nonrecurrent Associative Memory Model Based on a Nonlinear Transformation in the Spectral Domain p. 1 Bobby Hunt, Mariappan S. Nadar, Paul Keller, Eric VonColln, Anupam Goyal Abstract -

More information

The Nearest Doubly Stochastic Matrix to a Real Matrix with the same First Moment

The Nearest Doubly Stochastic Matrix to a Real Matrix with the same First Moment he Nearest Doubly Stochastic Matrix to a Real Matrix with the same First Moment William Glunt 1, homas L. Hayden 2 and Robert Reams 2 1 Department of Mathematics and Computer Science, Austin Peay State

More information

Introduction to Machine Learning

Introduction to Machine Learning 10-701 Introduction to Machine Learning PCA Slides based on 18-661 Fall 2018 PCA Raw data can be Complex, High-dimensional To understand a phenomenon we measure various related quantities If we knew what

More information

Lecture 4.6: Some special orthogonal functions

Lecture 4.6: Some special orthogonal functions Lecture 4.6: Some special orthogonal functions Matthew Macauley Department of Mathematical Sciences Clemson University http://www.math.clemson.edu/~macaule/ Math 4340, Advanced Engineering Mathematics

More information

One-dimensional harmonic oscillator. -motivation. -equation, energy levels. -eigenfunctions, Hermite polynomials. -classical analogy

One-dimensional harmonic oscillator. -motivation. -equation, energy levels. -eigenfunctions, Hermite polynomials. -classical analogy One-dimensional harmonic oscillator -motivation -equation, energy levels -eigenfunctions, Hermite polynomials -classical analogy One-dimensional harmonic oscillator 05/0 Harmonic oscillator = potential

More information

Active Sonar Target Classification Using Classifier Ensembles

Active Sonar Target Classification Using Classifier Ensembles International Journal of Engineering Research and Technology. ISSN 0974-3154 Volume 11, Number 12 (2018), pp. 2125-2133 International Research Publication House http://www.irphouse.com Active Sonar Target

More information

Scientific Computing: An Introductory Survey

Scientific Computing: An Introductory Survey Scientific Computing: An Introductory Survey Chapter 7 Interpolation Prof. Michael T. Heath Department of Computer Science University of Illinois at Urbana-Champaign Copyright c 2002. Reproduction permitted

More information

L-statistics based Modification of Reconstruction Algorithms for Compressive Sensing in the Presence of Impulse Noise

L-statistics based Modification of Reconstruction Algorithms for Compressive Sensing in the Presence of Impulse Noise L-statistics based Modification of Reconstruction Algorithms for Compressive Sensing in the Presence of Impulse Noise Srdjan Stanković, Irena Orović and Moeness Amin 1 Abstract- A modification of standard

More information

A Novel Approach to the 2D Analytic Signal? Thomas Bulow and Gerald Sommer. Christian{Albrechts{Universitat zu Kiel

A Novel Approach to the 2D Analytic Signal? Thomas Bulow and Gerald Sommer. Christian{Albrechts{Universitat zu Kiel A Novel Approach to the 2D Analytic Signal? Thomas Bulow and Gerald Sommer Christian{Albrechts{Universitat zu Kiel Institute of Computer Science, Cognitive Systems Preuerstrae 1{9, 24105 Kiel Tel:+49 431

More information

Fitting Aggregation Functions to Data: Part II Idempotization

Fitting Aggregation Functions to Data: Part II Idempotization Fitting Aggregation Functions to Data: Part II Idempotization Maciej Bartoszuk 1, Gleb Beliakov 2, Marek Gagolewski 3,1, and Simon James 2 1 Faculty of Mathematics and Information Science, Warsaw University

More information

CS 450 Numerical Analysis. Chapter 8: Numerical Integration and Differentiation

CS 450 Numerical Analysis. Chapter 8: Numerical Integration and Differentiation Lecture slides based on the textbook Scientific Computing: An Introductory Survey by Michael T. Heath, copyright c 2018 by the Society for Industrial and Applied Mathematics. http://www.siam.org/books/cl80

More information

Lecture 3 Dynamics 29

Lecture 3 Dynamics 29 Lecture 3 Dynamics 29 30 LECTURE 3. DYNAMICS 3.1 Introduction Having described the states and the observables of a quantum system, we shall now introduce the rules that determine their time evolution.

More information

Fourier and Wavelet Signal Processing

Fourier and Wavelet Signal Processing Ecole Polytechnique Federale de Lausanne (EPFL) Audio-Visual Communications Laboratory (LCAV) Fourier and Wavelet Signal Processing Martin Vetterli Amina Chebira, Ali Hormati Spring 2011 2/25/2011 1 Outline

More information

arxiv: v1 [quant-ph] 8 Sep 2010

arxiv: v1 [quant-ph] 8 Sep 2010 Few-Body Systems, (8) Few- Body Systems c by Springer-Verlag 8 Printed in Austria arxiv:9.48v [quant-ph] 8 Sep Two-boson Correlations in Various One-dimensional Traps A. Okopińska, P. Kościk Institute

More information

1 Distributions (due January 22, 2009)

1 Distributions (due January 22, 2009) Distributions (due January 22, 29). The distribution derivative of the locally integrable function ln( x ) is the principal value distribution /x. We know that, φ = lim φ(x) dx. x ɛ x Show that x, φ =

More information

Lie algebraic quantum control mechanism analysis

Lie algebraic quantum control mechanism analysis Lie algebraic quantum control mechanism analysis June 4, 2 Existing methods for quantum control mechanism identification do not exploit analytic features of control systems to delineate mechanism (or,

More information

Lecture 19 (Nov. 15, 2017)

Lecture 19 (Nov. 15, 2017) Lecture 19 8.31 Quantum Theory I, Fall 017 8 Lecture 19 Nov. 15, 017) 19.1 Rotations Recall that rotations are transformations of the form x i R ij x j using Einstein summation notation), where R is an

More information

CS 6375 Machine Learning

CS 6375 Machine Learning CS 6375 Machine Learning Decision Trees Instructor: Yang Liu 1 Supervised Classifier X 1 X 2. X M Ref class label 2 1 Three variables: Attribute 1: Hair = {blond, dark} Attribute 2: Height = {tall, short}

More information

Computing Phase Noise Eigenfunctions Directly from Steady-State Jacobian Matrices

Computing Phase Noise Eigenfunctions Directly from Steady-State Jacobian Matrices Computing Phase Noise Eigenfunctions Directly from Steady-State Jacobian Matrices Alper Demir David Long Jaijeet Roychowdhury Bell Laboratories Murray Hill New Jersey USA Abstract The main effort in oscillator

More information

Reproducing Kernel Hilbert Spaces

Reproducing Kernel Hilbert Spaces 9.520: Statistical Learning Theory and Applications February 10th, 2010 Reproducing Kernel Hilbert Spaces Lecturer: Lorenzo Rosasco Scribe: Greg Durrett 1 Introduction In the previous two lectures, we

More information

THE information capacity is one of the most important

THE information capacity is one of the most important 256 IEEE TRANSACTIONS ON INFORMATION THEORY, VOL. 44, NO. 1, JANUARY 1998 Capacity of Two-Layer Feedforward Neural Networks with Binary Weights Chuanyi Ji, Member, IEEE, Demetri Psaltis, Senior Member,

More information

Unsupervised Clustering of Human Pose Using Spectral Embedding

Unsupervised Clustering of Human Pose Using Spectral Embedding Unsupervised Clustering of Human Pose Using Spectral Embedding Muhammad Haseeb and Edwin R Hancock Department of Computer Science, The University of York, UK Abstract In this paper we use the spectra of

More information

Solutions: Problem Set 3 Math 201B, Winter 2007

Solutions: Problem Set 3 Math 201B, Winter 2007 Solutions: Problem Set 3 Math 201B, Winter 2007 Problem 1. Prove that an infinite-dimensional Hilbert space is a separable metric space if and only if it has a countable orthonormal basis. Solution. If

More information

Decision Trees (Cont.)

Decision Trees (Cont.) Decision Trees (Cont.) R&N Chapter 18.2,18.3 Side example with discrete (categorical) attributes: Predicting age (3 values: less than 30, 30-45, more than 45 yrs old) from census data. Attributes (split

More information

Radial-Basis Function Networks

Radial-Basis Function Networks Radial-Basis Function etworks A function is radial () if its output depends on (is a nonincreasing function of) the distance of the input from a given stored vector. s represent local receptors, as illustrated

More information

CSC242: Intro to AI. Lecture 21

CSC242: Intro to AI. Lecture 21 CSC242: Intro to AI Lecture 21 Administrivia Project 4 (homeworks 18 & 19) due Mon Apr 16 11:59PM Posters Apr 24 and 26 You need an idea! You need to present it nicely on 2-wide by 4-high landscape pages

More information

NUMERICAL SOLUTION OF DELAY DIFFERENTIAL EQUATIONS VIA HAAR WAVELETS

NUMERICAL SOLUTION OF DELAY DIFFERENTIAL EQUATIONS VIA HAAR WAVELETS TWMS J Pure Appl Math V5, N2, 24, pp22-228 NUMERICAL SOLUTION OF DELAY DIFFERENTIAL EQUATIONS VIA HAAR WAVELETS S ASADI, AH BORZABADI Abstract In this paper, Haar wavelet benefits are applied to the delay

More information

Neural Network Weight Space Symmetries Can Speed up Genetic Learning

Neural Network Weight Space Symmetries Can Speed up Genetic Learning Neural Network Weight Space Symmetries Can Speed up Genetic Learning ROMAN NERUDA Λ Institue of Computer Science Academy of Sciences of the Czech Republic P.O. Box 5, 187 Prague, Czech Republic tel: (4)665375,fax:(4)8585789

More information

Lecture 3: Central Limit Theorem

Lecture 3: Central Limit Theorem Lecture 3: Central Limit Theorem Scribe: Jacy Bird (Division of Engineering and Applied Sciences, Harvard) February 8, 003 The goal of today s lecture is to investigate the asymptotic behavior of P N (

More information

Vector Space Models. wine_spectral.r

Vector Space Models. wine_spectral.r Vector Space Models 137 wine_spectral.r Latent Semantic Analysis Problem with words Even a small vocabulary as in wine example is challenging LSA Reduce number of columns of DTM by principal components

More information

Widely Linear Estimation with Complex Data

Widely Linear Estimation with Complex Data Widely Linear Estimation with Complex Data Bernard Picinbono, Pascal Chevalier To cite this version: Bernard Picinbono, Pascal Chevalier. Widely Linear Estimation with Complex Data. IEEE Transactions on

More information

APPENDIX A. Background Mathematics. A.1 Linear Algebra. Vector algebra. Let x denote the n-dimensional column vector with components x 1 x 2.

APPENDIX A. Background Mathematics. A.1 Linear Algebra. Vector algebra. Let x denote the n-dimensional column vector with components x 1 x 2. APPENDIX A Background Mathematics A. Linear Algebra A.. Vector algebra Let x denote the n-dimensional column vector with components 0 x x 2 B C @. A x n Definition 6 (scalar product). The scalar product

More information

Lecture Notes Special: Using Neural Networks for Mean-Squared Estimation. So: How can we evaluate E(X Y )? What is a neural network? How is it useful?

Lecture Notes Special: Using Neural Networks for Mean-Squared Estimation. So: How can we evaluate E(X Y )? What is a neural network? How is it useful? Lecture Notes Special: Using Neural Networks for Mean-Squared Estimation The Story So Far... So: How can we evaluate E(X Y )? What is a neural network? How is it useful? EE 278: Using Neural Networks for

More information

MATH Linear Algebra

MATH Linear Algebra MATH 304 - Linear Algebra In the previous note we learned an important algorithm to produce orthogonal sequences of vectors called the Gramm-Schmidt orthogonalization process. Gramm-Schmidt orthogonalization

More information

Gaussian processes. Chuong B. Do (updated by Honglak Lee) November 22, 2008

Gaussian processes. Chuong B. Do (updated by Honglak Lee) November 22, 2008 Gaussian processes Chuong B Do (updated by Honglak Lee) November 22, 2008 Many of the classical machine learning algorithms that we talked about during the first half of this course fit the following pattern:

More information

arxiv:math/ v2 [math.ap] 3 Oct 2006

arxiv:math/ v2 [math.ap] 3 Oct 2006 THE TAYLOR SERIES OF THE GAUSSIAN KERNEL arxiv:math/0606035v2 [math.ap] 3 Oct 2006 L. ESCAURIAZA From some people one can learn more than mathematics Abstract. We describe a formula for the Taylor series

More information

arxiv: v3 [math.ca] 20 Aug 2015

arxiv: v3 [math.ca] 20 Aug 2015 A note on mean-value properties of harmonic functions on the hypercube arxiv:506.0703v3 [math.ca] 20 Aug 205 P. P. Petrov,a a Faculty of Mathematics and Informatics, Sofia University, 5 James Bourchier

More information

Statistical Methods for SVM

Statistical Methods for SVM Statistical Methods for SVM Support Vector Machines Here we approach the two-class classification problem in a direct way: We try and find a plane that separates the classes in feature space. If we cannot,

More information

Dunkl operators and Clifford algebras II

Dunkl operators and Clifford algebras II Translation operator for the Clifford Research Group Department of Mathematical Analysis Ghent University Hong Kong, March, 2011 Translation operator for the Hermite polynomials Translation operator for

More information

Gaussian interval quadrature rule for exponential weights

Gaussian interval quadrature rule for exponential weights Gaussian interval quadrature rule for exponential weights Aleksandar S. Cvetković, a, Gradimir V. Milovanović b a Department of Mathematics, Faculty of Mechanical Engineering, University of Belgrade, Kraljice

More information

Vectors in Function Spaces

Vectors in Function Spaces Jim Lambers MAT 66 Spring Semester 15-16 Lecture 18 Notes These notes correspond to Section 6.3 in the text. Vectors in Function Spaces We begin with some necessary terminology. A vector space V, also

More information

Application of the Bernstein Polynomials for Solving Volterra Integral Equations with Convolution Kernels

Application of the Bernstein Polynomials for Solving Volterra Integral Equations with Convolution Kernels Filomat 3:4 (216), 145 152 DOI 1.2298/FIL16445A Published by Faculty of Sciences and Mathematics, University of Niš, Serbia Available at: http://www.pmf.ni.ac.rs/filomat Application of the Bernstein Polynomials

More information

1. Background: The SVD and the best basis (questions selected from Ch. 6- Can you fill in the exercises?)

1. Background: The SVD and the best basis (questions selected from Ch. 6- Can you fill in the exercises?) Math 35 Exam Review SOLUTIONS Overview In this third of the course we focused on linear learning algorithms to model data. summarize: To. Background: The SVD and the best basis (questions selected from

More information

Support Vector Machines

Support Vector Machines Support Vector Machines Here we approach the two-class classification problem in a direct way: We try and find a plane that separates the classes in feature space. If we cannot, we get creative in two

More information

REVIEW: The Matching Method Algorithm

REVIEW: The Matching Method Algorithm Lecture 26: Numerov Algorithm for Solving the Time-Independent Schrödinger Equation 1 REVIEW: The Matching Method Algorithm Need for a more general method The shooting method for solving the time-independent

More information

Recall that any inner product space V has an associated norm defined by

Recall that any inner product space V has an associated norm defined by Hilbert Spaces Recall that any inner product space V has an associated norm defined by v = v v. Thus an inner product space can be viewed as a special kind of normed vector space. In particular every inner

More information

Radial-Basis Function Networks

Radial-Basis Function Networks Radial-Basis Function etworks A function is radial basis () if its output depends on (is a non-increasing function of) the distance of the input from a given stored vector. s represent local receptors,

More information

Matrices and Vectors. Definition of Matrix. An MxN matrix A is a two-dimensional array of numbers A =

Matrices and Vectors. Definition of Matrix. An MxN matrix A is a two-dimensional array of numbers A = 30 MATHEMATICS REVIEW G A.1.1 Matrices and Vectors Definition of Matrix. An MxN matrix A is a two-dimensional array of numbers A = a 11 a 12... a 1N a 21 a 22... a 2N...... a M1 a M2... a MN A matrix can

More information

x 3 x 1 ix 2 x 1 + ix 2 x 3

x 3 x 1 ix 2 x 1 + ix 2 x 3 Peeter Joot peeterjoot@pm.me PHY2403H Quantum Field Theory. Lecture 19: Pauli matrices, Weyl spinors, SL2,c, Weyl action, Weyl equation, Dirac matrix, Dirac action, Dirac Lagrangian. Taught by Prof. Erich

More information

Linear Algebra: Matrix Eigenvalue Problems

Linear Algebra: Matrix Eigenvalue Problems CHAPTER8 Linear Algebra: Matrix Eigenvalue Problems Chapter 8 p1 A matrix eigenvalue problem considers the vector equation (1) Ax = λx. 8.0 Linear Algebra: Matrix Eigenvalue Problems Here A is a given

More information

Performance Evaluation of Generalized Polynomial Chaos

Performance Evaluation of Generalized Polynomial Chaos Performance Evaluation of Generalized Polynomial Chaos Dongbin Xiu, Didier Lucor, C.-H. Su, and George Em Karniadakis 1 Division of Applied Mathematics, Brown University, Providence, RI 02912, USA, gk@dam.brown.edu

More information

3 Orthogonality and Fourier series

3 Orthogonality and Fourier series 3 Orthogonality and Fourier series We now turn to the concept of orthogonality which is a key concept in inner product spaces and Hilbert spaces. We start with some basic definitions. Definition 3.1. Let

More information

Cooperative Communication with Feedback via Stochastic Approximation

Cooperative Communication with Feedback via Stochastic Approximation Cooperative Communication with Feedback via Stochastic Approximation Utsaw Kumar J Nicholas Laneman and Vijay Gupta Department of Electrical Engineering University of Notre Dame Email: {ukumar jnl vgupta}@ndedu

More information

Solving Linear Time Varying Systems by Orthonormal Bernstein Polynomials

Solving Linear Time Varying Systems by Orthonormal Bernstein Polynomials Science Journal of Applied Mathematics and Statistics 2015; 3(4): 194-198 Published online July 27, 2015 (http://www.sciencepublishinggroup.com/j/sjams) doi: 10.11648/j.sjams.20150304.15 ISSN: 2376-9491

More information

Lecture 5: Logistic Regression. Neural Networks

Lecture 5: Logistic Regression. Neural Networks Lecture 5: Logistic Regression. Neural Networks Logistic regression Comparison with generative models Feed-forward neural networks Backpropagation Tricks for training neural networks COMP-652, Lecture

More information

Eigenmodes for coupled harmonic vibrations. Algebraic Method for Harmonic Oscillator.

Eigenmodes for coupled harmonic vibrations. Algebraic Method for Harmonic Oscillator. PHYS208 spring 2008 Eigenmodes for coupled harmonic vibrations. Algebraic Method for Harmonic Oscillator. 07.02.2008 Adapted from the text Light - Atom Interaction PHYS261 autumn 2007 Go to list of topics

More information

Quantum Mechanics for Scientists and Engineers. David Miller

Quantum Mechanics for Scientists and Engineers. David Miller Quantum Mechanics for Scientists and Engineers David Miller Vector spaces, operators and matrices Vector spaces, operators and matrices Vector space Vector space We need a space in which our vectors exist

More information

Lecture 6 Quantum Mechanical Systems and Measurements

Lecture 6 Quantum Mechanical Systems and Measurements Lecture 6 Quantum Mechanical Systems and Measurements Today s Program: 1. Simple Harmonic Oscillator (SHO). Principle of spectral decomposition. 3. Predicting the results of measurements, fourth postulate

More information

Functional Radial Basis Function Networks (FRBFN)

Functional Radial Basis Function Networks (FRBFN) Bruges (Belgium), 28-3 April 24, d-side publi., ISBN 2-9337-4-8, pp. 313-318 Functional Radial Basis Function Networks (FRBFN) N. Delannay 1, F. Rossi 2, B. Conan-Guez 2, M. Verleysen 1 1 Université catholique

More information

CS 195-5: Machine Learning Problem Set 1

CS 195-5: Machine Learning Problem Set 1 CS 95-5: Machine Learning Problem Set Douglas Lanman dlanman@brown.edu 7 September Regression Problem Show that the prediction errors y f(x; ŵ) are necessarily uncorrelated with any linear function of

More information

Feature Extraction with Weighted Samples Based on Independent Component Analysis

Feature Extraction with Weighted Samples Based on Independent Component Analysis Feature Extraction with Weighted Samples Based on Independent Component Analysis Nojun Kwak Samsung Electronics, Suwon P.O. Box 105, Suwon-Si, Gyeonggi-Do, KOREA 442-742, nojunk@ieee.org, WWW home page:

More information

Degree Master of Science in Mathematical Modelling and Scientific Computing Mathematical Methods I Thursday, 12th January 2012, 9:30 a.m.- 11:30 a.m.

Degree Master of Science in Mathematical Modelling and Scientific Computing Mathematical Methods I Thursday, 12th January 2012, 9:30 a.m.- 11:30 a.m. Degree Master of Science in Mathematical Modelling and Scientific Computing Mathematical Methods I Thursday, 12th January 2012, 9:30 a.m.- 11:30 a.m. Candidates should submit answers to a maximum of four

More information

A Simple Implementation of the Stochastic Discrimination for Pattern Recognition

A Simple Implementation of the Stochastic Discrimination for Pattern Recognition A Simple Implementation of the Stochastic Discrimination for Pattern Recognition Dechang Chen 1 and Xiuzhen Cheng 2 1 University of Wisconsin Green Bay, Green Bay, WI 54311, USA chend@uwgb.edu 2 University

More information

d 1 µ 2 Θ = 0. (4.1) consider first the case of m = 0 where there is no azimuthal dependence on the angle φ.

d 1 µ 2 Θ = 0. (4.1) consider first the case of m = 0 where there is no azimuthal dependence on the angle φ. 4 Legendre Functions In order to investigate the solutions of Legendre s differential equation d ( µ ) dθ ] ] + l(l + ) m dµ dµ µ Θ = 0. (4.) consider first the case of m = 0 where there is no azimuthal

More information

A Reverse Technique for Lumping High Dimensional Model Representation Method

A Reverse Technique for Lumping High Dimensional Model Representation Method A Reverse Technique for Lumping High Dimensional Model Representation Method M. ALPER TUNGA Bahçeşehir University Department of Software Engineering Beşiktaş, 34349, İstanbul TURKEY TÜRKİYE) alper.tunga@bahcesehir.edu.tr

More information

arxiv: v1 [cond-mat.stat-mech] 14 Apr 2009

arxiv: v1 [cond-mat.stat-mech] 14 Apr 2009 arxiv:0904.2126v1 [cond-mat.stat-mech] 14 Apr 2009 Critical exponents for Gaussian fixed point of renormalization Witold Haliniak 1 and Wojciech Wislicki 2 1 University of Warsaw, Faculty of Mathematics,

More information

Decomposing Bent Functions

Decomposing Bent Functions 2004 IEEE TRANSACTIONS ON INFORMATION THEORY, VOL. 49, NO. 8, AUGUST 2003 Decomposing Bent Functions Anne Canteaut and Pascale Charpin Abstract In a recent paper [1], it is shown that the restrictions

More information

MATH 590: Meshfree Methods

MATH 590: Meshfree Methods MATH 590: Meshfree Methods The Connection to Green s Kernels Greg Fasshauer Department of Applied Mathematics Illinois Institute of Technology Fall 2014 fasshauer@iit.edu MATH 590 1 Outline 1 Introduction

More information

a subset of these N input variables. A naive method is to train a new neural network on this subset to determine this performance. Instead of the comp

a subset of these N input variables. A naive method is to train a new neural network on this subset to determine this performance. Instead of the comp Input Selection with Partial Retraining Pierre van de Laar, Stan Gielen, and Tom Heskes RWCP? Novel Functions SNN?? Laboratory, Dept. of Medical Physics and Biophysics, University of Nijmegen, The Netherlands.

More information

Pattern Recognition Prof. P. S. Sastry Department of Electronics and Communication Engineering Indian Institute of Science, Bangalore

Pattern Recognition Prof. P. S. Sastry Department of Electronics and Communication Engineering Indian Institute of Science, Bangalore Pattern Recognition Prof. P. S. Sastry Department of Electronics and Communication Engineering Indian Institute of Science, Bangalore Lecture - 27 Multilayer Feedforward Neural networks with Sigmoidal

More information

Advanced Spectroscopy. Dr. P. Hunt Rm 167 (Chemistry) web-site:

Advanced Spectroscopy. Dr. P. Hunt Rm 167 (Chemistry) web-site: Advanced Spectroscopy Dr. P. Hunt p.hunt@imperial.ac.uk Rm 167 (Chemistry) web-site: http://www.ch.ic.ac.uk/hunt Maths! Coordinate transformations rotations! example 18.1 p501 whole chapter on Matrices

More information

A Necessary Condition for Learning from Positive Examples

A Necessary Condition for Learning from Positive Examples Machine Learning, 5, 101-113 (1990) 1990 Kluwer Academic Publishers. Manufactured in The Netherlands. A Necessary Condition for Learning from Positive Examples HAIM SHVAYTSER* (HAIM%SARNOFF@PRINCETON.EDU)

More information

Cheng Soon Ong & Christian Walder. Canberra February June 2018

Cheng Soon Ong & Christian Walder. Canberra February June 2018 Cheng Soon Ong & Christian Walder Research Group and College of Engineering and Computer Science Canberra February June 2018 Outlines Overview Introduction Linear Algebra Probability Linear Regression

More information

VC-dimension of a context-dependent perceptron

VC-dimension of a context-dependent perceptron 1 VC-dimension of a context-dependent perceptron Piotr Ciskowski Institute of Engineering Cybernetics, Wroc law University of Technology, Wybrzeże Wyspiańskiego 27, 50 370 Wroc law, Poland cis@vectra.ita.pwr.wroc.pl

More information

SOME FUNDAMENTAL THEOREMS IN BANACH SPACES AND HILBERT SPACES

SOME FUNDAMENTAL THEOREMS IN BANACH SPACES AND HILBERT SPACES SOME FUNDAMENTAL THEOREMS IN BANACH SPACES AND HILBERT SPACES Sanjay Kumar Department of Mathematics Central University of Jammu, India sanjaykmath@gmail.com Sanjay Kumar (Central University of Jammu)

More information

LECTURE # - NEURAL COMPUTATION, Feb 04, Linear Regression. x 1 θ 1 output... θ M x M. Assumes a functional form

LECTURE # - NEURAL COMPUTATION, Feb 04, Linear Regression. x 1 θ 1 output... θ M x M. Assumes a functional form LECTURE # - EURAL COPUTATIO, Feb 4, 4 Linear Regression Assumes a functional form f (, θ) = θ θ θ K θ (Eq) where = (,, ) are the attributes and θ = (θ, θ, θ ) are the function parameters Eample: f (, θ)

More information

y(n) Time Series Data

y(n) Time Series Data Recurrent SOM with Local Linear Models in Time Series Prediction Timo Koskela, Markus Varsta, Jukka Heikkonen, and Kimmo Kaski Helsinki University of Technology Laboratory of Computational Engineering

More information

Differential Equations

Differential Equations Electricity and Magnetism I (P331) M. R. Shepherd October 14, 2008 Differential Equations The purpose of this note is to provide some supplementary background on differential equations. The problems discussed

More information

Conjugate Directions for Stochastic Gradient Descent

Conjugate Directions for Stochastic Gradient Descent Conjugate Directions for Stochastic Gradient Descent Nicol N Schraudolph Thore Graepel Institute of Computational Science ETH Zürich, Switzerland {schraudo,graepel}@infethzch Abstract The method of conjugate

More information

Examination paper for TMA4215 Numerical Mathematics

Examination paper for TMA4215 Numerical Mathematics Department of Mathematical Sciences Examination paper for TMA425 Numerical Mathematics Academic contact during examination: Trond Kvamsdal Phone: 93058702 Examination date: 6th of December 207 Examination

More information

Iterative Construction of Sparse Polynomial Approximations

Iterative Construction of Sparse Polynomial Approximations Iterative Construction of Sparse Polynomial Approximations Terence D. Sanger Massachusetts Institute of Technology Room E25-534 Cambridge, MA 02139 tds@ai.mit.edu Richard S. Sutton GTE Laboratories Incorporated

More information

Deep Learning: Approximation of Functions by Composition

Deep Learning: Approximation of Functions by Composition Deep Learning: Approximation of Functions by Composition Zuowei Shen Department of Mathematics National University of Singapore Outline 1 A brief introduction of approximation theory 2 Deep learning: approximation

More information