Maximum a posteriori estimation for Markov chains based on Gaussian Markov random fields

Size: px
Start display at page:

Download "Maximum a posteriori estimation for Markov chains based on Gaussian Markov random fields"

Transcription

1 Proceda Computer Scence Proceda Computer Scence (1) 1 1 Maxmum a posteror estmaton for Markov chans based on Gaussan Markov random felds H. Wu, F. Noé Free Unversty of Berln, Arnmallee 6, Berln, Germany Abstract In ths paper, we present a Gaussan Markov random feld (GMRF) model for the transton matrces (TMs) of Markov chans (MCs) by assumng the exstence of a neghborhood relatonshp between states, and develop the maxmum a posteror (MAP) estmators under dfferent observaton condtons. Unlke earler work on TM estmaton, our method can make full use of the smlarty between dfferent states to mprove the estmated accuracy, and the estmator can be performed very effcently by solvng a convex programmng problem. In addton, we dscuss the parameter choce of the proposed model, and ntroduce a Monte Carlo cross valdaton (MCCV) method. The numercal smulatons of a dffuson process are employed to show the effectveness of the proposed models and algorthms. Keywords: Markov chan, Gaussan Markov random feld, maxmum a posteror, cross valdaton 1. Introducton Markov chan (MC) models provde a general modelng framework for descrbng state evolutons of stochastc and memoryless systems, and are now mportant and powerful tools for an enormous range of mathematcal applcatons, ncludng scence, economcs, and engneerng. Here we only focus on the fnte dscrete-tme homogeneous MC model, whch s one of the most common MC models, and whose dynamcs can be smply characterzed by a transton matrx (TM) T = [ T R n n wth T the transton probablty from the -th state to the -th state. In most applcatons, the man problem s to estmate the transton probabltes from observed data. In the past few decades, a lot of dfferent technques have been proposed to estmated the TMs. Many early researches devoted to the least-square (LS) approaches [1 3, for MC models Correspondng author Emal addresses: hwu@zedat.fu-berln.de (H. Wu), frank.noe@fu-berln.de (F. Noé)

2 H. Wu, F. Noé / Proceda Computer Scence (1) can be transformed to lnear stochastc systems wth zero-mean nose. However, the conventonal LS estmators may volate the nonnegatve constrants on TMs. Thus, some restrcted LS methods [4 6 based on constraned quadratc programmng algorthms were developed to avod ths problem. Some researchers [2, 5 suggested utlzng the weghted LS and weghted restrcted methods to solve the problem of heteroscedastcty. By now, the best known and most popular estmaton method of MC models s maxmum lkelhood (ML) estmator whch was proposed n [7, for t s consstent and asymptotcally normally dstrbuted as the sample sze ncreases [8, and can be effcently calculated by countng transton pars. Some experments show ML estmator s superor to the LS estmators [9. Moreover, the ML method can be appled to revsersble TM estmaton for some physcal and chemcal processes [1. Recently, the Bayesan approach [11, 12 to TM estmaton has receved a good deal of attenton. In ths approach, an unknown TM s assumed to be a realzaton of some pror model, and the posteror dstrbuton gven observed data can be obtaned by Bayes rule. Comparng to the non-bayesan methods, the Bayesan estmator can provde much more nformaton than a sngle pont estmate, and s more relable for small sze data set f the pror model s approprately desgned. The most commonly used pror dstrbuton s the matrx beta dstrbuton wth densty p (T Θ), T θ 1. It s a conugate pror and can be easly analyzed and effcently sampled snce each row of T follows the Drchlet dstrbuton. In some applcatons, Θ = 1 and Θ = are recommended, because p (T Θ) s equvalent to the unform dstrbuton when Θ = 1 [13, and Θ = makes the posteror mean of the TM dentcal to the ML estmate [14. The matrx Θ can also be optmzed by usng the emprcal Bayes approach [15. The matrx beta pror dstrbuton based Bayesan estmaton of revsersble TM was nvestgated n [13. The shortcomng of the matrx beta pror s that t does not take nto account possble correlatons between dfferent rows of the transton matrx. Assoudou and Essebbar [16, 17 proposed the Jeffreys pror (a non-nformatve pror) model for TMs to overcome ths problem, and no extra parameter s requred n ths model. However, the Jeffreys pror dstrbuton s too complcated for dervng the Bayesan estmator, and can only be appled to MC models wth very few states n practce. The maor obectve of ths paper s to propose a new pror model for MCs based on the Gaussan Markov random fled (GMRF). The GMRF [18 21 model s a specfc Gaussan feld model, and frequently used n spatal statstcs and mage processng, whch constructs a global dstrbuton of a spatal functon by consderng the local correlatons between ponts or regons. In ths paper, we assume that the state space of the MC has neghborhood structure and the adacent states have smlar transton behavors. Ths assumpton generally holds for the grd based approxmate models of contnuous space MCs, and the case that the state space has a dstance metrc. A GMRF pror model of TMs s then desgned accordng to the assumpton, and the correspondng maxmum a posteror (MAP) estmator s developed. In comparson wth the exstng models, the new pror model s able to utlze the smlarty relatonshp of states better. And there s only one extra parameter s requred, whch can be selected by the cross valdaton (CV) method. Moreover the estmaton problem wth nosy data s consdered, and the expectaton maxmzaton (EM) algorthm s used to get the MAP estmate. 2. Background 2.1. Gaussan Markov random felds Let G = (V, E) be an undrected graph wthout loop edges, where V s the set of vertces and E V V s the edge set. And vertces u, v V are sad to be adacent ff (u, v) E, whch s

3 H. Wu, F. Noé / Proceda Computer Scence (1) denoted by u v. It s clear that u, v V, v v and u v v u. A Gaussan Markov random feld (GMRF) Y on G s a Gaussan stochastc functon that assgns to each vertex v a real number Y (v). Here we only ntroduce the wdely used ntrnsc GMRF model [19, 22, whch s often specfed through the followng dstrbuton p GMRF (y σ) exp 1 ( ) 2 Y (u) Y (v) 2σ 2 d 2 (u, v) (1) where y = {Y (v) v V}, σ s a parameter, and d (, ) denotes a dstance measure between vertces. It s clear that the neghborng data ponts are desred to have the smlar values Markov chans We consder a tme-homogeneous Marv chan (MC) {x t t } on the fnte state space S = {s 1,..., s n }. Its probablty model can be descrbed by a transton matrx (TM) T = [ T R n n whose entres are gven by T = p ( ) x t+1 = s x t = s (2) where u v T = 1, T > (3) Here we defne Ω n = {T T R n n s a stochasc matrx}, whch s a convex set. And the probablty dstrbuton of the fnte-length state sequence {x, x 1,..., x m } gven T can be expressed as p (x :m T) = T C (4) where entres of count matrx C = [ C are numbers of observed transton pars wth, C = { (xt, x t+1 ) x t = s, x t+1 = s, t m 1 } (5) 3. GMRF Based MC Model Estmaton 3.1. GMRF pror Gven an MC state space S = {s 1,..., s n }, the purpose of ths subsecton s to provde a GMRF model based pror dstrbuton for the TM T = [ T. Assumng a neghborhood structure on the state space, we construct a neghborhood relaton between the transton pars as ( s, s ) (sk, s l ) ( s, s ) ( sk {s k }) ( s l {s l }) \ {(s k, s l )} (6) Then the unknown matrx T can be modeled by GMRF wth dstrbuton p GMRF (T σ) exp ( u (T, σ)) (7) where u (T, σ) = 1 2σ 2 (s,s ) (s k,s l ) T T kl d 2 kl 2 (8)

4 H. Wu, F. Noé / Proceda Computer Scence (1) d kl s the dstance between ( s, s ) and (sk, s l ), and here defned as d sk = d 2 (s, s k ) + d 2 ( s, s l ) (9) However, the realzaton of dstrbuton (7) does not satsfy (3) n the general case. Therefore we modfy the pror dstrbuton as 1 p GMC (T σ) = p GMRF (T σ, T Ω n ) z(σ) exp ( u (T, σ)), = T Ω n (1), T Ω n where z (σ) = exp ( u (T, σ)) dt Ω n (11) 3.2. MAP estmaton The maxmum a posteror (MAP) estmate of the TM T of an MC from observed data {x,..., x t } wth count matrx C = [ C s gven by { } ˆT = arg max log p (C T) + log p (T) T Usng the proposed GMRF pror model and assumng the parameter σ s known, (12) s equvalent to the followng optmzaton problem ˆT (σ) = arg mn T Ω n C log T + u (T, σ) (13), It s a convex problem and can be solved wthout any spurous local mnma. In ths paper, we perform the optmzaton by the dagonalzed Newton (DN) method (see [23 for detals) Choce of σ We now consder the case that σ s unknown. Motvated by the above analyss, t seems reasonable to ontly estmate T and σ by MAP method. But t s ntractable to compute the ont pror dstrbuton p(t, σ) = p(σ)p GMC (T σ) for z (σ) has no closed form. So here we use cross-valdaton (CV) approach to select the value of σ, and adopt the Monte Carlo cross-valdaton (MCCV) method proposed n [24. The MCCV of a σ s conducted by the followng steps: Step 1. Partton the set of observed state transton pars randomly nto tran and test subsets, where the tran subset s a fracton β (typcally.5) of the overall set, and the correspondng count matrces are denoted by C tran k and C test k. Step 2. Calculate and the predctve log-lkelhood (12) ˆT k (σ) = arg max T Ω n { log p ( C tran k T ) u (T, σ) } (14) CV k (σ) = log p ( C test k ˆT k (σ) ) (15)

5 H. Wu, F. Noé / Proceda Computer Scence (1) Step 3. Repeat the above steps for k = 1,..., K and select wth CV (σ) = k CV k (σ) /K. σ = arg max CV (σ) (16) σ It can be seen from (15) that CV k (σ) f the (, )-th entry of C test k s postve and that of ˆT k (σ) convergences to. In order to avod the possble sngularty, we approxmate the logarthmc functon as log ( ) ( ) 1 ( T PLη T = T η η 1) (17) when calculatng CV k (σ), where η (, 1) s a small number (η = n ths paper). It s easy to prove that lm η PL η (x) = log(x) for x >. 4. Estmaton wth Stochastc Observatons In ths secton, we wll take nto account that the actual state transtons are unknown, and only stochastc observatons o t x t p (o t x t ) (18) for t =,..., m are avalable. In ths case, the MAP estmator of the TM wth pror parameter σ can be expressed by { } ˆT (σ) = arg max log p (O T) u (T, σ) (19) T Ω n where O = {o,..., o m }, and computed wth the expectaton maxmzaton (EM) algorthm [25 consstng of the followng steps: Step 1. Choose an ntal T () Ω n and let k =. Step 2. Compute the functonal Q ( T T (k)) = E [ log (C (X) T) u (T, σ) T (k), O = C log T u (T, σ) (), where X = {x,..., x m }, C (X) = [ C (X) denotes the count matrx of X, and C = [ C = E [ C (X) T (k), O (21) Step 3. Fnd T (k+1) whch maxmzes the functon Q ( T T (k)) as T (k+1) = arg mn T Ω n C log T + u (T, σ), (22) Step 4. Termnate f ( ( ) log p O T (k+1) u ( T (k+1), σ )) ( log p ( O T (k)) u ( T (k), σ )) s small enough.

6 Step 5. Let k = k + 1 and go to Step 2. H. Wu, F. Noé / Proceda Computer Scence (1) Note that (22) has the same form as (13) wth C for any,, so (22) s a convex optmzaton problem and can be solved by the DN algorthm too. Further, n a smlar manner to Secton 3.3, the value of σ can be desgned through the MCCV algorthm. Due to space lmtatons, we omt detals here. 5. Smulatons 5.1. Brownan dynamcs model In ths secton, the estmaton method proposed n ths paper wll be appled to a Brownan dynamcs (BD) model, whch s descrbed as dr = f (r) dt + ρdw (23) where ρ = 1.4, W s a standard Brownan moton, f (r) = dv (r) /dr and V (r) s the potental functon (see Fg. 1) gven by 111.1r r r , r < r r r ,.75 r < 1 V (r) = r r 2 (24) 595.6r r < r r r , 1.25 < r Dscretzng the moton equaton (23) wth tme step t = 1 3 and decomposng the state space {r r 2} nto n = cells S = {s 1,..., s n } wth s = 2 1 n, we can get the grd based approxmate MC model p ( ( ) s x k+1 = s x k = s exp s + t f (s ) ) 2 2ρ 2 t (25) The correspondng TM T = [ T s shown n Fg. 2. Furthermore, the neghborhood structure on S s here defned by s = {s 1, s +1 } S wth dstance measure d ( s, s ) = TM estmaton Here, we wll use the MAP method presented n Secton 3 to estmate the TM T based on a realzaton {r (t) t 3} of (23) (see Fg. 3), and compare t wth the ML method [11. Fg. 4 plots the MCCV results of σ and the optmal σ = The comparsons of the dfferent estmators are based on the Kullback-Lebler (KL) dvergence rate metrc [26 defned as KLR ( ˆT T ) = ˆπ ˆT log ˆT T (26) where ˆπ = [ˆπ denotes the statonary dstrbuton of TM ˆT = [ ˆT. It can measure the dstances between Markov chans on the same state space.

7 H. Wu, F. Noé / Proceda Computer Scence (1) V r Fgure 1: Potental Functon Fgure 2: T r t Fgure 3: r (t) CV σ 1 1 Fgure 4: CV (σ) Fg. 5 shows the estmaton results of the proposed MAP method wth dfferent σ and ML method. Clearly, the ML method fals to estmate the values T wth [1, 7 [34, 44 [91, for there are few x k are sampled wthn the ranges. The GMRF pror based MAP estmator overcome ths problem by nterpolatng from the other T accordng to the GMRF model. Moreover, as observed from the fgures, the parameter σ determnes the overall smoothness of the estmated TM, and the MCCV approach can provde an approprate value of σ TM estmaton wth nosy data In ths subsecton, we study the performance of our proposed algorthms for estmatng T from nosy observatons o (t) r (t) N ( r (t), v 2) (27) wth v =. The MAP estmator wth GMRF pror n Secton 4 wll now be compared to the ML estmator mplemented usng Baum-Welch algorthm [27. The MCCV results are shown n Fg. 6 and the optmal σ = 947.

8 H. Wu, F. Noé / Proceda Computer Scence (1) (a) MAP (σ = σ ), KLR ( ˆT T ) =.872 (b) MAP (σ = 3σ ), KLR ( ˆT T ) = (c) MAP (σ = σ /3), KLR ( ˆT T ) =.729 (d) ML, KLR ( ˆT T ) =.2316 Fgure 5: ˆT 1 CV σ Fgure 6: CV (σ)

9 H. Wu, F. Noé / Proceda Computer Scence (1) (a) MAP (σ = σ ), KLR ( ˆT T ) = 7 (b) MAP (σ = 3σ ), KLR ( ˆT T ) = (c) MAP (σ =.6159), KLR ( ˆT T ) =.492 (d) ML, KLR ( ˆT T ) = Fgure 7: ˆT In Fg. 7, we show plots for ˆT obtaned usng our MAP estmator wth σ = σ, 3σ and.6159 (σ n Subsecton 5.2) and ML estmator. As can be seen from Fg. 7d, the ML estmator exhbts strong overfttng. Wth comparson to ML method, the proposed MAP estmator avods overfttng by the regularzaton term u (T, σ), whch penalzes excessvely large value of T. Note that here σ = 947 s bgger than the σ =.6159 n the prevous subsecton, whch may be related to nosy observaton and nsuffcent sample sze. From Fgs. 5a and 7c, we can see that the observaton nose makes ˆT obtaned from {o,..., o m } smoother than that drectly estmated by states {x,..., x m }. Therefore the MCCV approach wll select a bgger σ to get a sutably smooth ˆT and maxmze the predctve lkelhood. 6. Conclusons The GMRF model of TMs dscussed n ths paper provdes a general and flexble framework for analyzng and estmatng MCs wth smooth TMs by extendng the neghborhood relatonshp between states to that between transton pars. Ths model s helpful to mprove the robustness and accuracy of estmators n many practcal cases, especally when the sample sze

10 H. Wu, F. Noé / Proceda Computer Scence (1) s small wth respect to the sze of state space. And the convex form of GMRF model benefts the numercal calculaton. The parameter choce s a dffcult problem for our model, but t can be solved by CV methods snce there s only one undetermned parameter. References [1 G. Mller, Fnte Markov processes n psychology, Psychometrka 17 (2) (1952) [2 A. Madansky, Least squares estmaton n fnte Markov processes, Psychometrka 24 (2) (1959) [3 L. Telser, Least-squares estmates of transton probabltes, Measurement n Economcs (1963) [4 T. Lee, G. Judge, T. Takayama, On estmatng the transton probabltes of a Markov process, Journal of Farm Economcs 47 (3) (1965) [5 H. Thel, G. Rey, A quadratc programmng approach to the estmaton of transton probabltes, Management Scence 12 (9) (1966) [6 G. Judge, T. Takayama, Inequalty restrctons n regresson analyss, Journal of the Amercan Statstcal Assocaton 61 (313) (1966) [7 T. Anderson, L. Goodman, Statstcal nference about Markov chans, The Annals of Mathematcal Statstcs 28 (1) (1957) [8 M. Kendall, A. Stuart, The advanced theory of statstcs, Vol. 2, Charles Grffn, London, [9 T. Lee, G. Judge, A. Zellner, Estmatng the parameters of the Markov probablty model from aggregate tme seres data, North-Holland, 197. [1 G. Bowman, K. Beauchamp, G. Boxer, V. Pande, Progress and challenges n the automated constructon of Markov state models for full proten systems, The Journal of Chemcal Physcs 131 (9) [11 T. Lee, G. Judge, A. Zellner, Maxmum lkelhood and Bayesan estmaton of transton probabltes, Journal of the Amercan Statstcal Assocaton 63 (324) (1968) [12 J. Martn, Bayesan decson problems and Markov chans, Wley New York, [13 F. Noé, Probablty dstrbutons of molecular observables computed from Markov models, The Journal of Chemcal Physcs 128 (8) [14 C. Fuh, T. Fan, A Bayesan bootstrap for fnte state Markov chans, Statstca Snca 7 (1997) 5 1. [15 M. Meshkan, L. Bllard, Emprcal Bayes estmators for a fnte Markov chan, Bometrka 79 (1) (1992) [16 S. Assoudou, B. Essebbar, A Bayesan Model for Markov Chans va Jeffrey s Pror, Communcatons n Statstcs 32 (11). [17 S. Assoudou, B. Essebbar, A Bayesan model for bnary Markov chans, Internatonal Journal of Mathematcs and Mathematcal Scences 4 (8) (4) [18 J. Besag, Spatal nteracton and the statstcal analyss of lattce systems, Journal of the Royal Statstcal Socety. Seres B (Methodologcal) 36 (2) (1974) [19 J. Besag, C. Kooperberg, On condtonal and ntrnsc autoregressons, Bometrka 82 (4) (1995) [ H. Rue, L. Held, Gaussan Markov random felds: theory and applcatons, Chapman & Hall, 5. [21 S. L, Markov random feld modelng n mage analyss, Sprnger, 9. [22 S. Saqub, C. Bouman, K. Sauer, ML parameter estmaton for Markov random felds wth applcatons to Bayesan tomography, IEEE Transactons on Image Processng 7 (7) (1998) [23 T. Larsson, M. Patrksson, C. Rydergren, An effcent soluton method for the stochastc transportaton problem, n: Optmzaton Methods for Analyss of Transportaton Networks, Lnköpng Studes n Scence and Technology, no. 72, Department of Mathematcs, Lnköpng Unversty, [24 J. Shao, Lnear model selecton by cross-valdaton, Journal of the Amercan Statstcal Assocaton (1993) [25 A. Dempster, N. Lard, D. Rubn, et al., Maxmum lkelhood from ncomplete data va the EM algorthm, Journal of the Royal Statstcal Socety. Seres B (Methodologcal) 39 (1) (1977) [26 Z. Rached, F. Alaa, L. Campbell, The Kullback-Lebler dvergence rate between Markov sources, IEEE Transactons on Informaton Theory 5 (5) (4) [27 L. Welch, Hdden Markov models and the Baum-Welch algorthm, IEEE Informaton Theory Socety Newsletter 53 (4) (3) 1 1.

On an Extension of Stochastic Approximation EM Algorithm for Incomplete Data Problems. Vahid Tadayon 1

On an Extension of Stochastic Approximation EM Algorithm for Incomplete Data Problems. Vahid Tadayon 1 On an Extenson of Stochastc Approxmaton EM Algorthm for Incomplete Data Problems Vahd Tadayon Abstract: The Stochastc Approxmaton EM (SAEM algorthm, a varant stochastc approxmaton of EM, s a versatle tool

More information

Hidden Markov Models & The Multivariate Gaussian (10/26/04)

Hidden Markov Models & The Multivariate Gaussian (10/26/04) CS281A/Stat241A: Statstcal Learnng Theory Hdden Markov Models & The Multvarate Gaussan (10/26/04) Lecturer: Mchael I. Jordan Scrbes: Jonathan W. Hu 1 Hdden Markov Models As a bref revew, hdden Markov models

More information

COMPARISON OF SOME RELIABILITY CHARACTERISTICS BETWEEN REDUNDANT SYSTEMS REQUIRING SUPPORTING UNITS FOR THEIR OPERATIONS

COMPARISON OF SOME RELIABILITY CHARACTERISTICS BETWEEN REDUNDANT SYSTEMS REQUIRING SUPPORTING UNITS FOR THEIR OPERATIONS Avalable onlne at http://sck.org J. Math. Comput. Sc. 3 (3), No., 6-3 ISSN: 97-537 COMPARISON OF SOME RELIABILITY CHARACTERISTICS BETWEEN REDUNDANT SYSTEMS REQUIRING SUPPORTING UNITS FOR THEIR OPERATIONS

More information

2E Pattern Recognition Solutions to Introduction to Pattern Recognition, Chapter 2: Bayesian pattern classification

2E Pattern Recognition Solutions to Introduction to Pattern Recognition, Chapter 2: Bayesian pattern classification E395 - Pattern Recognton Solutons to Introducton to Pattern Recognton, Chapter : Bayesan pattern classfcaton Preface Ths document s a soluton manual for selected exercses from Introducton to Pattern Recognton

More information

Motion Perception Under Uncertainty. Hongjing Lu Department of Psychology University of Hong Kong

Motion Perception Under Uncertainty. Hongjing Lu Department of Psychology University of Hong Kong Moton Percepton Under Uncertanty Hongjng Lu Department of Psychology Unversty of Hong Kong Outlne Uncertanty n moton stmulus Correspondence problem Qualtatve fttng usng deal observer models Based on sgnal

More information

Numerical Heat and Mass Transfer

Numerical Heat and Mass Transfer Master degree n Mechancal Engneerng Numercal Heat and Mass Transfer 06-Fnte-Dfference Method (One-dmensonal, steady state heat conducton) Fausto Arpno f.arpno@uncas.t Introducton Why we use models and

More information

Composite Hypotheses testing

Composite Hypotheses testing Composte ypotheses testng In many hypothess testng problems there are many possble dstrbutons that can occur under each of the hypotheses. The output of the source s a set of parameters (ponts n a parameter

More information

Module 3 LOSSY IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur

Module 3 LOSSY IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur Module 3 LOSSY IMAGE COMPRESSION SYSTEMS Verson ECE IIT, Kharagpur Lesson 6 Theory of Quantzaton Verson ECE IIT, Kharagpur Instructonal Objectves At the end of ths lesson, the students should be able to:

More information

Classification as a Regression Problem

Classification as a Regression Problem Target varable y C C, C,, ; Classfcaton as a Regresson Problem { }, 3 L C K To treat classfcaton as a regresson problem we should transform the target y nto numercal values; The choce of numercal class

More information

Computation of Higher Order Moments from Two Multinomial Overdispersion Likelihood Models

Computation of Higher Order Moments from Two Multinomial Overdispersion Likelihood Models Computaton of Hgher Order Moments from Two Multnomal Overdsperson Lkelhood Models BY J. T. NEWCOMER, N. K. NEERCHAL Department of Mathematcs and Statstcs, Unversty of Maryland, Baltmore County, Baltmore,

More information

Non-Mixture Cure Model for Interval Censored Data: Simulation Study ABSTRACT

Non-Mixture Cure Model for Interval Censored Data: Simulation Study ABSTRACT Malaysan Journal of Mathematcal Scences 8(S): 37-44 (2014) Specal Issue: Internatonal Conference on Mathematcal Scences and Statstcs 2013 (ICMSS2013) MALAYSIAN JOURNAL OF MATHEMATICAL SCIENCES Journal

More information

Parametric fractional imputation for missing data analysis. Jae Kwang Kim Survey Working Group Seminar March 29, 2010

Parametric fractional imputation for missing data analysis. Jae Kwang Kim Survey Working Group Seminar March 29, 2010 Parametrc fractonal mputaton for mssng data analyss Jae Kwang Km Survey Workng Group Semnar March 29, 2010 1 Outlne Introducton Proposed method Fractonal mputaton Approxmaton Varance estmaton Multple mputaton

More information

Econ107 Applied Econometrics Topic 3: Classical Model (Studenmund, Chapter 4)

Econ107 Applied Econometrics Topic 3: Classical Model (Studenmund, Chapter 4) I. Classcal Assumptons Econ7 Appled Econometrcs Topc 3: Classcal Model (Studenmund, Chapter 4) We have defned OLS and studed some algebrac propertes of OLS. In ths topc we wll study statstcal propertes

More information

Introduction to Hidden Markov Models

Introduction to Hidden Markov Models Introducton to Hdden Markov Models Alperen Degrmenc Ths document contans dervatons and algorthms for mplementng Hdden Markov Models. The content presented here s a collecton of my notes and personal nsghts

More information

Retrieval Models: Language models

Retrieval Models: Language models CS-590I Informaton Retreval Retreval Models: Language models Luo S Department of Computer Scence Purdue Unversty Introducton to language model Ungram language model Document language model estmaton Maxmum

More information

Week 5: Neural Networks

Week 5: Neural Networks Week 5: Neural Networks Instructor: Sergey Levne Neural Networks Summary In the prevous lecture, we saw how we can construct neural networks by extendng logstc regresson. Neural networks consst of multple

More information

CS 2750 Machine Learning. Lecture 5. Density estimation. CS 2750 Machine Learning. Announcements

CS 2750 Machine Learning. Lecture 5. Density estimation. CS 2750 Machine Learning. Announcements CS 750 Machne Learnng Lecture 5 Densty estmaton Mlos Hauskrecht mlos@cs.ptt.edu 539 Sennott Square CS 750 Machne Learnng Announcements Homework Due on Wednesday before the class Reports: hand n before

More information

Simulated Power of the Discrete Cramér-von Mises Goodness-of-Fit Tests

Simulated Power of the Discrete Cramér-von Mises Goodness-of-Fit Tests Smulated of the Cramér-von Mses Goodness-of-Ft Tests Steele, M., Chaselng, J. and 3 Hurst, C. School of Mathematcal and Physcal Scences, James Cook Unversty, Australan School of Envronmental Studes, Grffth

More information

Lecture 12: Discrete Laplacian

Lecture 12: Discrete Laplacian Lecture 12: Dscrete Laplacan Scrbe: Tanye Lu Our goal s to come up wth a dscrete verson of Laplacan operator for trangulated surfaces, so that we can use t n practce to solve related problems We are mostly

More information

A PROBABILITY-DRIVEN SEARCH ALGORITHM FOR SOLVING MULTI-OBJECTIVE OPTIMIZATION PROBLEMS

A PROBABILITY-DRIVEN SEARCH ALGORITHM FOR SOLVING MULTI-OBJECTIVE OPTIMIZATION PROBLEMS HCMC Unversty of Pedagogy Thong Nguyen Huu et al. A PROBABILITY-DRIVEN SEARCH ALGORITHM FOR SOLVING MULTI-OBJECTIVE OPTIMIZATION PROBLEMS Thong Nguyen Huu and Hao Tran Van Department of mathematcs-nformaton,

More information

MATH 829: Introduction to Data Mining and Analysis The EM algorithm (part 2)

MATH 829: Introduction to Data Mining and Analysis The EM algorithm (part 2) 1/16 MATH 829: Introducton to Data Mnng and Analyss The EM algorthm (part 2) Domnque Gullot Departments of Mathematcal Scences Unversty of Delaware Aprl 20, 2016 Recall 2/16 We are gven ndependent observatons

More information

LOW BIAS INTEGRATED PATH ESTIMATORS. James M. Calvin

LOW BIAS INTEGRATED PATH ESTIMATORS. James M. Calvin Proceedngs of the 007 Wnter Smulaton Conference S G Henderson, B Bller, M-H Hseh, J Shortle, J D Tew, and R R Barton, eds LOW BIAS INTEGRATED PATH ESTIMATORS James M Calvn Department of Computer Scence

More information

Global Sensitivity. Tuesday 20 th February, 2018

Global Sensitivity. Tuesday 20 th February, 2018 Global Senstvty Tuesday 2 th February, 28 ) Local Senstvty Most senstvty analyses [] are based on local estmates of senstvty, typcally by expandng the response n a Taylor seres about some specfc values

More information

Linear Approximation with Regularization and Moving Least Squares

Linear Approximation with Regularization and Moving Least Squares Lnear Approxmaton wth Regularzaton and Movng Least Squares Igor Grešovn May 007 Revson 4.6 (Revson : March 004). 5 4 3 0.5 3 3.5 4 Contents: Lnear Fttng...4. Weghted Least Squares n Functon Approxmaton...

More information

Grenoble, France Grenoble University, F Grenoble Cedex, France

Grenoble, France   Grenoble University, F Grenoble Cedex, France MODIFIED K-MEA CLUSTERIG METHOD OF HMM STATES FOR IITIALIZATIO OF BAUM-WELCH TRAIIG ALGORITHM Paulne Larue 1, Perre Jallon 1, Bertrand Rvet 2 1 CEA LETI - MIATEC Campus Grenoble, France emal: perre.jallon@cea.fr

More information

Expectation Maximization Mixture Models HMMs

Expectation Maximization Mixture Models HMMs -755 Machne Learnng for Sgnal Processng Mture Models HMMs Class 9. 2 Sep 200 Learnng Dstrbutons for Data Problem: Gven a collecton of eamples from some data, estmate ts dstrbuton Basc deas of Mamum Lelhood

More information

Supporting Information

Supporting Information Supportng Informaton The neural network f n Eq. 1 s gven by: f x l = ReLU W atom x l + b atom, 2 where ReLU s the element-wse rectfed lnear unt, 21.e., ReLUx = max0, x, W atom R d d s the weght matrx to

More information

Checking Pairwise Relationships. Lecture 19 Biostatistics 666

Checking Pairwise Relationships. Lecture 19 Biostatistics 666 Checkng Parwse Relatonshps Lecture 19 Bostatstcs 666 Last Lecture: Markov Model for Multpont Analyss X X X 1 3 X M P X 1 I P X I P X 3 I P X M I 1 3 M I 1 I I 3 I M P I I P I 3 I P... 1 IBD states along

More information

Homework Assignment 3 Due in class, Thursday October 15

Homework Assignment 3 Due in class, Thursday October 15 Homework Assgnment 3 Due n class, Thursday October 15 SDS 383C Statstcal Modelng I 1 Rdge regresson and Lasso 1. Get the Prostrate cancer data from http://statweb.stanford.edu/~tbs/elemstatlearn/ datasets/prostate.data.

More information

Lecture 3: Shannon s Theorem

Lecture 3: Shannon s Theorem CSE 533: Error-Correctng Codes (Autumn 006 Lecture 3: Shannon s Theorem October 9, 006 Lecturer: Venkatesan Guruswam Scrbe: Wdad Machmouch 1 Communcaton Model The communcaton model we are usng conssts

More information

Convexity preserving interpolation by splines of arbitrary degree

Convexity preserving interpolation by splines of arbitrary degree Computer Scence Journal of Moldova, vol.18, no.1(52), 2010 Convexty preservng nterpolaton by splnes of arbtrary degree Igor Verlan Abstract In the present paper an algorthm of C 2 nterpolaton of dscrete

More information

Lecture Notes on Linear Regression

Lecture Notes on Linear Regression Lecture Notes on Lnear Regresson Feng L fl@sdueducn Shandong Unversty, Chna Lnear Regresson Problem In regresson problem, we am at predct a contnuous target value gven an nput feature vector We assume

More information

Using T.O.M to Estimate Parameter of distributions that have not Single Exponential Family

Using T.O.M to Estimate Parameter of distributions that have not Single Exponential Family IOSR Journal of Mathematcs IOSR-JM) ISSN: 2278-5728. Volume 3, Issue 3 Sep-Oct. 202), PP 44-48 www.osrjournals.org Usng T.O.M to Estmate Parameter of dstrbutons that have not Sngle Exponental Famly Jubran

More information

Regularized Discriminant Analysis for Face Recognition

Regularized Discriminant Analysis for Face Recognition 1 Regularzed Dscrmnant Analyss for Face Recognton Itz Pma, Mayer Aladem Department of Electrcal and Computer Engneerng, Ben-Guron Unversty of the Negev P.O.Box 653, Beer-Sheva, 845, Israel. Abstract Ths

More information

Markov Chain Monte Carlo (MCMC), Gibbs Sampling, Metropolis Algorithms, and Simulated Annealing Bioinformatics Course Supplement

Markov Chain Monte Carlo (MCMC), Gibbs Sampling, Metropolis Algorithms, and Simulated Annealing Bioinformatics Course Supplement Markov Chan Monte Carlo MCMC, Gbbs Samplng, Metropols Algorthms, and Smulated Annealng 2001 Bonformatcs Course Supplement SNU Bontellgence Lab http://bsnuackr/ Outlne! Markov Chan Monte Carlo MCMC! Metropols-Hastngs

More information

Maximizing Overlap of Large Primary Sampling Units in Repeated Sampling: A comparison of Ernst s Method with Ohlsson s Method

Maximizing Overlap of Large Primary Sampling Units in Repeated Sampling: A comparison of Ernst s Method with Ohlsson s Method Maxmzng Overlap of Large Prmary Samplng Unts n Repeated Samplng: A comparson of Ernst s Method wth Ohlsson s Method Red Rottach and Padrac Murphy 1 U.S. Census Bureau 4600 Slver Hll Road, Washngton DC

More information

Hidden Markov Models

Hidden Markov Models Hdden Markov Models Namrata Vaswan, Iowa State Unversty Aprl 24, 204 Hdden Markov Model Defntons and Examples Defntons:. A hdden Markov model (HMM) refers to a set of hdden states X 0, X,..., X t,...,

More information

CIS526: Machine Learning Lecture 3 (Sept 16, 2003) Linear Regression. Preparation help: Xiaoying Huang. x 1 θ 1 output... θ M x M

CIS526: Machine Learning Lecture 3 (Sept 16, 2003) Linear Regression. Preparation help: Xiaoying Huang. x 1 θ 1 output... θ M x M CIS56: achne Learnng Lecture 3 (Sept 6, 003) Preparaton help: Xaoyng Huang Lnear Regresson Lnear regresson can be represented by a functonal form: f(; θ) = θ 0 0 +θ + + θ = θ = 0 ote: 0 s a dummy attrbute

More information

Support Vector Machines. Vibhav Gogate The University of Texas at dallas

Support Vector Machines. Vibhav Gogate The University of Texas at dallas Support Vector Machnes Vbhav Gogate he Unversty of exas at dallas What We have Learned So Far? 1. Decson rees. Naïve Bayes 3. Lnear Regresson 4. Logstc Regresson 5. Perceptron 6. Neural networks 7. K-Nearest

More information

VQ widely used in coding speech, image, and video

VQ widely used in coding speech, image, and video at Scalar quantzers are specal cases of vector quantzers (VQ): they are constraned to look at one sample at a tme (memoryless) VQ does not have such constrant better RD perfomance expected Source codng

More information

Kernel Methods and SVMs Extension

Kernel Methods and SVMs Extension Kernel Methods and SVMs Extenson The purpose of ths document s to revew materal covered n Machne Learnng 1 Supervsed Learnng regardng support vector machnes (SVMs). Ths document also provdes a general

More information

A Robust Method for Calculating the Correlation Coefficient

A Robust Method for Calculating the Correlation Coefficient A Robust Method for Calculatng the Correlaton Coeffcent E.B. Nven and C. V. Deutsch Relatonshps between prmary and secondary data are frequently quantfed usng the correlaton coeffcent; however, the tradtonal

More information

U.C. Berkeley CS294: Beyond Worst-Case Analysis Luca Trevisan September 5, 2017

U.C. Berkeley CS294: Beyond Worst-Case Analysis Luca Trevisan September 5, 2017 U.C. Berkeley CS94: Beyond Worst-Case Analyss Handout 4s Luca Trevsan September 5, 07 Summary of Lecture 4 In whch we ntroduce semdefnte programmng and apply t to Max Cut. Semdefnte Programmng Recall that

More information

Power law and dimension of the maximum value for belief distribution with the max Deng entropy

Power law and dimension of the maximum value for belief distribution with the max Deng entropy Power law and dmenson of the maxmum value for belef dstrbuton wth the max Deng entropy Bngy Kang a, a College of Informaton Engneerng, Northwest A&F Unversty, Yanglng, Shaanx, 712100, Chna. Abstract Deng

More information

ISSN: ISO 9001:2008 Certified International Journal of Engineering and Innovative Technology (IJEIT) Volume 3, Issue 1, July 2013

ISSN: ISO 9001:2008 Certified International Journal of Engineering and Innovative Technology (IJEIT) Volume 3, Issue 1, July 2013 ISSN: 2277-375 Constructon of Trend Free Run Orders for Orthogonal rrays Usng Codes bstract: Sometmes when the expermental runs are carred out n a tme order sequence, the response can depend on the run

More information

Mean Field / Variational Approximations

Mean Field / Variational Approximations Mean Feld / Varatonal Appromatons resented by Jose Nuñez 0/24/05 Outlne Introducton Mean Feld Appromaton Structured Mean Feld Weghted Mean Feld Varatonal Methods Introducton roblem: We have dstrbuton but

More information

Lossy Compression. Compromise accuracy of reconstruction for increased compression.

Lossy Compression. Compromise accuracy of reconstruction for increased compression. Lossy Compresson Compromse accuracy of reconstructon for ncreased compresson. The reconstructon s usually vsbly ndstngushable from the orgnal mage. Typcally, one can get up to 0:1 compresson wth almost

More information

Lecture 12: Classification

Lecture 12: Classification Lecture : Classfcaton g Dscrmnant functons g The optmal Bayes classfer g Quadratc classfers g Eucldean and Mahalanobs metrcs g K Nearest Neghbor Classfers Intellgent Sensor Systems Rcardo Guterrez-Osuna

More information

ANOMALIES OF THE MAGNITUDE OF THE BIAS OF THE MAXIMUM LIKELIHOOD ESTIMATOR OF THE REGRESSION SLOPE

ANOMALIES OF THE MAGNITUDE OF THE BIAS OF THE MAXIMUM LIKELIHOOD ESTIMATOR OF THE REGRESSION SLOPE P a g e ANOMALIES OF THE MAGNITUDE OF THE BIAS OF THE MAXIMUM LIKELIHOOD ESTIMATOR OF THE REGRESSION SLOPE Darmud O Drscoll ¹, Donald E. Ramrez ² ¹ Head of Department of Mathematcs and Computer Studes

More information

An adaptive SMC scheme for ABC. Bayesian Computation (ABC)

An adaptive SMC scheme for ABC. Bayesian Computation (ABC) An adaptve SMC scheme for Approxmate Bayesan Computaton (ABC) (ont work wth Prof. Mke West) Department of Statstcal Scence - Duke Unversty Aprl/2011 Approxmate Bayesan Computaton (ABC) Problems n whch

More information

A Bayesian Approach to Arrival Rate Forecasting for Inhomogeneous Poisson Processes for Mobile Calls

A Bayesian Approach to Arrival Rate Forecasting for Inhomogeneous Poisson Processes for Mobile Calls A Bayesan Approach to Arrval Rate Forecastng for Inhomogeneous Posson Processes for Moble Calls Mchael N. Nawar Department of Computer Engneerng Caro Unversty Caro, Egypt mchaelnawar@eee.org Amr F. Atya

More information

DERIVATION OF THE PROBABILITY PLOT CORRELATION COEFFICIENT TEST STATISTICS FOR THE GENERALIZED LOGISTIC DISTRIBUTION

DERIVATION OF THE PROBABILITY PLOT CORRELATION COEFFICIENT TEST STATISTICS FOR THE GENERALIZED LOGISTIC DISTRIBUTION Internatonal Worshop ADVANCES IN STATISTICAL HYDROLOGY May 3-5, Taormna, Italy DERIVATION OF THE PROBABILITY PLOT CORRELATION COEFFICIENT TEST STATISTICS FOR THE GENERALIZED LOGISTIC DISTRIBUTION by Sooyoung

More information

Gaussian Mixture Models

Gaussian Mixture Models Lab Gaussan Mxture Models Lab Objectve: Understand the formulaton of Gaussan Mxture Models (GMMs) and how to estmate GMM parameters. You ve already seen GMMs as the observaton dstrbuton n certan contnuous

More information

Finite Mixture Models and Expectation Maximization. Most slides are from: Dr. Mario Figueiredo, Dr. Anil Jain and Dr. Rong Jin

Finite Mixture Models and Expectation Maximization. Most slides are from: Dr. Mario Figueiredo, Dr. Anil Jain and Dr. Rong Jin Fnte Mxture Models and Expectaton Maxmzaton Most sldes are from: Dr. Maro Fgueredo, Dr. Anl Jan and Dr. Rong Jn Recall: The Supervsed Learnng Problem Gven a set of n samples X {(x, y )},,,n Chapter 3 of

More information

Chapter - 2. Distribution System Power Flow Analysis

Chapter - 2. Distribution System Power Flow Analysis Chapter - 2 Dstrbuton System Power Flow Analyss CHAPTER - 2 Radal Dstrbuton System Load Flow 2.1 Introducton Load flow s an mportant tool [66] for analyzng electrcal power system network performance. Load

More information

Hidden Markov Models

Hidden Markov Models CM229S: Machne Learnng for Bonformatcs Lecture 12-05/05/2016 Hdden Markov Models Lecturer: Srram Sankararaman Scrbe: Akshay Dattatray Shnde Edted by: TBD 1 Introducton For a drected graph G we can wrte

More information

Stat260: Bayesian Modeling and Inference Lecture Date: February 22, Reference Priors

Stat260: Bayesian Modeling and Inference Lecture Date: February 22, Reference Priors Stat60: Bayesan Modelng and Inference Lecture Date: February, 00 Reference Prors Lecturer: Mchael I. Jordan Scrbe: Steven Troxler and Wayne Lee In ths lecture, we assume that θ R; n hgher-dmensons, reference

More information

Learning undirected Models. Instructor: Su-In Lee University of Washington, Seattle. Mean Field Approximation

Learning undirected Models. Instructor: Su-In Lee University of Washington, Seattle. Mean Field Approximation Readngs: K&F 0.3, 0.4, 0.6, 0.7 Learnng undrected Models Lecture 8 June, 0 CSE 55, Statstcal Methods, Sprng 0 Instructor: Su-In Lee Unversty of Washngton, Seattle Mean Feld Approxmaton Is the energy functonal

More information

Parametric fractional imputation for missing data analysis

Parametric fractional imputation for missing data analysis Secton on Survey Research Methods JSM 2008 Parametrc fractonal mputaton for mssng data analyss Jae Kwang Km Wayne Fuller Abstract Under a parametrc model for mssng data, the EM algorthm s a popular tool

More information

Appendix B: Resampling Algorithms

Appendix B: Resampling Algorithms 407 Appendx B: Resamplng Algorthms A common problem of all partcle flters s the degeneracy of weghts, whch conssts of the unbounded ncrease of the varance of the mportance weghts ω [ ] of the partcles

More information

Research Article Green s Theorem for Sign Data

Research Article Green s Theorem for Sign Data Internatonal Scholarly Research Network ISRN Appled Mathematcs Volume 2012, Artcle ID 539359, 10 pages do:10.5402/2012/539359 Research Artcle Green s Theorem for Sgn Data Lous M. Houston The Unversty of

More information

DETERMINATION OF UNCERTAINTY ASSOCIATED WITH QUANTIZATION ERRORS USING THE BAYESIAN APPROACH

DETERMINATION OF UNCERTAINTY ASSOCIATED WITH QUANTIZATION ERRORS USING THE BAYESIAN APPROACH Proceedngs, XVII IMEKO World Congress, June 7, 3, Dubrovn, Croata Proceedngs, XVII IMEKO World Congress, June 7, 3, Dubrovn, Croata TC XVII IMEKO World Congress Metrology n the 3rd Mllennum June 7, 3,

More information

Portfolios with Trading Constraints and Payout Restrictions

Portfolios with Trading Constraints and Payout Restrictions Portfolos wth Tradng Constrants and Payout Restrctons John R. Brge Northwestern Unversty (ont wor wth Chrs Donohue Xaodong Xu and Gongyun Zhao) 1 General Problem (Very) long-term nvestor (eample: unversty

More information

BOOTSTRAP METHOD FOR TESTING OF EQUALITY OF SEVERAL MEANS. M. Krishna Reddy, B. Naveen Kumar and Y. Ramu

BOOTSTRAP METHOD FOR TESTING OF EQUALITY OF SEVERAL MEANS. M. Krishna Reddy, B. Naveen Kumar and Y. Ramu BOOTSTRAP METHOD FOR TESTING OF EQUALITY OF SEVERAL MEANS M. Krshna Reddy, B. Naveen Kumar and Y. Ramu Department of Statstcs, Osmana Unversty, Hyderabad -500 007, Inda. nanbyrozu@gmal.com, ramu0@gmal.com

More information

A Hybrid Variational Iteration Method for Blasius Equation

A Hybrid Variational Iteration Method for Blasius Equation Avalable at http://pvamu.edu/aam Appl. Appl. Math. ISSN: 1932-9466 Vol. 10, Issue 1 (June 2015), pp. 223-229 Applcatons and Appled Mathematcs: An Internatonal Journal (AAM) A Hybrd Varatonal Iteraton Method

More information

Chapter Newton s Method

Chapter Newton s Method Chapter 9. Newton s Method After readng ths chapter, you should be able to:. Understand how Newton s method s dfferent from the Golden Secton Search method. Understand how Newton s method works 3. Solve

More information

Bayesian predictive Configural Frequency Analysis

Bayesian predictive Configural Frequency Analysis Psychologcal Test and Assessment Modelng, Volume 54, 2012 (3), 285-292 Bayesan predctve Confgural Frequency Analyss Eduardo Gutérrez-Peña 1 Abstract Confgural Frequency Analyss s a method for cell-wse

More information

Lecture 10 Support Vector Machines II

Lecture 10 Support Vector Machines II Lecture 10 Support Vector Machnes II 22 February 2016 Taylor B. Arnold Yale Statstcs STAT 365/665 1/28 Notes: Problem 3 s posted and due ths upcomng Frday There was an early bug n the fake-test data; fxed

More information

DETERMINATION OF TEMPERATURE DISTRIBUTION FOR ANNULAR FINS WITH TEMPERATURE DEPENDENT THERMAL CONDUCTIVITY BY HPM

DETERMINATION OF TEMPERATURE DISTRIBUTION FOR ANNULAR FINS WITH TEMPERATURE DEPENDENT THERMAL CONDUCTIVITY BY HPM Ganj, Z. Z., et al.: Determnaton of Temperature Dstrbuton for S111 DETERMINATION OF TEMPERATURE DISTRIBUTION FOR ANNULAR FINS WITH TEMPERATURE DEPENDENT THERMAL CONDUCTIVITY BY HPM by Davood Domr GANJI

More information

Efficient nonresponse weighting adjustment using estimated response probability

Efficient nonresponse weighting adjustment using estimated response probability Effcent nonresponse weghtng adjustment usng estmated response probablty Jae Kwang Km Department of Appled Statstcs, Yonse Unversty, Seoul, 120-749, KOREA Key Words: Regresson estmator, Propensty score,

More information

CSci 6974 and ECSE 6966 Math. Tech. for Vision, Graphics and Robotics Lecture 21, April 17, 2006 Estimating A Plane Homography

CSci 6974 and ECSE 6966 Math. Tech. for Vision, Graphics and Robotics Lecture 21, April 17, 2006 Estimating A Plane Homography CSc 6974 and ECSE 6966 Math. Tech. for Vson, Graphcs and Robotcs Lecture 21, Aprl 17, 2006 Estmatng A Plane Homography Overvew We contnue wth a dscusson of the major ssues, usng estmaton of plane projectve

More information

The Second Anti-Mathima on Game Theory

The Second Anti-Mathima on Game Theory The Second Ant-Mathma on Game Theory Ath. Kehagas December 1 2006 1 Introducton In ths note we wll examne the noton of game equlbrum for three types of games 1. 2-player 2-acton zero-sum games 2. 2-player

More information

Appendix B. The Finite Difference Scheme

Appendix B. The Finite Difference Scheme 140 APPENDIXES Appendx B. The Fnte Dfference Scheme In ths appendx we present numercal technques whch are used to approxmate solutons of system 3.1 3.3. A comprehensve treatment of theoretcal and mplementaton

More information

Probability Theory (revisited)

Probability Theory (revisited) Probablty Theory (revsted) Summary Probablty v.s. plausblty Random varables Smulaton of Random Experments Challenge The alarm of a shop rang. Soon afterwards, a man was seen runnng n the street, persecuted

More information

Generalized Linear Methods

Generalized Linear Methods Generalzed Lnear Methods 1 Introducton In the Ensemble Methods the general dea s that usng a combnaton of several weak learner one could make a better learner. More formally, assume that we have a set

More information

Durban Watson for Testing the Lack-of-Fit of Polynomial Regression Models without Replications

Durban Watson for Testing the Lack-of-Fit of Polynomial Regression Models without Replications Durban Watson for Testng the Lack-of-Ft of Polynomal Regresson Models wthout Replcatons Ruba A. Alyaf, Maha A. Omar, Abdullah A. Al-Shha ralyaf@ksu.edu.sa, maomar@ksu.edu.sa, aalshha@ksu.edu.sa Department

More information

Statistical analysis using matlab. HY 439 Presented by: George Fortetsanakis

Statistical analysis using matlab. HY 439 Presented by: George Fortetsanakis Statstcal analyss usng matlab HY 439 Presented by: George Fortetsanaks Roadmap Probablty dstrbutons Statstcal estmaton Fttng data to probablty dstrbutons Contnuous dstrbutons Contnuous random varable X

More information

Uncertainty and auto-correlation in. Measurement

Uncertainty and auto-correlation in. Measurement Uncertanty and auto-correlaton n arxv:1707.03276v2 [physcs.data-an] 30 Dec 2017 Measurement Markus Schebl Federal Offce of Metrology and Surveyng (BEV), 1160 Venna, Austra E-mal: markus.schebl@bev.gv.at

More information

Comparison of the Population Variance Estimators. of 2-Parameter Exponential Distribution Based on. Multiple Criteria Decision Making Method

Comparison of the Population Variance Estimators. of 2-Parameter Exponential Distribution Based on. Multiple Criteria Decision Making Method Appled Mathematcal Scences, Vol. 7, 0, no. 47, 07-0 HIARI Ltd, www.m-hkar.com Comparson of the Populaton Varance Estmators of -Parameter Exponental Dstrbuton Based on Multple Crtera Decson Makng Method

More information

Convergence of random processes

Convergence of random processes DS-GA 12 Lecture notes 6 Fall 216 Convergence of random processes 1 Introducton In these notes we study convergence of dscrete random processes. Ths allows to characterze phenomena such as the law of large

More information

A Comparative Study for Estimation Parameters in Panel Data Model

A Comparative Study for Estimation Parameters in Panel Data Model A Comparatve Study for Estmaton Parameters n Panel Data Model Ahmed H. Youssef and Mohamed R. Abonazel hs paper examnes the panel data models when the regresson coeffcents are fxed random and mxed and

More information

Quantifying Uncertainty

Quantifying Uncertainty Partcle Flters Quantfyng Uncertanty Sa Ravela M. I. T Last Updated: Sprng 2013 1 Quantfyng Uncertanty Partcle Flters Partcle Flters Appled to Sequental flterng problems Can also be appled to smoothng problems

More information

Markov Chain Monte Carlo Lecture 6

Markov Chain Monte Carlo Lecture 6 where (x 1,..., x N ) X N, N s called the populaton sze, f(x) f (x) for at least one {1, 2,..., N}, and those dfferent from f(x) are called the tral dstrbutons n terms of mportance samplng. Dfferent ways

More information

The Gaussian classifier. Nuno Vasconcelos ECE Department, UCSD

The Gaussian classifier. Nuno Vasconcelos ECE Department, UCSD he Gaussan classfer Nuno Vasconcelos ECE Department, UCSD Bayesan decson theory recall that we have state of the world X observatons g decson functon L[g,y] loss of predctng y wth g Bayes decson rule s

More information

Unified Subspace Analysis for Face Recognition

Unified Subspace Analysis for Face Recognition Unfed Subspace Analyss for Face Recognton Xaogang Wang and Xaoou Tang Department of Informaton Engneerng The Chnese Unversty of Hong Kong Shatn, Hong Kong {xgwang, xtang}@e.cuhk.edu.hk Abstract PCA, LDA

More information

CHAPTER-5 INFORMATION MEASURE OF FUZZY MATRIX AND FUZZY BINARY RELATION

CHAPTER-5 INFORMATION MEASURE OF FUZZY MATRIX AND FUZZY BINARY RELATION CAPTER- INFORMATION MEASURE OF FUZZY MATRI AN FUZZY BINARY RELATION Introducton The basc concept of the fuzz matr theor s ver smple and can be appled to socal and natural stuatons A branch of fuzz matr

More information

Credit Card Pricing and Impact of Adverse Selection

Credit Card Pricing and Impact of Adverse Selection Credt Card Prcng and Impact of Adverse Selecton Bo Huang and Lyn C. Thomas Unversty of Southampton Contents Background Aucton model of credt card solctaton - Errors n probablty of beng Good - Errors n

More information

Predictive Analytics : QM901.1x Prof U Dinesh Kumar, IIMB. All Rights Reserved, Indian Institute of Management Bangalore

Predictive Analytics : QM901.1x Prof U Dinesh Kumar, IIMB. All Rights Reserved, Indian Institute of Management Bangalore Sesson Outlne Introducton to classfcaton problems and dscrete choce models. Introducton to Logstcs Regresson. Logstc functon and Logt functon. Maxmum Lkelhood Estmator (MLE) for estmaton of LR parameters.

More information

See Book Chapter 11 2 nd Edition (Chapter 10 1 st Edition)

See Book Chapter 11 2 nd Edition (Chapter 10 1 st Edition) Count Data Models See Book Chapter 11 2 nd Edton (Chapter 10 1 st Edton) Count data consst of non-negatve nteger values Examples: number of drver route changes per week, the number of trp departure changes

More information

Tracking with Kalman Filter

Tracking with Kalman Filter Trackng wth Kalman Flter Scott T. Acton Vrgna Image and Vdeo Analyss (VIVA), Charles L. Brown Department of Electrcal and Computer Engneerng Department of Bomedcal Engneerng Unversty of Vrgna, Charlottesvlle,

More information

Entropy of Markov Information Sources and Capacity of Discrete Input Constrained Channels (from Immink, Coding Techniques for Digital Recorders)

Entropy of Markov Information Sources and Capacity of Discrete Input Constrained Channels (from Immink, Coding Techniques for Digital Recorders) Entropy of Marov Informaton Sources and Capacty of Dscrete Input Constraned Channels (from Immn, Codng Technques for Dgtal Recorders). Entropy of Marov Chans We have already ntroduced the noton of entropy

More information

Lecture 7: Boltzmann distribution & Thermodynamics of mixing

Lecture 7: Boltzmann distribution & Thermodynamics of mixing Prof. Tbbtt Lecture 7 etworks & Gels Lecture 7: Boltzmann dstrbuton & Thermodynamcs of mxng 1 Suggested readng Prof. Mark W. Tbbtt ETH Zürch 13 März 018 Molecular Drvng Forces Dll and Bromberg: Chapters

More information

Degradation Data Analysis Using Wiener Process and MCMC Approach

Degradation Data Analysis Using Wiener Process and MCMC Approach Engneerng Letters 5:3 EL_5_3_0 Degradaton Data Analyss Usng Wener Process and MCMC Approach Chunpng L Hubng Hao Abstract Tradtonal relablty assessment methods are based on lfetme data. However the lfetme

More information

and V is a p p positive definite matrix. A normal-inverse-gamma distribution.

and V is a p p positive definite matrix. A normal-inverse-gamma distribution. OSR Journal of athematcs (OSR-J) e-ssn: 78-578, p-ssn: 39-765X. Volume 3, ssue 3 Ver. V (ay - June 07), PP 68-7 www.osrjournals.org Comparng The Performance of Bayesan And Frequentst Analyss ethods of

More information

Logistic Regression. CAP 5610: Machine Learning Instructor: Guo-Jun QI

Logistic Regression. CAP 5610: Machine Learning Instructor: Guo-Jun QI Logstc Regresson CAP 561: achne Learnng Instructor: Guo-Jun QI Bayes Classfer: A Generatve model odel the posteror dstrbuton P(Y X) Estmate class-condtonal dstrbuton P(X Y) for each Y Estmate pror dstrbuton

More information

STAT 511 FINAL EXAM NAME Spring 2001

STAT 511 FINAL EXAM NAME Spring 2001 STAT 5 FINAL EXAM NAME Sprng Instructons: Ths s a closed book exam. No notes or books are allowed. ou may use a calculator but you are not allowed to store notes or formulas n the calculator. Please wrte

More information

NUMERICAL DIFFERENTIATION

NUMERICAL DIFFERENTIATION NUMERICAL DIFFERENTIATION 1 Introducton Dfferentaton s a method to compute the rate at whch a dependent output y changes wth respect to the change n the ndependent nput x. Ths rate of change s called the

More information

Information Geometry of Gibbs Sampler

Information Geometry of Gibbs Sampler Informaton Geometry of Gbbs Sampler Kazuya Takabatake Neuroscence Research Insttute AIST Central 2, Umezono 1-1-1, Tsukuba JAPAN 305-8568 k.takabatake@ast.go.jp Abstract: - Ths paper shows some nformaton

More information

Maximum Likelihood Estimation of Binary Dependent Variables Models: Probit and Logit. 1. General Formulation of Binary Dependent Variables Models

Maximum Likelihood Estimation of Binary Dependent Variables Models: Probit and Logit. 1. General Formulation of Binary Dependent Variables Models ECO 452 -- OE 4: Probt and Logt Models ECO 452 -- OE 4 Maxmum Lkelhood Estmaton of Bnary Dependent Varables Models: Probt and Logt hs note demonstrates how to formulate bnary dependent varables models

More information

Bayesian Planning of Hit-Miss Inspection Tests

Bayesian Planning of Hit-Miss Inspection Tests Bayesan Plannng of Ht-Mss Inspecton Tests Yew-Meng Koh a and Wllam Q Meeker a a Center for Nondestructve Evaluaton, Department of Statstcs, Iowa State Unversty, Ames, Iowa 5000 Abstract Although some useful

More information