Second order approximations for probability models
|
|
- Suzan Ramsey
- 5 years ago
- Views:
Transcription
1 Second order approxmatons for probablty models lbert Kappen Department of Bophyscs Njmegen Unversty Njmegen, the Netherlands bertmbfysunnl Wm Wegernc Department of Bophyscs Njmegen Unversty Njmegen, the Netherlands wmwmbfysunnl Abstract In ths paper, we derve a second order mean feld theory for drected graphcal probablty models By usng an nformaton theoretc argument t s shown how ths can be done n the absense of a partton functon Ths method s a drect generalsaton of the well-nown TA approxmaton for Boltzmann Machnes In a numercal example, t s shown that the method greatly mproves the frst order mean feld approxmaton For a restrcted class of graphcal models, so-called sngle overlap graphs, the second order method has comparable complexty to the frst order method For sgmod belef networs, the method s shown to be partcularly fast and effectve 1 Introducton Recently, a number of authors have proposed determnstc methods for approxmate nference n large graphcal models The smplest approach gves a lower bound on the probablty of a subset of varables usng enssen s nequalty (Saul et al, 1996) The method nvolves the mnmzaton of the K dvergence between the target probablty dstrbuton and some smple varatonal dstrbuton The method can be appled to a large class of probablty models, such as sgmod belef networs, DAGs and Boltzmann Machnes (BM) For Boltzmann-Gbbs dstrbutons, t s possble to derve the lower bound as the frst term n a Taylor seres expanson of the free energy around a factorzed model The free energy s gven by, where s the normalzaton constant of the Boltzmann-Gbbs dstrbuton: $ "!#! Ths Taylor seres can be contnued and the second order term s nown as the TA correcton (lefa, 1982; Kappen and Rodríguez, 1998) The second order term sgnfcantly mproves the qualty of the approxmaton, but s no longer a bound For probablty dstrbutons that are not Boltzmann-Gbbs dstrbutons, t s not obvous how to obtan the second order approxmaton owever, there s an alternatve way to compute the hgher order correctons, based on an nformaton theoretc argument Recently, ths argument was appled to stochastc neural networs wth asymmetrc connectvty (Kappen and Spanjers, 1999) ere, we apply ths dea to drected graphcal models
2 C C : 2 The method et be an -dmensonal vector, wth tang on dscrete values et # be a drected graphcal model on We wll assume that can be wrtten as a product of potentals n the followng way: (1) ere, # denotes the condtonal probablty table of varable gven the values of ts parents denotes the subset of varables that appear n potental and # otentals can be overlappng, "! $#, and $% We wsh to compute the margnal probablty that & has some specfc value ' n the presence of some evdence We therefore denote )( ' where ( denote the subset of varables that consttute the evdence, and ' denotes the remander of the varables The margnal s gven as ' ( )( (2) Both numerator and denomnator contan sums over hdden states These sums scale exponentally wth the sze of the problem, and therefore the computaton of margnals s ntractable We propose to approxmate ths problem by usng a mean feld approach Consder a factorzed dstrbuton on the hdden varables * : ' 3) We wsh to fnd the factorzed dstrbuton that best approxmates ' ( Consder as a dstance measure +-, It s easy to see that the that mnmzes ' ( +3, 0/ ' ( ' 21 satsfes: () ' (5) We now thn of the manfold of all probablty dstrbutons of the form Eq 1, spanned by the coordnates # 5 76 For each, s a table of numbers, ndexed by Ths manfold contans a submanfold of factorzed probablty dstrbutons n whch the potentals factorze: 98 ;: )< 8 : ;< # # 7=?> *, ' ( reduces to ' When n addton, Assume now that ' ( s somehow close to the factorzed submanfold The dfference ' s then small, and we can expand ths small dfference n terms of changes n the parameters A B6 : ;C /GF I1K ED F M / F D N O F F Q 1R Q hgher order terms (6)
3 F The dfferentals are evaluated n the factorzed dstrbuton The left-hand sze of Eq 6 s zero because of Eq 5 and we solve for ' Ths factorzed dstrbuton gves the desred margnals up to the order of the expanson of It s straghtforward to compute the dervatves: F F F F Q ' ( ( Q ' ( Q ( ' ( Q ' ( ( Q ( (7) We ntroduce the notaton and as the expectaton values wth respect to the factorzed dstrbutons ' ( and (, respectvely We defne We obtan M : To frst order, settng Eq 8 equal to zero we obtan hgher order terms (8) # # ' (9) where we have absorbed all terms ndependent of = nto a constant Thus, we fnd the soluton ' (10) n whch the constants follow from normalsaton The frst order term s equvalent to the standard mean feld equatons, obtaned from ensens nequalty The correcton wth second order terms s obtaned n the same way, agan droppng terms ndependent of = : ' # M : (11) were, agan, the constants follow from normalsaton These equatons, whch form the man result of ths paper, are generalzaton of the mean feld equatons wth TA correctons for drected graphcal models Both left and rght-hand sze of Eqs 10 and 11 depend on the unnown probablty dstrbuton ' and can be solved by fxed pont teraton 3 Complexty and sngle-overlap graphs The complexty of the frst order equatons Eq 10 s exponental n the number of varables n the potentals of : f the maxmal clque sze s, then for each = we need of the order of computatons, where s the number of clques that contan node = The second term scales worse, snce one must compute averages over the unon of two overlappng clques and because of the double sum owever, thngs are not so bad Frst
4 vst to Asa? Smong? vst to Asa? Smong? Tuberculoss? ung cancer? Bronchts? Tuberculoss? ung cancer? Bronchts? Ether tub or lung canc? Ether tub or lung canc? postve X-ray? Dyspnoea? postve X-ray? Dyspnoea? Fgure 1: An example of a sngle-overlap graph eft: The chest clnc model (ASIA)(aurtzen and Spegelhalter, 1988) Rght: nodes wthn one potental a re grouped together, showng that potentals share at most one node of all, notce that the sum over and can be restrcted to overlappng clques (! # ) and that = must be n ether or or both (= > % ) Denote by the number of clques that have at least one varable n common wth clque and denote by Then, the sum over and contans not more than terms M Each term s an average over the unon of two clques, whch can be worse case of sze (when only one varable s shared) owever, snce ( means expectaton wrt condtoned on the varables n ) we can precompute for all pars of overlappng clques, for all states n Therefore, the worse case complexty of the second order term s less than Thus, we see that the second order method has the same exponental complexty as the frst order method, but wth a dfferent polynomal prefactor Therefore, the frst or second order method can be appled to drected graphcal models as long as the number of parents s reasonably small The fact that the second order term has a worse complexty than the frst order term s n contrast to Boltzmann machnes, n whch the TA approxmaton has the same complexty as the standard mean feld approxmaton Ths phenomenon also occurs for a specal class of DAGs, whch we call sngle-overlap graphs These are graphs n whch the potentals share at most one node Fgure 1 shows an example of a sngle-overlap graph For sngle overlap graphs, we can use the frst order result Eq 9 to smplfy the second order correcton The dervaton rather tedous and we just present the result ' M : ;< : ;< (12) whch has a complexty that s of order For probablty dstrbutons wth many small potentals that share nodes wth many other potentals, Eq 12 s more effcent than Eq 11 For nstance, for Boltzmann Machnes and M In ths case, Eq 12 s dentcal to the TA equatons (Thouless et al, 1977) Sgmod belef networs In ths secton, we consder sgmod belef networs as an nterestng class of drected graphcal models The reason s, that one can expand n terms of the couplngs nstead of the potentals whch s more effcent The sgmod belef networ s defned as # * (13)
5 8 ( 6 l l (a) (b) (c) (d) Fgure 2: Interpretaton of dfferent nteracton terms appearng n Eq 16 The open and shaded nodes are hdden and evdence nodes, respectvely (except n (a), where can be any node) Sold arrows ndcate the graphcal structure n the networ Dashed arrows ndcate nteracton terms that appear n Eq 16 where # M, and * s the local feld: * We separate the varables n evdence varables ( and hdden varables ' : ' ( When couplngs from hdden nodes to ether hdden or evdence nodes are zero,, =?> ( ' and > ', the probablty dstrbutons ' ( and )( reduce to ' ( ' )< ' (1) )( ( ( (15) ;< where $8 < depends on the evdence We expand to second order around ths tractable dstrbuton and obtan 6 M ( ( 6 6 < : < < 6 )( ( ( ( 6 < < : < M B 6 < : < wth 6 ' ' and s gven by Eq 15 ( ( (16) The dfferent terms that appear n ths equaton can be easly nterpreted The frst term descrbes the lowest order forward nfluence on node = from ts parents arents can be ether evdence or hdden nodes (fg 2a) The second term s the bas The thrd term descrbes to lowest order the effect of Bayes rule: t affects 6 such that the observed evdence on ts chldren becomes most probable (fg 2b) Note, that ths term s absent when the evdence s explaned by the evdence nodes themselves: ( The fourth and ffth terms are the quadratc contrbutons to the frst and thrd terms, respectvely The sxth term descrbes explanng away It descrbes the effect of hdden node on node =, when both have a common observed chld (fg 2c) The last term descrbes the effect on node = when ts grandchld s observed (fg 2d) Note, that these equatons are dfferent from Eq 10 When one apples Eq 10 to sgmod belef networs, one requres addtonal approxmatons to compute * (Saul et al, 1996)
6 M A B cpu tme (sec) RMS C D n Fgure 3: Second order approxmaton for fully connected sgmod belef networ of nodes a) nodes are hdden (whte) and nodes are clamped (grey), ; b) CU tme for exact nference (dashed) and second order approxmaton (sold) versus ( ); c) RMS of hdden node exact margnals (sold) and RMS error of second order approxmaton (dashed) versus couplng strength, ( A ) Snce only feed-forward connectons are present, one can order the nodes such that for = Then the frst order mean feld equatons can be solved n one sngle sweep startng wth node 1 The full second order equatons can be solved by teraton, startng wth the frst order soluton 5 Numercal results We llustrate the theory wth two toy problems The frst one s nference n aurtzen s chest clnc model (ASIA), defned on 8 bnary varables 0 (see fgure 1, and (aurtzen and Spegelhalter, 1988) for more detals about the model) We computed exact margnals and approxmate margnals usng the approxmatng methods up to frst (Eq 10) and second order (Eq 11), respectvely The approxmate margnals are determned by sequental teraton of (10) and (11), startng at # for all varables = The maxmal error n the margnals usng the frst and second order method s 0213 and 0061, respectvely We verfed that the sngle-overlap expresson Eq 12 gave smlar results In fg 3, we assess the accuracy and CU tme of the second order approxmaton Eq 16 for sgmod belef networs We generate random fully connected sgmod belef networs wth from a normal dstrbuton wth mean zero and varance and We observe n fg 3b that the computaton tme s very fast: For, we have obtaned convergence n 37 second on a entum 300 Mhz processor The accuracy of the method depends on the sze of the weghts and s computed for a networ of (fg 3c) In (Kappen and Wegernc, 2001), we compare ths approach to Saul s varatonal approach (Saul et al, 1996) and show that our approach s much faster and slghtly more accurate
7 6 Dscusson In ths paper, we computed a second order mean feld approxmaton for drected graphcal models We show that the second order approxmaton gves a sgnfcant mprovement over the frst order result The method does not use explctly that the graph s drected Therefore, the result s equally vald for Marov graphs The complexty of the frst and second order approxmaton s of and, respectvely, wth the number of varables n the largest potental For sngle-overlap graphs, one can rewrte the second order equaton such that the computatonal complexty reduces to Boltzmann machnes and the Asa networ are examples of sngle-overlap graphs For large, addtonal approxmatons are requred, as was proposed by (Saul et al, 1996) for the frst order mean feld equatons It s evdent, that such addtonal approxmatons are then also requred for the second order mean feld equatons It has been reported (Barber and Wegernc, 1999; Wegernc and Kappen, 1999) that smlar numercal mprovements can be obtaned by usng a very dfferent approach, whch s to use an approxmatng dstrbuton that s not factorzed, but stll tractable A promsng way to proceed s therefore to combne both approaches and to do a second order expanson aroud a manfold of dstrbutons wth non-factorzed yet tractable dstrbutons In ths approach the suffcent statstcs of the tractable structure s expanded, rather than the margnal probabltes Acnowledgments Ths research was supported n part by the Dutch Technology Foundaton (STW) References Barber, D and Wegernc, W (1999) Tractable varatonal structures for approxmatng graphcal models In Kearns, M, Solla, S, and Cohn, D, edtors, Advances n Neural Informaton rocessng Systems, volume 11 of Advances n Neural Informaton rocessng Systems, pages MIT ress Kappen, and Rodríguez, F (1998) Effcent learnng n Boltzmann Machnes usng lnear response theory Neural Computaton, 10: Kappen, and Spanjers, (1999) Mean feld theory for asymmetrc neural networs hyscal Revew E, 61: Kappen, and Wegernc, W (2001) Mean feld theory for graphcal models In Saad, D and Opper, M, edtors, Advanced mean feld theory MIT ress aurtzen, S and Spegelhalter, D (1988) ocal computatons wth probabltes on graphcal structures and ther applcaton to expert systems Royal Statstcal socety B, 50: lefa, T (1982) Convergence condton of the TA equaton for the nfnte-range Isng spn glass model ournal of hyscs A, 15: Saul,, aaola, T, and ordan, M (1996) Mean feld theory for sgmod belef networs ournal of artfcal ntellgence research, :61 76 Thouless, D, Anderson,, and almer, R (1977) Soluton of Solvable Model of a Spn Glass hl Mag, 35: Wegernc, W and Kappen, (1999) Approxmatons of bayesan networs through l mnmsaton New Generaton Computng, 18:
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 informationEEE 241: Linear Systems
EEE : Lnear Systems Summary #: Backpropagaton BACKPROPAGATION The perceptron rule as well as the Wdrow Hoff learnng were desgned to tran sngle layer networks. They suffer from the same dsadvantage: they
More informationFor now, let us focus on a specific model of neurons. These are simplified from reality but can achieve remarkable results.
Neural Networks : Dervaton compled by Alvn Wan from Professor Jtendra Malk s lecture Ths type of computaton s called deep learnng and s the most popular method for many problems, such as computer vson
More informationThe optimal delay of the second test is therefore approximately 210 hours earlier than =2.
THE IEC 61508 FORMULAS 223 The optmal delay of the second test s therefore approxmately 210 hours earler than =2. 8.4 The IEC 61508 Formulas IEC 61508-6 provdes approxmaton formulas for the PF for smple
More informationOutline. Bayesian Networks: Maximum Likelihood Estimation and Tree Structure Learning. Our Model and Data. Outline
Outlne Bayesan Networks: Maxmum Lkelhood Estmaton and Tree Structure Learnng Huzhen Yu janey.yu@cs.helsnk.f Dept. Computer Scence, Unv. of Helsnk Probablstc Models, Sprng, 200 Notces: I corrected a number
More informationOne-sided finite-difference approximations suitable for use with Richardson extrapolation
Journal of Computatonal Physcs 219 (2006) 13 20 Short note One-sded fnte-dfference approxmatons sutable for use wth Rchardson extrapolaton Kumar Rahul, S.N. Bhattacharyya * Department of Mechancal Engneerng,
More informationLinear 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 informationFinding Dense Subgraphs in G(n, 1/2)
Fndng Dense Subgraphs n Gn, 1/ Atsh Das Sarma 1, Amt Deshpande, and Rav Kannan 1 Georga Insttute of Technology,atsh@cc.gatech.edu Mcrosoft Research-Bangalore,amtdesh,annan@mcrosoft.com Abstract. Fndng
More informationCHAPTER 14 GENERAL PERTURBATION THEORY
CHAPTER 4 GENERAL PERTURBATION THEORY 4 Introducton A partcle n orbt around a pont mass or a sphercally symmetrc mass dstrbuton s movng n a gravtatonal potental of the form GM / r In ths potental t moves
More informationGeneralized 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 informationWhy BP Works STAT 232B
Why BP Works STAT 232B Free Energes Helmholz & Gbbs Free Energes 1 Dstance between Probablstc Models - K-L dvergence b{ KL b{ p{ = b{ ln { } p{ Here, p{ s the eact ont prob. b{ s the appromaton, called
More informationKernel 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 informationSupporting 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 informationLinear Regression Analysis: Terminology and Notation
ECON 35* -- Secton : Basc Concepts of Regresson Analyss (Page ) Lnear Regresson Analyss: Termnology and Notaton Consder the generc verson of the smple (two-varable) lnear regresson model. It s represented
More informationGlobal 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 informationLecture 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 informationCollege of Computer & Information Science Fall 2009 Northeastern University 20 October 2009
College of Computer & Informaton Scence Fall 2009 Northeastern Unversty 20 October 2009 CS7880: Algorthmc Power Tools Scrbe: Jan Wen and Laura Poplawsk Lecture Outlne: Prmal-dual schema Network Desgn:
More information1 Convex Optimization
Convex Optmzaton We wll consder convex optmzaton problems. Namely, mnmzaton problems where the objectve s convex (we assume no constrants for now). Such problems often arse n machne learnng. For example,
More informationExpectation propagation
Expectaton propagaton Lloyd Ellott May 17, 2011 Suppose p(x) s a pdf and we have a factorzaton p(x) = 1 Z n f (x). (1) =1 Expectaton propagaton s an nference algorthm desgned to approxmate the factors
More informationCOMPARISON 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 informationMath 426: Probability MWF 1pm, Gasson 310 Homework 4 Selected Solutions
Exercses from Ross, 3, : Math 26: Probablty MWF pm, Gasson 30 Homework Selected Solutons 3, p. 05 Problems 76, 86 3, p. 06 Theoretcal exercses 3, 6, p. 63 Problems 5, 0, 20, p. 69 Theoretcal exercses 2,
More informationANSWERS. Problem 1. and the moment generating function (mgf) by. defined for any real t. Use this to show that E( U) var( U)
Econ 413 Exam 13 H ANSWERS Settet er nndelt 9 deloppgaver, A,B,C, som alle anbefales å telle lkt for å gøre det ltt lettere å stå. Svar er gtt . Unfortunately, there s a prntng error n the hnt of
More informationProblem Set 9 Solutions
Desgn and Analyss of Algorthms May 4, 2015 Massachusetts Insttute of Technology 6.046J/18.410J Profs. Erk Demane, Srn Devadas, and Nancy Lynch Problem Set 9 Solutons Problem Set 9 Solutons Ths problem
More informationA 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 informationMASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 12 10/21/2013. Martingale Concentration Inequalities and Applications
MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.65/15.070J Fall 013 Lecture 1 10/1/013 Martngale Concentraton Inequaltes and Applcatons Content. 1. Exponental concentraton for martngales wth bounded ncrements.
More informationHidden 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 informationDepartment of Quantitative Methods & Information Systems. Time Series and Their Components QMIS 320. Chapter 6
Department of Quanttatve Methods & Informaton Systems Tme Seres and Ther Components QMIS 30 Chapter 6 Fall 00 Dr. Mohammad Zanal These sldes were modfed from ther orgnal source for educatonal purpose only.
More informationThe Multiple Classical Linear Regression Model (CLRM): Specification and Assumptions. 1. Introduction
ECONOMICS 5* -- NOTE (Summary) ECON 5* -- NOTE The Multple Classcal Lnear Regresson Model (CLRM): Specfcaton and Assumptons. Introducton CLRM stands for the Classcal Lnear Regresson Model. The CLRM s also
More informationErrors for Linear Systems
Errors for Lnear Systems When we solve a lnear system Ax b we often do not know A and b exactly, but have only approxmatons  and ˆb avalable. Then the best thng we can do s to solve ˆx ˆb exactly whch
More informationLecture 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 informationChapter 5. Solution of System of Linear Equations. Module No. 6. Solution of Inconsistent and Ill Conditioned Systems
Numercal Analyss by Dr. Anta Pal Assstant Professor Department of Mathematcs Natonal Insttute of Technology Durgapur Durgapur-713209 emal: anta.bue@gmal.com 1 . Chapter 5 Soluton of System of Lnear Equatons
More informationConjugacy and the Exponential Family
CS281B/Stat241B: Advanced Topcs n Learnng & Decson Makng Conjugacy and the Exponental Famly Lecturer: Mchael I. Jordan Scrbes: Bran Mlch 1 Conjugacy In the prevous lecture, we saw conjugate prors for the
More information3.1 Expectation of Functions of Several Random Variables. )' be a k-dimensional discrete or continuous random vector, with joint PMF p (, E X E X1 E X
Statstcs 1: Probablty Theory II 37 3 EPECTATION OF SEVERAL RANDOM VARIABLES As n Probablty Theory I, the nterest n most stuatons les not on the actual dstrbuton of a random vector, but rather on a number
More informationReport on Image warping
Report on Image warpng Xuan Ne, Dec. 20, 2004 Ths document summarzed the algorthms of our mage warpng soluton for further study, and there s a detaled descrpton about the mplementaton of these algorthms.
More informationEcon107 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 informationWorkshop: Approximating energies and wave functions Quantum aspects of physical chemistry
Workshop: Approxmatng energes and wave functons Quantum aspects of physcal chemstry http://quantum.bu.edu/pltl/6/6.pdf Last updated Thursday, November 7, 25 7:9:5-5: Copyrght 25 Dan Dll (dan@bu.edu) Department
More informationHidden 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 informationNumerical 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 informationStat260: 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 informationFormulas for the Determinant
page 224 224 CHAPTER 3 Determnants e t te t e 2t 38 A = e t 2te t e 2t e t te t 2e 2t 39 If 123 A = 345, 456 compute the matrx product A adj(a) What can you conclude about det(a)? For Problems 40 43, use
More informationInformation 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 informationNumerical Solution of Ordinary Differential Equations
Numercal Methods (CENG 00) CHAPTER-VI Numercal Soluton of Ordnar Dfferental Equatons 6 Introducton Dfferental equatons are equatons composed of an unknown functon and ts dervatves The followng are examples
More informationFoundations of Arithmetic
Foundatons of Arthmetc Notaton We shall denote the sum and product of numbers n the usual notaton as a 2 + a 2 + a 3 + + a = a, a 1 a 2 a 3 a = a The notaton a b means a dvdes b,.e. ac = b where c s an
More information18.1 Introduction and Recap
CS787: Advanced Algorthms Scrbe: Pryananda Shenoy and Shjn Kong Lecturer: Shuch Chawla Topc: Streamng Algorthmscontnued) Date: 0/26/2007 We contnue talng about streamng algorthms n ths lecture, ncludng
More informationChapter 8 Indicator Variables
Chapter 8 Indcator Varables In general, e explanatory varables n any regresson analyss are assumed to be quanttatve n nature. For example, e varables lke temperature, dstance, age etc. are quanttatve n
More informationCHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE
CHAPTER 5 NUMERICAL EVALUATION OF DYNAMIC RESPONSE Analytcal soluton s usually not possble when exctaton vares arbtrarly wth tme or f the system s nonlnear. Such problems can be solved by numercal tmesteppng
More informationEPR Paradox and the Physical Meaning of an Experiment in Quantum Mechanics. Vesselin C. Noninski
EPR Paradox and the Physcal Meanng of an Experment n Quantum Mechancs Vesseln C Nonnsk vesselnnonnsk@verzonnet Abstract It s shown that there s one purely determnstc outcome when measurement s made on
More informationOpen Systems: Chemical Potential and Partial Molar Quantities Chemical Potential
Open Systems: Chemcal Potental and Partal Molar Quanttes Chemcal Potental For closed systems, we have derved the followng relatonshps: du = TdS pdv dh = TdS + Vdp da = SdT pdv dg = VdP SdT For open systems,
More informationJoint Statistical Meetings - Biopharmaceutical Section
Iteratve Ch-Square Test for Equvalence of Multple Treatment Groups Te-Hua Ng*, U.S. Food and Drug Admnstraton 1401 Rockvlle Pke, #200S, HFM-217, Rockvlle, MD 20852-1448 Key Words: Equvalence Testng; Actve
More informationBasically, if you have a dummy dependent variable you will be estimating a probability.
ECON 497: Lecture Notes 13 Page 1 of 1 Metropoltan State Unversty ECON 497: Research and Forecastng Lecture Notes 13 Dummy Dependent Varable Technques Studenmund Chapter 13 Bascally, f you have a dummy
More informationLecture 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 informationFormal solvers of the RT equation
Formal solvers of the RT equaton Formal RT solvers Runge- Kutta (reference solver) Pskunov N.: 979, Master Thess Long characterstcs (Feautrer scheme) Cannon C.J.: 970, ApJ 6, 55 Short characterstcs (Hermtan
More informationA New Evolutionary Computation Based Approach for Learning Bayesian Network
Avalable onlne at www.scencedrect.com Proceda Engneerng 15 (2011) 4026 4030 Advanced n Control Engneerng and Informaton Scence A New Evolutonary Computaton Based Approach for Learnng Bayesan Network Yungang
More informationEnsemble Methods: Boosting
Ensemble Methods: Boostng Ncholas Ruozz Unversty of Texas at Dallas Based on the sldes of Vbhav Gogate and Rob Schapre Last Tme Varance reducton va baggng Generate new tranng data sets by samplng wth replacement
More information8 : Learning in Fully Observed Markov Networks. 1 Why We Need to Learn Undirected Graphical Models. 2 Structural Learning for Completely Observed MRF
10-708: Probablstc Graphcal Models 10-708, Sprng 2014 8 : Learnng n Fully Observed Markov Networks Lecturer: Erc P. Xng Scrbes: Meng Song, L Zhou 1 Why We Need to Learn Undrected Graphcal Models In the
More information2E 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 informationn α j x j = 0 j=1 has a nontrivial solution. Here A is the n k matrix whose jth column is the vector for all t j=0
MODULE 2 Topcs: Lnear ndependence, bass and dmenson We have seen that f n a set of vectors one vector s a lnear combnaton of the remanng vectors n the set then the span of the set s unchanged f that vector
More informationChapter 13: Multiple Regression
Chapter 13: Multple Regresson 13.1 Developng the multple-regresson Model The general model can be descrbed as: It smplfes for two ndependent varables: The sample ft parameter b 0, b 1, and b are used to
More informationNotes on Frequency Estimation in Data Streams
Notes on Frequency Estmaton n Data Streams In (one of) the data streamng model(s), the data s a sequence of arrvals a 1, a 2,..., a m of the form a j = (, v) where s the dentty of the tem and belongs to
More informationLectures - Week 4 Matrix norms, Conditioning, Vector Spaces, Linear Independence, Spanning sets and Basis, Null space and Range of a Matrix
Lectures - Week 4 Matrx norms, Condtonng, Vector Spaces, Lnear Independence, Spannng sets and Bass, Null space and Range of a Matrx Matrx Norms Now we turn to assocatng a number to each matrx. We could
More informationCase A. P k = Ni ( 2L i k 1 ) + (# big cells) 10d 2 P k.
THE CELLULAR METHOD In ths lecture, we ntroduce the cellular method as an approach to ncdence geometry theorems lke the Szemeréd-Trotter theorem. The method was ntroduced n the paper Combnatoral complexty
More informationChapter 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 informationLecture 13 APPROXIMATION OF SECOMD ORDER DERIVATIVES
COMPUTATIONAL FLUID DYNAMICS: FDM: Appromaton of Second Order Dervatves Lecture APPROXIMATION OF SECOMD ORDER DERIVATIVES. APPROXIMATION OF SECOND ORDER DERIVATIVES Second order dervatves appear n dffusve
More informationTransfer Functions. Convenient representation of a linear, dynamic model. A transfer function (TF) relates one input and one output: ( ) system
Transfer Functons Convenent representaton of a lnear, dynamc model. A transfer functon (TF) relates one nput and one output: x t X s y t system Y s The followng termnology s used: x y nput output forcng
More informationUsing 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 informationExcess Error, Approximation Error, and Estimation Error
E0 370 Statstcal Learnng Theory Lecture 10 Sep 15, 011 Excess Error, Approxaton Error, and Estaton Error Lecturer: Shvan Agarwal Scrbe: Shvan Agarwal 1 Introducton So far, we have consdered the fnte saple
More informationLimited Dependent Variables
Lmted Dependent Varables. What f the left-hand sde varable s not a contnuous thng spread from mnus nfnty to plus nfnty? That s, gven a model = f (, β, ε, where a. s bounded below at zero, such as wages
More informationArtificial Intelligence Bayesian Networks
Artfcal Intellgence Bayesan Networks Adapted from sldes by Tm Fnn and Mare desjardns. Some materal borrowed from Lse Getoor. 1 Outlne Bayesan networks Network structure Condtonal probablty tables Condtonal
More informationMore metrics on cartesian products
More metrcs on cartesan products If (X, d ) are metrc spaces for 1 n, then n Secton II4 of the lecture notes we defned three metrcs on X whose underlyng topologes are the product topology The purpose of
More information1 Motivation and Introduction
Instructor: Dr. Volkan Cevher EXPECTATION PROPAGATION September 30, 2008 Rce Unversty STAT 63 / ELEC 633: Graphcal Models Scrbes: Ahmad Beram Andrew Waters Matthew Nokleby Index terms: Approxmate nference,
More informationMean 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 informationAdditional Codes using Finite Difference Method. 1 HJB Equation for Consumption-Saving Problem Without Uncertainty
Addtonal Codes usng Fnte Dfference Method Benamn Moll 1 HJB Equaton for Consumpton-Savng Problem Wthout Uncertanty Before consderng the case wth stochastc ncome n http://www.prnceton.edu/~moll/ HACTproect/HACT_Numercal_Appendx.pdf,
More informationChapter 5 Multilevel Models
Chapter 5 Multlevel Models 5.1 Cross-sectonal multlevel models 5.1.1 Two-level models 5.1.2 Multple level models 5.1.3 Multple level modelng n other felds 5.2 Longtudnal multlevel models 5.2.1 Two-level
More informationThis column is a continuation of our previous column
Comparson of Goodness of Ft Statstcs for Lnear Regresson, Part II The authors contnue ther dscusson of the correlaton coeffcent n developng a calbraton for quanttatve analyss. Jerome Workman Jr. and Howard
More informationSection 8.3 Polar Form of Complex Numbers
80 Chapter 8 Secton 8 Polar Form of Complex Numbers From prevous classes, you may have encountered magnary numbers the square roots of negatve numbers and, more generally, complex numbers whch are the
More informationOn the correction of the h-index for career length
1 On the correcton of the h-ndex for career length by L. Egghe Unverstet Hasselt (UHasselt), Campus Depenbeek, Agoralaan, B-3590 Depenbeek, Belgum 1 and Unverstet Antwerpen (UA), IBW, Stadscampus, Venusstraat
More informationProf. Dr. I. Nasser Phys 630, T Aug-15 One_dimensional_Ising_Model
EXACT OE-DIMESIOAL ISIG MODEL The one-dmensonal Isng model conssts of a chan of spns, each spn nteractng only wth ts two nearest neghbors. The smple Isng problem n one dmenson can be solved drectly n several
More informationComputing MLE Bias Empirically
Computng MLE Bas Emprcally Kar Wa Lm Australan atonal Unversty January 3, 27 Abstract Ths note studes the bas arses from the MLE estmate of the rate parameter and the mean parameter of an exponental dstrbuton.
More informationThe equation of motion of a dynamical system is given by a set of differential equations. That is (1)
Dynamcal Systems Many engneerng and natural systems are dynamcal systems. For example a pendulum s a dynamcal system. State l The state of the dynamcal system specfes t condtons. For a pendulum n the absence
More informationInductance Calculation for Conductors of Arbitrary Shape
CRYO/02/028 Aprl 5, 2002 Inductance Calculaton for Conductors of Arbtrary Shape L. Bottura Dstrbuton: Internal Summary In ths note we descrbe a method for the numercal calculaton of nductances among conductors
More information= z 20 z n. (k 20) + 4 z k = 4
Problem Set #7 solutons 7.2.. (a Fnd the coeffcent of z k n (z + z 5 + z 6 + z 7 + 5, k 20. We use the known seres expanson ( n+l ( z l l z n below: (z + z 5 + z 6 + z 7 + 5 (z 5 ( + z + z 2 + z + 5 5
More informationSolving Nonlinear Differential Equations by a Neural Network Method
Solvng Nonlnear Dfferental Equatons by a Neural Network Method Luce P. Aarts and Peter Van der Veer Delft Unversty of Technology, Faculty of Cvlengneerng and Geoscences, Secton of Cvlengneerng Informatcs,
More informationRobert Eisberg Second edition CH 09 Multielectron atoms ground states and x-ray excitations
Quantum Physcs 量 理 Robert Esberg Second edton CH 09 Multelectron atoms ground states and x-ray exctatons 9-01 By gong through the procedure ndcated n the text, develop the tme-ndependent Schroednger equaton
More informationA new construction of 3-separable matrices via an improved decoding of Macula s construction
Dscrete Optmzaton 5 008 700 704 Contents lsts avalable at ScenceDrect Dscrete Optmzaton journal homepage: wwwelsevercom/locate/dsopt A new constructon of 3-separable matrces va an mproved decodng of Macula
More informationU.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 informationMarkov 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 informationUncertainty 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 informationAn Interactive Optimisation Tool for Allocation Problems
An Interactve Optmsaton ool for Allocaton Problems Fredr Bonäs, Joam Westerlund and apo Westerlund Process Desgn Laboratory, Faculty of echnology, Åbo Aadem Unversty, uru 20500, Fnland hs paper presents
More informationWeek3, Chapter 4. Position and Displacement. Motion in Two Dimensions. Instantaneous Velocity. Average Velocity
Week3, Chapter 4 Moton n Two Dmensons Lecture Quz A partcle confned to moton along the x axs moves wth constant acceleraton from x =.0 m to x = 8.0 m durng a 1-s tme nterval. The velocty of the partcle
More informationAffine transformations and convexity
Affne transformatons and convexty The purpose of ths document s to prove some basc propertes of affne transformatons nvolvng convex sets. Here are a few onlne references for background nformaton: http://math.ucr.edu/
More information1 Derivation of Rate Equations from Single-Cell Conductance (Hodgkin-Huxley-like) Equations
Physcs 171/271 -Davd Klenfeld - Fall 2005 (revsed Wnter 2011) 1 Dervaton of Rate Equatons from Sngle-Cell Conductance (Hodgkn-Huxley-lke) Equatons We consder a network of many neurons, each of whch obeys
More informationNP-Completeness : Proofs
NP-Completeness : Proofs Proof Methods A method to show a decson problem Π NP-complete s as follows. (1) Show Π NP. (2) Choose an NP-complete problem Π. (3) Show Π Π. A method to show an optmzaton problem
More informationChapter 11: Simple Linear Regression and Correlation
Chapter 11: Smple Lnear Regresson and Correlaton 11-1 Emprcal Models 11-2 Smple Lnear Regresson 11-3 Propertes of the Least Squares Estmators 11-4 Hypothess Test n Smple Lnear Regresson 11-4.1 Use of t-tests
More informationMAXIMUM A POSTERIORI TRANSDUCTION
MAXIMUM A POSTERIORI TRANSDUCTION LI-WEI WANG, JU-FU FENG School of Mathematcal Scences, Peng Unversty, Bejng, 0087, Chna Center for Informaton Scences, Peng Unversty, Bejng, 0087, Chna E-MIAL: {wanglw,
More informationOn 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 informationDifference Equations
Dfference Equatons c Jan Vrbk 1 Bascs Suppose a sequence of numbers, say a 0,a 1,a,a 3,... s defned by a certan general relatonshp between, say, three consecutve values of the sequence, e.g. a + +3a +1
More informationHomework 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 informationTHE SUMMATION NOTATION Ʃ
Sngle Subscrpt otaton THE SUMMATIO OTATIO Ʃ Most of the calculatons we perform n statstcs are repettve operatons on lsts of numbers. For example, we compute the sum of a set of numbers, or the sum of the
More informationCS 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 informationMATH 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