CART: Classification and Regression Trees

Size: px
Start display at page:

Download "CART: Classification and Regression Trees"

Transcription

1 CART: Claifiation and Regreion Tree Günther Sawitzki Otober, 04 Literature [] L. Breiman, J. H. Friedman, R. A. Olhen, and C. J. Stone. Claifiation and Regreion Tree. Wadworth, Belmont, CA., 984. [] J. H. Friedman. Loal learning baed on reurie oering. Tehnial report, Department of Statiti, Stanford Unierity, 996. [3] J. H. Friedman. On bia, ariane, 0/ lo, and the ure-of-dimenionality. Data Min. Knowl. Dio., ():55 77, 997. [4] T. Hatie, R. Tibhirani, and J. H. Friedman. The Element of Statitial Learning. Springer Serie in Statiti. Springer, nd edition, 009. [5] B.D. Ripley. Pattern reognition and neural network. Cambridge Unierity Pre, 996. Günther Sawitzki: CART Otober, 04 Günther Sawitzki: CART Otober, Coariable X with alue in X, for example X = R p. Repone Y with alue in Y, for example Y = R. and find model funtion m from loal model funtion m l : Elementary ae: m l ontant on X l. Example (regreion): m l = Y (x) x Xl Example (laifiation): m l = majority{y (x) x Xl } Günther Sawitzki: CART Otober, 04 3 Günther Sawitzki: CART Otober, Coariable X with alue in X, for example X = R p. Repone Y with alue in Y, for example Y = R. and find model funtion m from loal model funtion m l : Elementary ae: m l ontant on X l. Coariable X with alue in X, for example X = R p. Repone Y with alue in Y, for example Y = R. and find model funtion m from loal model funtion m l : Elementary ae: m l ontant on X l. Example (regreion): m l = Y (x) x Xl Example (laifiation): m l = majority{y (x) x Xl } Example (regreion): m l = Y (x) x Xl Example (laifiation): m l = majority{y (x) x Xl } Günther Sawitzki: CART Otober, 04 5 Coariable X with alue in X, for example X = R p. Repone Y with alue in Y, for example Y = R. and find model funtion m from loal model funtion m l : 7 Build partition by reurie plitting: X 0 = X Günther Sawitzki: CART Otober, 04 6 Coariable X with alue in X, for example X = R p. Repone Y with alue in Y, for example Y = R. and find model funtion m from loal model funtion m l : 8 Build partition by reurie plitting: X 0 = X for ome l for ome l Günther Sawitzki: CART Otober, 04 7 Günther Sawitzki: CART Otober, 04 8

2 9 0 Coariable X with alue in X, for example X = R p. Repone Y with alue in Y, for example Y = R. View on CART Reurie plitting ha two iew: iew Deiion tree and find model funtion m from loal model funtion m l : Build partition by reurie plitting: X 0 = X petal width irginia etoa eriolor irginia etoa Petal.Length <.45 Petal.Width <.75 Petal.Length < 4.95 eriolor irginia irginia Günther Sawitzki: CART Otober, 04 9 for ome l petal length We will ue the tree terminology node, edge here. Günther Sawitzki: CART Otober, 04 0 Elementary CART Only one ariable ued per plit. Continuou ariable: for ome l = {x X l : X k x 0 } {x X l : X k > x 0 } Fator: repet order, if appliable. Günther Sawitzki: CART Otober, 04 3 Günther Sawitzki: CART Otober, 04 4 Elementary CART a an optimiation problem For all urrent node X l and all ariable X k, find optimal plit point x 0. Selet optimal node and ariable. Need a quality meaure. For now, think of regreion a goodne of fit laifiation a milaifiation rate predition a predition error Elementary CART a an optimiation problem For all urrent node X l and all ariable X k, find optimal plit point x 0. Selet optimal node and ariable. Need a quality meaure. For now, think of regreion a goodne of fit laifiation a milaifiation rate predition a predition error Günther Sawitzki: CART Otober, 04 3 I 5 Günther Sawitzki: CART Otober, 04 4 II 6 Oberation X l with alue in X, for example X = R p. Clae Z l with alue,, K. Idea: reurie partition of X baed on a parameter X l, uh that Z l i homogeneou on the ell of the partition. Definition For a probability meaure P on,, K, let p k = P(k). A purity index i a funtion P Φ(P) uh that Φ ha exatly one maximum at P = ( K,, K ) Φ ha minima only at the meaure P = (, 0,, 0), P = (0,,, 0),..., P = (0, 0,, ) Φ i ymmetri, i.e. Φ(p,, p K ) = Φ(p π(),, p π(k) ) for any permutation π of,, K. Günther Sawitzki: CART Otober, 04 5 Günther Sawitzki: CART Otober, 04 6

3 III 7 8 Example (Shannon-Information) Φ(P) = K k= p klogp k Example (Gini Index of Dierity) Φ(P) = k k p kp k For X 0 X, let P X0 (k) = P(Z = k X X ). Let P X0 = #l:z l=k&x l X #l:x l X denote the orreponding empirial ditribution. The purity index for a partition i the weighted um P(X l )Φ(P Xl ). The purity index of a tree i the weighted um P(X l )Φ(P Xl ) l l:x l terminal node Günther Sawitzki: CART Otober, Günther Sawitzki: CART Otober, A imple regreion tree For all urrent node X l and all ariable X k, find optimal plit point x 0. Selet optimal node and ariable. Quality meaure: um of reidual quare. Note: thi i additie. So for eah node, we an ole the optimiation problem loally, and then elet the node giing the maximum redution in the um of reidual quare. ToDo Example A imple regreion tree For all urrent node X l and all ariable X k, find optimal plit point x 0. Selet optimal node and ariable. Quality meaure: um of reidual quare. Note: thi i additie. So for eah node, we an ole the optimiation problem loally, and then elet the node giing the maximum redution in the um of reidual quare. ToDo Example Günther Sawitzki: CART Otober, 04 9 Günther Sawitzki: CART Otober, 04 0 A imple regreion tree For all urrent node X l and eah ariable X k, find optimal plit point x 0. Selet optimal node and ariable. Quality meaure: e.g. um of reidual quare. Problem Degenerate to point et: need topping and pruning. A imple regreion tree For all urrent node X l and eah ariable X k, find optimal plit point x 0. Selet optimal node and ariable. Quality meaure: e.g. um of reidual quare. Problem Degenerate to point et: need topping and pruning. Data-baed um of reidual quare i biaed (oer-optimiti). Realiti quality meaure i needed. Data-baed um of reidual quare i biaed (oer-optimiti). Realiti quality meaure i needed. Günther Sawitzki: CART Otober, 04 A imple regreion tree 3 For all urrent node X l and eah ariable X k, find optimal plit point x 0. Selet optimal node and ariable. Quality meaure: e.g. um of reidual quare. Problem Degenerate to point et: need topping and pruning. Günther Sawitzki: CART Otober, 04 4 Reurie partitioning degenerate to point et: need topping and pruning. Early hope: Speify a topping trategy to aoid degenerate plit. Data-baed um of reidual quare i biaed (oer-optimiti). Realiti quality meaure i needed. Günther Sawitzki: CART Otober, 04 3 Günther Sawitzki: CART Otober, 04 4

4 5 6 Ad ho Do not plit a node if it ontain at mot n 0 oberation for ome hoen n 0 or Y i ontant or X i ontant n 0 = 5 or n 0 = 0, ay. Aoid degeneration, but till leae large ariane Still tree may need ome redution (pruning) in an additional tep. (To ome... ) Ad ho Do not plit a node if it ontain at mot n 0 oberation for ome hoen n 0 or Y i ontant or X i ontant n 0 = 5 or n 0 = 0, ay. Aoid degeneration, but till leae large ariane Still tree may need ome redution (pruning) in an additional tep. (To ome... ) Günther Sawitzki: CART Otober, Günther Sawitzki: CART Otober, Ad ho Do not plit a node if it ontain at mot n 0 oberation for ome hoen n 0 or Y i ontant or X i ontant n 0 = 5 or n 0 = 0, ay. Simple CART i a greedy algorithm, and only look at one tep. To improe algorithm, a multi tep iew i neeary. Complexity? There are aymptotially about.5 L ubtree of a balaned binary tree with L leae. P. Feigin, aording to [] p. 84. Aoid degeneration, but till leae large ariane Still tree may need ome redution (pruning) in an additional tep. (To ome... ) Günther Sawitzki: CART Otober, Günther Sawitzki: CART Otober, Simple CART i a greedy algorithm, and only look at one tep. To improe algorithm, a multi tep iew i neeary. Complexity? There are aymptotially about.5 L ubtree of a balaned binary tree with L leae. P. Feigin, aording to [] p. 84. Simple CART i a greedy algorithm, and only look at one tep. To improe algorithm, a multi tep iew i neeary. Complexity? There are aymptotially about.5 L ubtree of a balaned binary tree with L leae. P. Feigin, aording to [] p. 84. Günther Sawitzki: CART Otober, Günther Sawitzki: CART Otober, Exhautie earh oon not feaible. Bold idea: Build one large tree until it hit a topping barrier. Ue thi tree a an exploration of the data and ut down unneeary branhe. Heuriti to find andidate for good tree: Start with a large tree. Find a good ubtree with ame root t 0. Gien T 0, a rooted tree i a tree with ame root a T 0. Günther Sawitzki: CART Otober, 04 3 Günther Sawitzki: CART Otober, 04 3

5 33 34 An attempt to formaliation: For a tree T, define ome rik etimate R(T ), e.g. the milaifiation rate uing the data. Thi i too optimiti, due to a tendeny to oer-fitting. Add a omplexity penalty R α (T ) := R(T ) + α omplexity(t ) for ome free parameter α, where α = 0 mean: no penalty. omplexity(t ) = # of node of T, or omplexity(t ) = # of leae of T, or... For a binary tree: # of node of T = # of leae of T. For now: omplexity(t ) = # of leae of T = T. An attempt to formaliation: For a tree T, define ome rik etimate R(T ), e.g. the milaifiation rate uing the data. Thi i too optimiti, due to a tendeny to oer-fitting. Add a omplexity penalty R α (T ) := R(T ) + α omplexity(t ) for ome free parameter α, where α = 0 mean: no penalty. omplexity(t ) = # of node of T, or omplexity(t ) = # of leae of T, or... For a binary tree: # of node of T = # of leae of T. For now: omplexity(t ) = # of leae of T = T. Günther Sawitzki: CART Otober, Günther Sawitzki: CART Otober, An attempt to formaliation: For a tree T, define ome rik etimate R(T ), e.g. the milaifiation rate uing the data. Thi i too optimiti, due to a tendeny to oer-fitting. Add a omplexity penalty R α (T ) := R(T ) + α omplexity(t ) for ome free parameter α, where α = 0 mean: no penalty. omplexity(t ) = # of node of T, or omplexity(t ) = # of leae of T, or... For a binary tree: # of node of T = # of leae of T. For now: omplexity(t ) = # of leae of T = T. Start with a large tree T 0. For eah α, find ome rooted ubtree T T 0 minimiing R α (T ) := R(T ) + α T Günther Sawitzki: CART Otober, Günther Sawitzki: CART Otober, T T 0 i minimiing R α (T ) := R(T ) + α T T T 0 i minimiing R α (T ) := R(T ) + α T iff T = argmin T =L T 0 R α (T ) := R(T ) + α T iff T = argmin T =L T 0 R α (T ) := R(T ) + α T So we an build a equene T L = T 0, T L,..., T = t 0 of tree, minimiing R α for ome α 0 = 0, α L,..., α. R α i ontant on the interal and T l i minimier on the interal. There an be eeral tree with the ame alue of R α (T ), and minimiing tree need not be neted. So we an build a equene T L = T 0, T L,..., T = t 0 of tree, minimiing R α for ome α 0 = 0, α L,..., α. R α i ontant on the interal and T l i minimier on the interal. There an be eeral tree with the ame alue of R α (T ), and minimiing tree need not be neted. Günther Sawitzki: CART Otober, Günther Sawitzki: CART Otober, Remark It i poible to build a equene of neted minimiing tree from T 0 to {t 0 }. If there i a tree whih i a ubtree of all tree optimiing R α (T ), it will be denoted by T (α). Lemma If T 0 i not triial, T (α) exit. For a non-triial tree T and a non-terminal node t, let T t be the ubtree rooted at t. Definition T i pruned at t if T t i replaed by {t}. Let Remark g(t, T ) > α R α (t) > R α (T t ) g(t, T ) := R(t) R(T t) T t {t} Günther Sawitzki: CART Otober, Günther Sawitzki: CART Otober, 04 40

6 4 4 Günther Sawitzki: CART Otober, Günther Sawitzki: CART Otober, Günther Sawitzki: CART Otober, Tet et enario 45 Degenerate to point et: need topping and pruning. Split data et into training et and ealuation et. Tree ontrution ue only training et. Quality i judged by applying regreion tree to ealuation et. RSS t = n eal (Y i n Ŷi) eal Sum of term with idential ditribution. Standard deiation i= ( ) ( ( 4 ( ) )) ) (E Y Ŷ E Y Ŷ neal Günther Sawitzki: CART Otober, Tet et enario 46 RSS t = n eal (Y i n Ŷi) eal Sum of term with idential ditribution. Standard deiation Etimated by i= ( ) ( ( 4 ( ) )) ) (E Y Ŷ E Y Ŷ neal ( neal n eal n eal i= (Y i Ŷi) 4 RSSt ) Günther Sawitzki: CART Otober, Cro alidation enario 47 Diide the oberation V in group of nearly equal ize V,..., V k. For etimation, ue the omplete data et. Define the k-fold ro-alidation etimate of the quality index a the aerage R CV = k Rj t k where Rj t i the tet ample etimate, uing V \ V j a a training ample, and V j a an ealuation ample. j= Speial ae: k = n i the leae-one-out etimate. Günther Sawitzki: CART Otober, Cro alidation enario 48 Diide the oberation V in group of nearly equal ize V,..., V k. For etimation, ue the omplete data et. Define the k-fold ro-alidation etimate of the quality index a the aerage R CV = k Rj t k where Rj t i the tet ample etimate, uing V \ V j a a training ample, and V j a an ealuation ample. j= Speial ae: k = n i the leae-one-out etimate. Günther Sawitzki: CART Otober, Günther Sawitzki: CART Otober, 04 48

7 Cro alidation enario 49 Diide the oberation V in group of nearly equal ize V,..., V k. For etimation, ue the omplete data et. Define the k-fold ro-alidation etimate of the quality index a the aerage R CV = k Rj t k where Rj t i the tet ample etimate, uing V \ V j a a training ample, and V j a an ealuation ample. j= Speial ae: k = n i the leae-one-out etimate. Cro alidation enario 50 Diide the oberation V in group of nearly equal ize V,..., V k. For etimation, ue the omplete data et. Define the k-fold ro-alidation etimate of the quality index a the aerage R CV = k Rj t k where Rj t i the tet ample etimate, uing V \ V j a a training ample, and V j a an ealuation ample. j= Speial ae: k = n i the leae-one-out etimate. Günther Sawitzki: CART Otober, Günther Sawitzki: CART Otober, Günther Sawitzki: CART Otober, 04 5 Günther Sawitzki: CART Otober, Referene I Leo Breiman, Jerome H. Friedman, Rihard A. Olhen, and Charle J. Stone. Claifiation and Regreion Tree. Wadworth, Belmont, CA., 984. Jerome H. Friedman. Loal learning baed on reurie oering. Tehnial report, Department of Statiti, Stanford Unierity, 996. Jerome H. Friedman. On bia, ariane, 0/ lo, and the ure-of-dimenionality. Data Min. Knowl. Dio., ():55 77, 997. Günther Sawitzki: CART Otober, Günther Sawitzki: CART Otober, Referene II Index I Treor Hatie, Robert Tibhirani, and Jerome H. Friedman. The Element of Statitial Learning. Springer Serie in Statiti. Springer, nd edition, 009. Brian D. Ripley. Pattern reognition and neural network. Cambridge Unierity Pre, Cambridge, 996. bagging, 57 booting, 57 random foret, 57 Günther Sawitzki: CART Otober, Günther Sawitzki: CART Otober, 04 56

8 57 Günther Sawitzki: CART Otober, 04 57

Discrimination and Classification

Discrimination and Classification Dirimination and Claifiation Nathaniel E. Helwig Aitant Profeor of Pyhology and Statiti Unierity of Minneota (Twin Citie) Updated 14-Mar-2017 Nathaniel E. Helwig (U of Minneota) Dirimination and Claifiation

More information

WEIGHTED PRINCIPAL COMPONENT ANALYSIS BASED ON FUZZY CLUSTERING MIKA SATO-ILIC. Received May 12, 2003

WEIGHTED PRINCIPAL COMPONENT ANALYSIS BASED ON FUZZY CLUSTERING MIKA SATO-ILIC. Received May 12, 2003 Sientiae Mathematiae Japoniae Online, Vol. 10, (2004), 359 368 359 WEIGHTED PRINCIPAL COMPONENT ANALYSIS BASED ON FUZZY CLUSTERING MIKA SATO-ILIC Reeied May 12, 2003 Abtrat. In thi paper, we propoe a weighted

More information

5.2.6 COMPARISON OF QUALITY CONTROL AND VERIFICATION TESTS

5.2.6 COMPARISON OF QUALITY CONTROL AND VERIFICATION TESTS 5..6 COMPARISON OF QUALITY CONTROL AND VERIFICATION TESTS Thi proedure i arried out to ompare two different et of multiple tet reult for finding the ame parameter. Typial example would be omparing ontrator

More information

OLIGONUCLEOTIDE microarrays are widely used

OLIGONUCLEOTIDE microarrays are widely used Evolution Strategy with Greedy Probe Seletion Heuriti for the Non-Unique Oligonuleotide Probe Seletion Problem Lili Wang, Alioune Ngom, Robin Gra and Lui Rueda Abtrat In order to aurately meaure the gene

More information

Boosted Partial Least-Squares. Regression

Boosted Partial Least-Squares. Regression Booted Partial Leat-Square Regreion Jean-Françoi Durand Montpellier II Unierity, Frane E-Mail: upport@jf-durand-pl.om Web ite: www.jf-durand-pl.om Booted PLS Regreion: Séminaire J.P. Fénelon 008- I. Introdution

More information

Chapter 12 Simple Linear Regression

Chapter 12 Simple Linear Regression Chapter 1 Simple Linear Regreion Introduction Exam Score v. Hour Studied Scenario Regreion Analyi ued to quantify the relation between (or more) variable o you can predict the value of one variable baed

More information

Kalman Filter. Wim van Drongelen, Introduction

Kalman Filter. Wim van Drongelen, Introduction alman Filter Wim an Drongelen alman Filter Wim an Drongelen, 03. Introduction Getting to undertand a ytem can be quite a challenge. One approach i to create a model, an abtraction of the ytem. The idea

More information

MINITAB Stat Lab 3

MINITAB Stat Lab 3 MINITAB Stat 20080 Lab 3. Statitical Inference In the previou lab we explained how to make prediction from a imple linear regreion model and alo examined the relationhip between the repone and predictor

More information

If Y is normally Distributed, then and 2 Y Y 10. σ σ

If Y is normally Distributed, then and 2 Y Y 10. σ σ ull Hypothei Significance Teting V. APS 50 Lecture ote. B. Dudek. ot for General Ditribution. Cla Member Uage Only. Chi-Square and F-Ditribution, and Diperion Tet Recall from Chapter 4 material on: ( )

More information

To determine the biasing conditions needed to obtain a specific gain each stage must be considered.

To determine the biasing conditions needed to obtain a specific gain each stage must be considered. PHYSIS 56 Experiment 9: ommon Emitter Amplifier A. Introdution A ommon-emitter oltage amplifier will be tudied in thi experiment. You will inetigate the fator that ontrol the midfrequeny gain and the low-and

More information

Lag-Lead Compensator Design

Lag-Lead Compensator Design Lag-Lead Compenator Deign ELEC 3 Spring 08 Lag or Lead Struture A bai ompenator onit of a gain, one real pole and one real zero Two type: phae-lead and phae-lag Phae-lead: provide poitive phae hift and

More information

THE SOLAR SYSTEM. We begin with an inertial system and locate the planet and the sun with respect to it. Then. F m. Then

THE SOLAR SYSTEM. We begin with an inertial system and locate the planet and the sun with respect to it. Then. F m. Then THE SOLAR SYSTEM We now want to apply what we have learned to the olar ytem. Hitorially thi wa the great teting ground for mehani and provided ome of it greatet triumph, uh a the diovery of the outer planet.

More information

q-expansions of vector-valued modular forms of negative weight

q-expansions of vector-valued modular forms of negative weight Ramanujan J 2012 27:1 13 DOI 101007/11139-011-9299-9 q-expanion of vetor-valued modular form of negative weight Joe Gimenez Wiam Raji Reeived: 19 July 2010 / Aepted: 2 February 2011 / Publihed online:

More information

SIMPLE LINEAR REGRESSION

SIMPLE LINEAR REGRESSION SIMPLE LINEAR REGRESSION In linear regreion, we conider the frequency ditribution of one variable (Y) at each of everal level of a econd variable (). Y i known a the dependent variable. The variable for

More information

Lecture 9: Shor s Algorithm

Lecture 9: Shor s Algorithm Quantum Computation (CMU 8-859BB, Fall 05) Lecture 9: Shor Algorithm October 7, 05 Lecturer: Ryan O Donnell Scribe: Sidhanth Mohanty Overview Let u recall the period finding problem that wa et up a a function

More information

Lecture 2: Computer Arithmetic: Adders

Lecture 2: Computer Arithmetic: Adders CMU 8-447 S 9 L2-29 8-447 Leture 2: Computer Arithmeti: Adder Jame C. Hoe Dept of ECE, CMU January 4, 29 Announement: No la on Monday Verilog Refreher next Wedneday Review P&H Ch 3 Handout: Lab and HW

More information

Lecture 16. Kinetics and Mass Transfer in Crystallization

Lecture 16. Kinetics and Mass Transfer in Crystallization Leture 16. Kineti and Ma Tranfer in Crytallization Crytallization Kineti Superaturation Nuleation - Primary nuleation - Seondary nuleation Crytal Growth - Diffuion-reation theory - Srew-diloation theory

More information

CART Classification and Regression Trees Trees can be viewed as basis expansions of simple functions. f(x) = c m 1(x R m )

CART Classification and Regression Trees Trees can be viewed as basis expansions of simple functions. f(x) = c m 1(x R m ) CART Classification and Regression Trees Trees can be viewed as basis expansions of simple functions with R 1,..., R m R p disjoint. f(x) = M c m 1(x R m ) m=1 The CART algorithm is a heuristic, adaptive

More information

Comparing Means: t-tests for Two Independent Samples

Comparing Means: t-tests for Two Independent Samples Comparing ean: t-tet for Two Independent Sample Independent-eaure Deign t-tet for Two Independent Sample Allow reearcher to evaluate the mean difference between two population uing data from two eparate

More information

BAGGING PREDICTORS AND RANDOM FOREST

BAGGING PREDICTORS AND RANDOM FOREST BAGGING PREDICTORS AND RANDOM FOREST DANA KANER M.SC. SEMINAR IN STATISTICS, MAY 2017 BAGIGNG PREDICTORS / LEO BREIMAN, 1996 RANDOM FORESTS / LEO BREIMAN, 2001 THE ELEMENTS OF STATISTICAL LEARNING (CHAPTERS

More information

SF2930 Regression Analysis

SF2930 Regression Analysis SF2930 Regression Analysis Alexandre Chotard Tree-based regression and classication 20 February 2017 1 / 30 Idag Overview Regression trees Pruning Bagging, random forests 2 / 30 Today Overview Regression

More information

Machine Leanring Theory and Applications: third lecture

Machine Leanring Theory and Applications: third lecture Machine Leanring Theory and Applications: third lecture Arnak Dalalyan ENSAE PARISTECH 12 avril 2016 Framework and notation Framework and notation We observe (X i, Y i ) X Y, i = 1,..., n independent randomly

More information

Regression. What is regression? Linear Regression. Cal State Northridge Ψ320 Andrew Ainsworth PhD

Regression. What is regression? Linear Regression. Cal State Northridge Ψ320 Andrew Ainsworth PhD Regreion Cal State Northridge Ψ30 Andrew Ainworth PhD What i regreion? How do we predict one variable from another? How doe one variable change a the other change? Caue and effect Linear Regreion A technique

More information

Suggested Answers To Exercises. estimates variability in a sampling distribution of random means. About 68% of means fall

Suggested Answers To Exercises. estimates variability in a sampling distribution of random means. About 68% of means fall Beyond Significance Teting ( nd Edition), Rex B. Kline Suggeted Anwer To Exercie Chapter. The tatitic meaure variability among core at the cae level. In a normal ditribution, about 68% of the core fall

More information

Relativity III. Review: Kinetic Energy. Example: He beam from THIA K = 300keV v =? Exact vs non-relativistic calculations Q.37-3.

Relativity III. Review: Kinetic Energy. Example: He beam from THIA K = 300keV v =? Exact vs non-relativistic calculations Q.37-3. Relatiity III Today: Time dilation eamples The Lorentz Transformation Four-dimensional spaetime The inariant interal Eamples Reiew: Kineti Energy General relation for total energy: Rest energy, 0: Kineti

More information

DISCHARGE MEASUREMENT IN TRAPEZOIDAL LINED CANALS UTILIZING HORIZONTAL AND VERTICAL TRANSITIONS

DISCHARGE MEASUREMENT IN TRAPEZOIDAL LINED CANALS UTILIZING HORIZONTAL AND VERTICAL TRANSITIONS Ninth International Water Tehnology Conferene, IWTC9 005, Sharm El-Sheikh, Egypt 63 DISCHARGE MEASUREMENT IN TRAPEZOIDAL LINED CANALS UTILIZING HORIZONTAL AND VERTICAL TRANSITIONS Haan Ibrahim Mohamed

More information

Decision Trees. CS57300 Data Mining Fall Instructor: Bruno Ribeiro

Decision Trees. CS57300 Data Mining Fall Instructor: Bruno Ribeiro Decision Trees CS57300 Data Mining Fall 2016 Instructor: Bruno Ribeiro Goal } Classification without Models Well, partially without a model } Today: Decision Trees 2015 Bruno Ribeiro 2 3 Why Trees? } interpretable/intuitive,

More information

Holdout and Cross-Validation Methods Overfitting Avoidance

Holdout and Cross-Validation Methods Overfitting Avoidance Holdout and Cross-Validation Methods Overfitting Avoidance Decision Trees Reduce error pruning Cost-complexity pruning Neural Networks Early stopping Adjusting Regularizers via Cross-Validation Nearest

More information

Entropic Graphs for Manifold Learning

Entropic Graphs for Manifold Learning Entropi Graphs for Manifold Learning Jose A. Costa and Alfred O. Hero III Department of Eletrial Engineering and Computer Siene Uniersity of Mihigan, Ann Arbor, MI 9 Email: josta@umih.edu, hero@ees.umih.edu

More information

Confusion matrices. True / False positives / negatives. INF 4300 Classification III Anne Solberg The agenda today: E.g., testing for cancer

Confusion matrices. True / False positives / negatives. INF 4300 Classification III Anne Solberg The agenda today: E.g., testing for cancer INF 4300 Claification III Anne Solberg 29.10.14 The agenda today: More on etimating claifier accuracy Cure of dimenionality knn-claification K-mean clutering x i feature vector for pixel i i- The cla label

More information

CSCI565 - Compiler Design

CSCI565 - Compiler Design CSCI565 - Compiler Deign Spring 6 Due Dte: Fe. 5, 6 t : PM in Cl Prolem [ point]: Regulr Expreion nd Finite Automt Develop regulr expreion (RE) tht detet the longet tring over the lphet {-} with the following

More information

Informal Definition: Telling things apart

Informal Definition: Telling things apart 9. Decision Trees Informal Definition: Telling things apart 2 Nominal data No numeric feature vector Just a list or properties: Banana: longish, yellow Apple: round, medium sized, different colors like

More information

μ + = σ = D 4 σ = D 3 σ = σ = All units in parts (a) and (b) are in V. (1) x chart: Center = μ = 0.75 UCL =

μ + = σ = D 4 σ = D 3 σ = σ = All units in parts (a) and (b) are in V. (1) x chart: Center = μ = 0.75 UCL = Our online Tutor are available 4*7 to provide Help with Proce control ytem Homework/Aignment or a long term Graduate/Undergraduate Proce control ytem Project. Our Tutor being experienced and proficient

More information

Lecture 7: Testing Distributions

Lecture 7: Testing Distributions CSE 5: Sublinear (and Streaming) Algorithm Spring 014 Lecture 7: Teting Ditribution April 1, 014 Lecturer: Paul Beame Scribe: Paul Beame 1 Teting Uniformity of Ditribution We return today to property teting

More information

Machine Learning. Nathalie Villa-Vialaneix - Formation INRA, Niveau 3

Machine Learning. Nathalie Villa-Vialaneix -  Formation INRA, Niveau 3 Machine Learning Nathalie Villa-Vialaneix - nathalie.villa@univ-paris1.fr http://www.nathalievilla.org IUT STID (Carcassonne) & SAMM (Université Paris 1) Formation INRA, Niveau 3 Formation INRA (Niveau

More information

A MATHEMATICAL MODEL FOR TWO DIMENSIONAL LOADING PROBLEM IN CROSS-DOCKING NETWORK DESIGN. İlker Küçükoğlu, Aslı Aksoy, Seval Ene and Nursel Öztürk

A MATHEMATICAL MODEL FOR TWO DIMENSIONAL LOADING PROBLEM IN CROSS-DOCKING NETWORK DESIGN. İlker Küçükoğlu, Aslı Aksoy, Seval Ene and Nursel Öztürk Mathematial and Computational Appliation Vol. 18 No. 3 pp. 273-282 2013 A MATHEMATICAL MODEL FOR TWO DIMENSIONAL LOADING PROBLEM IN CROSS-DOCKING NETWORK DESIGN İlker Küçükoğlu Alı Akoy Seval Ene and Nurel

More information

Advanced Digital Signal Processing. Stationary/nonstationary signals. Time-Frequency Analysis... Some nonstationary signals. Time-Frequency Analysis

Advanced Digital Signal Processing. Stationary/nonstationary signals. Time-Frequency Analysis... Some nonstationary signals. Time-Frequency Analysis Advanced Digital ignal Proceing Prof. Nizamettin AYDIN naydin@yildiz.edu.tr Time-Frequency Analyi http://www.yildiz.edu.tr/~naydin 2 tationary/nontationary ignal Time-Frequency Analyi Fourier Tranform

More information

Einstein s theory of special relativity

Einstein s theory of special relativity Einstein s theory of speial relatiity Announements: First homework assignment is online. You will need to read about time dilation (1.8) to answer problem #3 and for the definition of γ for problem #4.

More information

Critical Percolation Probabilities for the Next-Nearest-Neighboring Site Problems on Sierpinski Carpets

Critical Percolation Probabilities for the Next-Nearest-Neighboring Site Problems on Sierpinski Carpets Critial Perolation Probabilitie for the Next-Nearet-Neighboring Site Problem on Sierpinki Carpet H. B. Nie, B. M. Yu Department of Phyi, Huazhong Univerity of Siene and Tehnology, Wuhan 430074, China K.

More information

Growing a Large Tree

Growing a Large Tree STAT 5703 Fall, 2004 Data Mining Methodology I Decision Tree I Growing a Large Tree Contents 1 A Single Split 2 1.1 Node Impurity.................................. 2 1.2 Computation of i(t)................................

More information

2. SYNTHESIS OF ANALOG FILTERS

2. SYNTHESIS OF ANALOG FILTERS ,. SYNTHESIS OF ANALO FILTERS 9. a) H BW ( j ) = -------------------------------- where ε =.A max ε N A max = log Hj ( ) = ( ) ε log( ) A min = log Hj ( ) = ( ) ε log N N.A min ----- ε = -------------------------------

More information

Data Mining. CS57300 Purdue University. Bruno Ribeiro. February 8, 2018

Data Mining. CS57300 Purdue University. Bruno Ribeiro. February 8, 2018 Data Mining CS57300 Purdue University Bruno Ribeiro February 8, 2018 Decision trees Why Trees? interpretable/intuitive, popular in medical applications because they mimic the way a doctor thinks model

More information

15.12 Applications of Suffix Trees

15.12 Applications of Suffix Trees 248 Algorithms in Bioinformatis II, SoSe 07, ZBIT, D. Huson, May 14, 2007 15.12 Appliations of Suffix Trees 1. Searhing for exat patterns 2. Minimal unique substrings 3. Maximum unique mathes 4. Maximum

More information

On-demand Mobile Charger Scheduling for Effective Coverage in Wireless Rechargeable Sensor Networks

On-demand Mobile Charger Scheduling for Effective Coverage in Wireless Rechargeable Sensor Networks On-demand Mobile Charger Sheduling for Effetive Coverage in Wirele Rehargeable Senor Network Lintong Jiang, Haipeng Dai, Xiaobing Wu, and Guihai Chen State Key Laboratory for Novel Software Tehnology,

More information

Constant Force: Projectile Motion

Constant Force: Projectile Motion Contant Force: Projectile Motion Abtract In thi lab, you will launch an object with a pecific initial velocity (magnitude and direction) and determine the angle at which the range i a maximum. Other tak,

More information

Chapter 7. Network Flow. CS 350: Winter 2018

Chapter 7. Network Flow. CS 350: Winter 2018 Chapter 7 Network Flow CS 3: Winter 1 1 Soviet Rail Network, Reference: On the hitory of the tranportation and maximum flow problem. Alexander Schrijver in Math Programming, 1: 3,. Maximum Flow and Minimum

More information

CART Classification and Regression Trees Trees can be viewed as basis expansions of simple functions

CART Classification and Regression Trees Trees can be viewed as basis expansions of simple functions CART Classification and Regression Trees Trees can be viewed as basis expansions of simple functions f (x) = M c m 1(x R m ) m=1 with R 1,..., R m R p disjoint. The CART algorithm is a heuristic, adaptive

More information

Probability Distributions

Probability Distributions STATGRAPHICS Re. 6805 Probability Distributions Summary... Data Input... Analysis Summary... 3 Analysis Options... 3 Cumulatie Distribution... 4 Inerse CDF... 5 Random Numbers... 6 DensityMass Funtion...

More information

SCOUR HOLE CHARACTERISTICS AROUND A VERTICAL PIER UNDER CLEARWATER SCOUR CONDITIONS

SCOUR HOLE CHARACTERISTICS AROUND A VERTICAL PIER UNDER CLEARWATER SCOUR CONDITIONS ARPN Journal of Engineering and Applied Siene 2006-2012 Aian Reearh Publihing Network (ARPN). All right reerved. www.arpnjournal.om SCOUR HOLE CHARACTERISTICS AROUND A VERTICAL PIER UNDER CLEARWATER SCOUR

More information

Statistical Consulting Topics Classification and Regression Trees (CART)

Statistical Consulting Topics Classification and Regression Trees (CART) Statistical Consulting Topics Classification and Regression Trees (CART) Suppose the main goal in a data analysis is the prediction of a categorical variable outcome. Such as in the examples below. Given

More information

Social Studies 201 Notes for March 18, 2005

Social Studies 201 Notes for March 18, 2005 1 Social Studie 201 Note for March 18, 2005 Etimation of a mean, mall ample ize Section 8.4, p. 501. When a reearcher ha only a mall ample ize available, the central limit theorem doe not apply to the

More information

Chapter 4. Simulations. 4.1 Introduction

Chapter 4. Simulations. 4.1 Introduction Chapter 4 Simulation 4.1 Introdution In the previou hapter, a methodology ha been developed that will be ued to perform the ontrol needed for atuator haraterization. A tudy uing thi methodology allowed

More information

Degree Ramsey numbers of graphs

Degree Ramsey numbers of graphs Degree Ramey number of graph William B. Kinnerley, Kein G. Milan, Dougla B. Wet Abtract Let H G mean that eery -coloring of E(H) produce a monochromatic copy of G in ome color cla. Let the -color degree

More information

Data Mining Classification Trees (2)

Data Mining Classification Trees (2) Data Mining Classification Trees (2) Ad Feelders Universiteit Utrecht September 14, 2017 Ad Feelders ( Universiteit Utrecht ) Data Mining September 14, 2017 1 / 46 Basic Tree Construction Algorithm Construct

More information

1. Tc from BCS 2. Ginzburg-Landau theory 3. Josephson effect

1. Tc from BCS 2. Ginzburg-Landau theory 3. Josephson effect Condensed Matter Physis 6 Leture 6/: More on superondutiity.. T from BCS. Ginzburg-Landau theory 3. Josephson effet Referenes: Ashroft & Mermin 34 Taylor & Heinonen 7.-7.5 7.7 blaboard Reall from last

More information

Suggestions - Problem Set (a) Show the discriminant condition (1) takes the form. ln ln, # # R R

Suggestions - Problem Set (a) Show the discriminant condition (1) takes the form. ln ln, # # R R Suggetion - Problem Set 3 4.2 (a) Show the dicriminant condition (1) take the form x D Ð.. Ñ. D.. D. ln ln, a deired. We then replace the quantitie. 3ß D3 by their etimate to get the proper form for thi

More information

ES 247 Fracture Mechanics Zhigang Suo. Applications of Fracture Mechanics

ES 247 Fracture Mechanics   Zhigang Suo. Applications of Fracture Mechanics Appliation of Frature Mehani Many appliation of frature mehani are baed on the equation σ a Γ = β. E Young modulu i uually known. Of the other four quantitie, if three are known, the equation predit the

More information

Chapter 7. Network Flow. Slides by Kevin Wayne. Copyright 2005 Pearson-Addison Wesley. All rights reserved.

Chapter 7. Network Flow. Slides by Kevin Wayne. Copyright 2005 Pearson-Addison Wesley. All rights reserved. Chapter 7 Network Flow Slide by Kevin Wayne. Copyright 5 Pearon-Addion Weley. All right reerved. Soviet Rail Network, 55 Reference: On the hitory of the tranportation and maximum flow problem. Alexander

More information

Probabilistic Graphical Models

Probabilistic Graphical Models Probabilisti Graphial Models David Sontag New York University Leture 12, April 19, 2012 Aknowledgement: Partially based on slides by Eri Xing at CMU and Andrew MCallum at UMass Amherst David Sontag (NYU)

More information

Asymptotic Normality of an Entropy Estimator with Exponentially Decaying Bias

Asymptotic Normality of an Entropy Estimator with Exponentially Decaying Bias Asymptotic Normality of an Entropy Estimator with Exponentially Decaying Bias Zhiyi Zhang Department of Mathematics and Statistics Uniersity of North Carolina at Charlotte Charlotte, NC 28223 Abstract

More information

Solution to Test #1.

Solution to Test #1. Solution to Tet #. Problem #. Lited below are recorded peed (in mile per hour, mph) of randomly elected car on a ection of Freeway 5 in Lo Angele. Data are orted. 56 58 58 59 60 60 6 6 65 65 65 65 66 66

More information

Dyadic Classification Trees via Structural Risk Minimization

Dyadic Classification Trees via Structural Risk Minimization Dyadic Classification Trees via Structural Risk Minimization Clayton Scott and Robert Nowak Department of Electrical and Computer Engineering Rice University Houston, TX 77005 cscott,nowak @rice.edu Abstract

More information

A Luenberger Soil Quality Indicator

A Luenberger Soil Quality Indicator A Luenberger Soil Quality Indicator Atakelty Hailu Univerity of Wetern Autralia atakelty.hailu@uwa.edu.au and Robert G. Chamber Univerity of Maryland and Univerity of Wetern Autralia rchamber@arec.umd.edu

More information

Einstein's Energy Formula Must Be Revised

Einstein's Energy Formula Must Be Revised Eintein' Energy Formula Mut Be Reied Le Van Cuong uong_le_an@yahoo.om Information from a iene journal how that the dilation of time in Eintein peial relatie theory wa proen by the experiment of ientit

More information

PROBABILITY AND STATISTICS. Least Squares Regression

PROBABILITY AND STATISTICS. Least Squares Regression PROBABILITY AND STATISTICS Leat Square Regreion LEAST-SQUARES REGRESSION What doe correlation give u? If a catterplot how a linear relationhip one wa to ummarize the overall pattern of the catterplot i

More information

Really useful mathematics

Really useful mathematics i Really ueful mathemati William Harwin Department of Cyberneti Shool of Sytem Engineering November, 7 To be publihed at www.rdg.a.uk/ hhawin/dnb And what i that to me whoe mind i full of indiie and urd,

More information

Yoram Gat. Technical report No. 548, March Abstract. A classier is said to have good generalization ability if it performs on

Yoram Gat. Technical report No. 548, March Abstract. A classier is said to have good generalization ability if it performs on A bound concerning the generalization ability of a certain cla of learning algorithm Yoram Gat Univerity of California, Berkeley Technical report No. 548, March 999 Abtract A claier i aid to have good

More information

IRANIAN JOURNAL OF ELECTRICAL AND COMPUTER ENGINEERING, VOL. 7, NO. 1, WINTER-SPRING M. Rahimi, H. Mokhtari, and Gh.

IRANIAN JOURNAL OF ELECTRICAL AND COMPUTER ENGINEERING, VOL. 7, NO. 1, WINTER-SPRING M. Rahimi, H. Mokhtari, and Gh. IRANIAN JOURNAL OF ELECTRICAL AND COMPUTER ENGINEERING VOL. 7 NO. WINTER-SPRING 008 6 A New Atie Method in Damping Poible Reonane in Atie Filter M. Rahimi H. Mokhtari and Gh. Zaarabadi Abtrat Thi paper

More information

Application of Fuzzy C-Means Clustering in Power System Model Reduction for Controller Design

Application of Fuzzy C-Means Clustering in Power System Model Reduction for Controller Design Proeeding of the 5th WSEAS Int. Conf. on COMPUAIONAL INELLIGENCE, MAN-MACHINE SYSEMS AND CYBERNEICS, Venie, Italy, November 20-22, 2006 223 Appliation of Fuzzy C-Mean Clutering in Power Sytem Model Redution

More information

Today. Flow review. Augmenting paths. Ford-Fulkerson Algorithm. Intro to cuts (reason: prove correctness)

Today. Flow review. Augmenting paths. Ford-Fulkerson Algorithm. Intro to cuts (reason: prove correctness) Today Flow review Augmenting path Ford-Fulkeron Algorithm Intro to cut (reaon: prove correctne) Flow Network = ource, t = ink. c(e) = capacity of edge e Capacity condition: f(e) c(e) Conervation condition:

More information

Decision trees COMS 4771

Decision trees COMS 4771 Decision trees COMS 4771 1. Prediction functions (again) Learning prediction functions IID model for supervised learning: (X 1, Y 1),..., (X n, Y n), (X, Y ) are iid random pairs (i.e., labeled examples).

More information

NEGATIVE z Scores. TABLE A-2 Standard Normal (z) Distribution: Cumulative Area from the LEFT. (continued)

NEGATIVE z Scores. TABLE A-2 Standard Normal (z) Distribution: Cumulative Area from the LEFT. (continued) NEGATIVE z Score z 0 TALE A- Standard Normal (z) Ditribution: Cumulative Area from the LEFT z.00.01.0.03.04.05.06.07.08.09-3.50 and lower.0001-3.4.0003.0003.0003.0003.0003.0003.0003.0003.0003.000-3.3.0005.0005.0005.0004.0004.0004.0004.0004.0004.0003-3..0007.0007.0006.0006.0006.0006.0006.0005.0005.0005-3.1.0010.0009.0009.0009.0008.0008.0008.0008.0007.0007-3.0.0013.0013.0013.001.001.0011.0011.0011.0010.0010

More information

OBSERVATIONS ON BAGGING

OBSERVATIONS ON BAGGING OBSERVATIONS ON BAGGING Andreas Buja and Werner Stuetzle Uniersity of Pennsylania and Uniersity of Washington Abstract: Bagging is a deice intended for reducing the prediction error of learning algorithms.

More information

Simplified Modeling, Analysis and Simulation of Permanent Magnet Brushless Direct Current Motors for Sensorless Operation

Simplified Modeling, Analysis and Simulation of Permanent Magnet Brushless Direct Current Motors for Sensorless Operation Amerian Journal of Applied Sienes 9 (7): 1046-1054, 2012 ISSN 1546-9239 2012 Siene Publiations Simplified Modeling, Analysis and Simulation of Permanent Magnet Brushless Diret Current Motors for Sensorless

More information

Complexity of Regularization RBF Networks

Complexity of Regularization RBF Networks Complexity of Regularization RBF Networks Mark A Kon Department of Mathematis and Statistis Boston University Boston, MA 02215 mkon@buedu Leszek Plaskota Institute of Applied Mathematis University of Warsaw

More information

Final Overview. Introduction to ML. Marek Petrik 4/25/2017

Final Overview. Introduction to ML. Marek Petrik 4/25/2017 Final Overview Introduction to ML Marek Petrik 4/25/2017 This Course: Introduction to Machine Learning Build a foundation for practice and research in ML Basic machine learning concepts: max likelihood,

More information

Sidelobe-Suppression Technique Applied To Binary Phase Barker Codes

Sidelobe-Suppression Technique Applied To Binary Phase Barker Codes Journal of Engineering and Development, Vol. 16, No.4, De. 01 ISSN 1813-78 Sidelobe-Suppreion Tehnique Applied To Binary Phae Barker Code Aitant Profeor Dr. Imail M. Jaber Al-Mutaniriya Univerity College

More information

Bio-Systems Modeling and Control

Bio-Systems Modeling and Control Bio-Sytem Modelig ad otrol Leture Ezyme ooperatio i Ezyme Dr. Zvi Roth FAU Ezyme with Multiple Bidig Site May ezyme have more tha oe bidig ite for ubtrate moleule. Example: Hemoglobi Hb, the oxygearryig

More information

BINARY TREE-STRUCTURED PARTITION AND CLASSIFICATION SCHEMES

BINARY TREE-STRUCTURED PARTITION AND CLASSIFICATION SCHEMES BINARY TREE-STRUCTURED PARTITION AND CLASSIFICATION SCHEMES DAVID MCDIARMID Abstract Binary tree-structured partition and classification schemes are a class of nonparametric tree-based approaches to classification

More information

Biplots. Some ecological data to illustrate regression biplots

Biplots. Some ecological data to illustrate regression biplots Biplot The biplot eten the iea of a imple atterplot of two variable to the ae of man variable, with the obetive of viualizing a maimum amount of information in the ata a poible. We firt look at how a regreion

More information

ON THE APPROXIMATION ERROR IN HIGH DIMENSIONAL MODEL REPRESENTATION. Xiaoqun Wang

ON THE APPROXIMATION ERROR IN HIGH DIMENSIONAL MODEL REPRESENTATION. Xiaoqun Wang Proceeding of the 2008 Winter Simulation Conference S. J. Maon, R. R. Hill, L. Mönch, O. Roe, T. Jefferon, J. W. Fowler ed. ON THE APPROXIMATION ERROR IN HIGH DIMENSIONAL MODEL REPRESENTATION Xiaoqun Wang

More information

Chapter ML:III. III. Decision Trees. Decision Trees Basics Impurity Functions Decision Tree Algorithms Decision Tree Pruning

Chapter ML:III. III. Decision Trees. Decision Trees Basics Impurity Functions Decision Tree Algorithms Decision Tree Pruning Chapter ML:III III. Decision Trees Decision Trees Basics Impurity Functions Decision Tree Algorithms Decision Tree Pruning ML:III-34 Decision Trees STEIN/LETTMANN 2005-2017 Splitting Let t be a leaf node

More information

Modeling Highway Traffic Volumes

Modeling Highway Traffic Volumes Modeling Highway Traffic Volumes Tomáš Šingliar1 and Miloš Hauskrecht 1 Computer Science Dept, Uniersity of Pittsburgh, Pittsburgh, PA 15260 {tomas, milos}@cs.pitt.edu Abstract. Most traffic management

More information

66 Lecture 3 Random Search Tree i unique. Lemma 3. Let X and Y be totally ordered et, and let be a function aigning a ditinct riority in Y to each ele

66 Lecture 3 Random Search Tree i unique. Lemma 3. Let X and Y be totally ordered et, and let be a function aigning a ditinct riority in Y to each ele Lecture 3 Random Search Tree In thi lecture we will decribe a very imle robabilitic data tructure that allow inert, delete, and memberhi tet (among other oeration) in exected logarithmic time. Thee reult

More information

The Effectiveness of the Linear Hull Effect

The Effectiveness of the Linear Hull Effect The Effetiveness of the Linear Hull Effet S. Murphy Tehnial Report RHUL MA 009 9 6 Otober 009 Department of Mathematis Royal Holloway, University of London Egham, Surrey TW0 0EX, England http://www.rhul.a.uk/mathematis/tehreports

More information

Model-based mixture discriminant analysis an experimental study

Model-based mixture discriminant analysis an experimental study Model-based mixture disriminant analysis an experimental study Zohar Halbe and Mayer Aladjem Department of Eletrial and Computer Engineering, Ben-Gurion University of the Negev P.O.Box 653, Beer-Sheva,

More information

Reinforcement Learning

Reinforcement Learning Reinforcement Learning Yihay Manour Google Inc. & Tel-Aviv Univerity Outline Goal of Reinforcement Learning Mathematical Model (MDP) Planning Learning Current Reearch iue 2 Goal of Reinforcement Learning

More information

Topics to be covered. Topology defined by TX range. Power/Topology Control. Power/Topology Control. Changes Topology

Topics to be covered. Topology defined by TX range. Power/Topology Control. Power/Topology Control. Changes Topology Power/Topology Control Topi to be overe Power/Topology Control problem. Importane of power ontrol Minimum-energy Path-preerving Graph Algorithm Performane of the algorithm Diuion/quetion/omment Wire network

More information

Communicating Special Relativity Theory s Mathematical Inconsistencies

Communicating Special Relativity Theory s Mathematical Inconsistencies Communiating Speial Relatiity Theory s Mathematial Inonsistenies Steen B Bryant Primitie Logi, In, 704 Sansome Street, San Franiso, California 94111 Stee.Bryant@RelatiityChallenge.Com Einstein s Speial

More information

Factor Analysis with Poisson Output

Factor Analysis with Poisson Output Factor Analyi with Poion Output Gopal Santhanam Byron Yu Krihna V. Shenoy, Department of Electrical Engineering, Neurocience Program Stanford Univerity Stanford, CA 94305, USA {gopal,byronyu,henoy}@tanford.edu

More information

S Mobile Communications Services and Systems

S Mobile Communications Services and Systems S-7.60 Mobile ommuniation Serie and Sytem Tutorial, Noember 9, 004. One a pyiit obert Wood did not top i ar beind te red trai ligt. He exue imel by uing oppler eet. Beaue o oppler it te red ligt ad turned

More information

Highway Capacity Manual 2010

Highway Capacity Manual 2010 RR = minimum number of lane change that mut be made by one ramp-toramp ehicle to execute the deired maneuer uccefully. MIN for two-ided weaing egment i gien by Equation 12-3: MIN RR For two-ided weaing

More information

Decision Tree Learning

Decision Tree Learning Decision Tree Learning Goals for the lecture you should understand the following concepts the decision tree representation the standard top-down approach to learning a tree Occam s razor entropy and information

More information

Decision Tree Learning

Decision Tree Learning Decision Tree Learning Goals for the lecture you should understand the following concepts the decision tree representation the standard top-down approach to learning a tree Occam s razor entropy and information

More information

Hands-on Tutorial on Optimization F. Eberle, R. Hoeksma, and N. Megow September 26, Branch & Bound

Hands-on Tutorial on Optimization F. Eberle, R. Hoeksma, and N. Megow September 26, Branch & Bound Hands-on Tutorial on Optimization F. Eberle, R. Hoeksma, and N. Megow September 6, 8 Branh & Bound Branh & Bound: A General Framework for ILPs Introdued in the 96 s by Land and Doig Based on two priniple

More information

SUPPLEMENTARY MATERIAL. Authors: Alan A. Stocker (1) and Eero P. Simoncelli (2)

SUPPLEMENTARY MATERIAL. Authors: Alan A. Stocker (1) and Eero P. Simoncelli (2) SUPPLEMENTARY MATERIAL Authors: Alan A. Stocker () and Eero P. Simoncelli () Affiliations: () Dept. of Psychology, Uniersity of Pennsylania 34 Walnut Street 33C Philadelphia, PA 94-68 U.S.A. () Howard

More information

Multiple Sequence Alignment. Progressive Alignment Iterative Pairwise Guide Tree ClustalW Co-linearity Multiple Sequence Alignment Editors

Multiple Sequence Alignment. Progressive Alignment Iterative Pairwise Guide Tree ClustalW Co-linearity Multiple Sequence Alignment Editors Yverdon Le Bain Introduction to Bioinformatic Introduction to Bioinformatic ami Khuri Department of Computer cience an Joé tate Univerity an Joé, California, UA khuri@c.ju.edu www.c.ju.edu/faculty/khuri

More information

Network Flow. Data Structures and Algorithms Andrei Bulatov

Network Flow. Data Structures and Algorithms Andrei Bulatov Nework Flow Daa Srucure and Algorihm Andrei Bulao Algorihm Nework Flow 24-2 Flow Nework Think of a graph a yem of pipe We ue hi yem o pump waer from he ource o ink Eery pipe/edge ha limied capaciy Flow

More information

Knowledge Discovery and Data Mining

Knowledge Discovery and Data Mining Knowledge Discovery and Data Mining Lecture 06 - Regression & Decision Trees Tom Kelsey School of Computer Science University of St Andrews http://tom.home.cs.st-andrews.ac.uk twk@st-andrews.ac.uk Tom

More information

Insights into Cross-validation

Insights into Cross-validation Noname manuscript No. (will be inserted by the editor) Insights into Cross-alidation Amit Dhurandhar Alin Dobra Receied: date / Accepted: date Abstract Cross-alidation is one of the most widely used techniques,

More information