# What is Candidate Sampling

Save this PDF as:

Size: px
Start display at page:

Download "What is Candidate Sampling"

## Transcription

1 What s Canddate Samplng Say we have a multclass or mult label problem where each tranng example ( x, T ) conssts of a context x a small (mult)set of target classes T out of a large unverse L of possble classes. For example, the problem mght be to predctng the next word (or the set of future words) n a sentence gven the prevous words. We wsh to learn a compatblty functon F (x, y ) whch says somethng about the compatblty of a class y wth a context x. For example the probablty of the class gven the context. Exhaustve tranng methods such as softmax and logstc regresson requre us to compute F (x, y ) for every class y L for every tranng example. When L s very large, ths can be prohbtvely expensve. Canddate Samplng tranng methods nvolve constructng a tranng task n whch for each tranng example ( x, T ), we only need to evaluate F (x, y ) for a small set of canddate classes C L. Typcally, the set of canddates C s the unon of the target classes wth a randomly chosen sample of (other) classes S L. C = T S The random choce of S may or may not depend on x and/or T. The tranng algorthm takes the form of a neural network, where the layer representng (x, ) F y s traned by back propagaton from a loss functon.

2 Table of Canddate Samplng Algorthms Postve tranng classes assocated wth tranng example ( x, T ) : Negatve tranng classes assocated wth tranng example ( x, T ) : Input to Tranng Loss G(x, y ) = Tranng Loss F (x, y ) gets traned to approxmate: P OS = NEG = Nose Contrastve Estmaton (NCE) T S F (x, y) l og(q(y x)) Logstc l og(p (y x)) Negatve Samplng T S F (x, y ) Logstc l og P ( Q(y x) (y x)) Sampled Logstc T ( S T ) F (x, y) l og(q(y x)) Logstc l ogodds(y x) = l og ( P 1 P (y x)) Full Logstc T ( L T ) F (x, y ) Logstc l og(odds(y x)) = l og ( P 1 P (y x)) Full Softmax T = { t } ( L T ) F (x, y ) Softmax l og(p (y x)) + K (x) Sampled Softmax T = { t } ( S T ) F (x, y) l og(q(y x)) Softmax l og(p (y x)) + K (x) Q (y x) s defned as the probablty (or expected count) accordng to the samplng algorthm of the class y n the (mult )set of sampled classes gven the context x. K (x) s an arbtrary functon that does not depend on the canddate class. Snce Softmax nvolves a normalzaton, addton of such a functon does not affect the computed probabltes. ( ) l ogstc tranng loss = l og(1 + exp( G(x, y )) + l og(1 + exp(g(x, y )) y P OS y NEG ( ( )) s oftmax tranng loss = G (x, t ) + log exp(g(x, y )) y P OS NEG NCE and Negatve Samplng generalze to the case where T s a multset. In ths case, P (y x) denotes the expected count of y n T. Smlarly, NCE, Negatve Samplng, and Sampled Logstc generalze to the case where S s a multset. In ths case Q(y x) denotes the expected count of y n S.

3 Sampled Softmax (A faster way to tran a softmax classfer) Reference: Assume that we have a sngle label problem. Each tranng example ( x, {t }) conssts of a context and one target class. We wrte P (y x) for the probablty of that the one target class s y gven that the context s x. We would lke to tran a functon F (x, y ) to produce softmax logts that s, relatve log probabltes of the class gven the context: F (x, y) log(p (y x)) + K (x) Where K(x) s an arbtrary functon that does not depend on y. In full softmax tranng, for every tranng example ( x, {t }), we would need to compute logts F (x, y ) for all classes n y L. Ths can get expensve f the unverse of classes L s very large. In Sampled Softmax, for each tranng example ( x, { t }), we pck a small set S L of sampled classes accordng to a chosen samplng functon Q (y x). Each class y L s ncluded n S ndependently wth probablty Q(y x ). P (S = S x ) = Q(y x ) (1 Q(y x )) y S y (L S) We create a set of canddates classes: C contanng the unon of the target class and the sampled t } C = S { Our tranng task s to fgure out, gven ths set C, whch of the classes n C s the target class. For each class y C, we want to compute the posteror probablty that y s the target class gven our knowledge of x and C. We call ths P (t = y x, C ) Applyng Bayes rule: (t x, ) (t, x ) / P (C x ) P = y C = P = y C (t x ) P (C t, ) / P (C x ) = P = y = y x P (y x ) P (C t, x ) / P (C x ) = = y

4 Now to compute P (C t = y, x ), we note that n order for ths to happen, S may or may not contan y, must contan all other elements of C, and must not contan any classes not n C. So: P (t = y x, C ) = P (y x ) Q (y x ) (1 Q (y x )) / P (C x ) P (y x = ) Q(y x ) y C P (y x = ) Q(y x ) C y C {y} y (L C ) Q (y x ) (1 Q (y x )) / P (C x ) / K(x, ) y (L C ) where K(x, C ) s a functon that does not depend on y. So: log(p (t = y x, C )) = log(p (y x )) log(q(y x )) + K (x, C ) These are the relatve logts that should feed nto a softmax classfer predctng whch of the canddates n s the true one. C Snce we are tryng to tran the functon F (x, y) to approxmate l og(p (y x)), we take the layer n our network representng F (x, y), subtract log(q(y x)), and pass the result to a softmax classfer predctng whch canddate s the true one. T ranng Sof tmax Input = F (x, y) l og(q(y x) Backpropagatng the gradents from that classfer trans F to gve us what we want.

5 Nose Contrastve Estmaton (NCE) Reference: Each tranng example ( x, T ) conssts of a context and a small multset of target classes. In practce, T x may always be a set or even a sngle class, but we use a multset here for generalty. We use the followng as a shorthand for the expected count of a class n the set of target classes for a context. In the case of sets wth no duplcates, ths s the probablty of the class gven the context: P (y x) : = E(T (y) x) We would lke to tran a functon F (x, y ) to approxmate the log expected count of the class gven the context, or n the case of a sets, the log probablty of the class gven the context. F (x, y) log (P (y x)) For each example ( x, T ), we pck a multset of sampled classes S. In practce, t probably makes sense to pck a set, but we use a multset here for generalty. Our samplng algorthm may or may not depend on x but may not depend on T. We construct a multset of canddates consstng of the sum of the target classes and the sampled classes. C = T + S Our tranng task s to dstngush the true canddates from the sampled canddates. We have one postve tranng meta example for each element of and one negatve tranng meta example for each element of S. We ntroduce the shorthand Q (y x) to denote the expected count, accordng to our samplng algorthm, of a partcular class n the set of sampled classes. If S never contans duplcates, then ths s a probablty. P l ogodds(y came from T vs S x) = l og ( Q(y x) (y x)) = l og (P (y x)) l og(q(y x)) T Q (y x) : = E (S(y) x)) The frst term, l og (P (y x)), s what we would lke to tran F (x, y ) to estmate.

6 We have a layer n our model whch represents F (x, y ). We add to t the second term, l og(q(y x)), whch we compute analytcally, and we pass the result to a logstc regresson loss whose label ndcates whether y came from T as opposed to S. L ogstc Regresson Input = F (x, y) log(q(y x)) The backpropagaton sgnal trans F (x, y ) to approxmate what we want t to.

7 Negatve Samplng Reference: dstrbuted representatons of words and phrases and ther co mpostonalty.pdf Negatve samplng s a smplfed varant of Nose Contrastve Estmaton where we neglect to subtract off l og(q(y x)) durng tranng. As a result, F (x, y ) s traned to approxmate l og (E(y x)) l og(q(y x)). It s noteworthy that n Negatve Samplng, we are optmzng F (x, y ) to approxmate somethng that depends on the samplng dstrbuton Q. Ths wll make the results hghly dependent on the choce of samplng dstrbuton. Ths s not true for the other algorthms descrbed here.

8 Sampled Logstc Sampled Logstc s a varant on Nose Contrastve Estmaton where we dscard wthout replacement all sampled classes that happen to also be target classes. Ths requres a set, as opposed to a multset, though S T to be may be a multset. As a result we learn an estmator of the log odds of a class as opposed to the log probablty of a class. The math changes from the NCE math as follows: P (y x) l ogodds(y came from T vs (S T ) x) = l og( Q(y x)(1 P (y x)) = l og( P 1 P (y x)) l og(q(y x)) ( P (y x) The frst term, l og 1 P (y x)), s what we would lke to tran F (x, y ) to estmate. We have a layer n our model, whch represents F (x, y ). We add to t the second term, l og(q(y x)), whch we compute analytcally, and we pass the result to a logstc regresson loss predctng whether y came from T vs ( S T ). The backpropagaton sgnal trans the L ogstc Regresson Input = F (x, y) log(q(y x) F (x, y ) layer to approxmate what we want t to. F (x, y) log ( P (y x) 1 P (y x))

9 Context Specfc vs. Generc Samplng In the methods dscussed, the samplng algorthm s allowed to depend on the context. It s possble that for some models, context specfc samplng wll be very useful, n that we can generate context dependent hard negatves and provde a more useful tranng sgnal. The authors have to ths pont focused on generc samplng algorthms such as unform samplng and ungram samplng, whch do not make use of the context. The reason s descrbed n the next secton. Batchwse Samplng We have focused on models whch use the same set S of sampled classes across a whole batch of tranng examples. Ths seems counterntutve shouldn t convergence be faster f we use dfferent sampled classes for each tranng example? The reason for usng the same sampled classes across a batch s computatonal. In many of our models, F (x, y ) s computed as the dot product of a feature vector for the context (the top hdden layer of a neural network), and an embeddng vector for the class. Computng the dot products of many feature vectors wth many embeddng vectors s a matrx multplcaton, whch s hghly effcent on modern hardware, especally on GPUs. Batchng lke ths often allows us to use hundreds or thousands of sampled classes wthout notceable slowdown. Another way to see t s that the overhead of fetchng a class embeddng across devces s greater than the tme t takes to compute ts dot products wth hundreds or even thousands of feature vectors. So f we are gong to use a sampled class wth one context, t s vrtually free to use t wth all of the other contexts n the batch as well.

### Forecasting the Direction and Strength of Stock Market Movement

Forecastng the Drecton and Strength of Stock Market Movement Jngwe Chen Mng Chen Nan Ye cjngwe@stanford.edu mchen5@stanford.edu nanye@stanford.edu Abstract - Stock market s one of the most complcated systems

More information

### Logistic Regression. Lecture 4: More classifiers and classes. Logistic regression. Adaboost. Optimization. Multiple class classification

Lecture 4: More classfers and classes C4B Machne Learnng Hlary 20 A. Zsserman Logstc regresson Loss functons revsted Adaboost Loss functons revsted Optmzaton Multple class classfcaton Logstc Regresson

More information

### benefit is 2, paid if the policyholder dies within the year, and probability of death within the year is ).

REVIEW OF RISK MANAGEMENT CONCEPTS LOSS DISTRIBUTIONS AND INSURANCE Loss and nsurance: When someone s subject to the rsk of ncurrng a fnancal loss, the loss s generally modeled usng a random varable or

More information

### Nonlinear data mapping by neural networks

Nonlnear data mappng by neural networks R.P.W. Dun Delft Unversty of Technology, Netherlands Abstract A revew s gven of the use of neural networks for nonlnear mappng of hgh dmensonal data on lower dmensonal

More information

### 1 Example 1: Axis-aligned rectangles

COS 511: Theoretcal Machne Learnng Lecturer: Rob Schapre Lecture # 6 Scrbe: Aaron Schld February 21, 2013 Last class, we dscussed an analogue for Occam s Razor for nfnte hypothess spaces that, n conjuncton

More information

### CS 2750 Machine Learning. Lecture 17a. Clustering. CS 2750 Machine Learning. Clustering

Lecture 7a Clusterng Mlos Hauskrecht mlos@cs.ptt.edu 539 Sennott Square Clusterng Groups together smlar nstances n the data sample Basc clusterng problem: dstrbute data nto k dfferent groups such that

More information

### Recurrence. 1 Definitions and main statements

Recurrence 1 Defntons and man statements Let X n, n = 0, 1, 2,... be a MC wth the state space S = (1, 2,...), transton probabltes p j = P {X n+1 = j X n = }, and the transton matrx P = (p j ),j S def.

More information

### v a 1 b 1 i, a 2 b 2 i,..., a n b n i.

SECTION 8.4 COMPLEX VECTOR SPACES AND INNER PRODUCTS 455 8.4 COMPLEX VECTOR SPACES AND INNER PRODUCTS All the vector spaces we have studed thus far n the text are real vector spaces snce the scalars are

More information

### Luby s Alg. for Maximal Independent Sets using Pairwise Independence

Lecture Notes for Randomzed Algorthms Luby s Alg. for Maxmal Independent Sets usng Parwse Independence Last Updated by Erc Vgoda on February, 006 8. Maxmal Independent Sets For a graph G = (V, E), an ndependent

More information

### An Interest-Oriented Network Evolution Mechanism for Online Communities

An Interest-Orented Network Evoluton Mechansm for Onlne Communtes Cahong Sun and Xaopng Yang School of Informaton, Renmn Unversty of Chna, Bejng 100872, P.R. Chna {chsun,yang}@ruc.edu.cn Abstract. Onlne

More information

### An Alternative Way to Measure Private Equity Performance

An Alternatve Way to Measure Prvate Equty Performance Peter Todd Parlux Investment Technology LLC Summary Internal Rate of Return (IRR) s probably the most common way to measure the performance of prvate

More information

### 1. Fundamentals of probability theory 2. Emergence of communication traffic 3. Stochastic & Markovian Processes (SP & MP)

6.3 / -- Communcaton Networks II (Görg) SS20 -- www.comnets.un-bremen.de Communcaton Networks II Contents. Fundamentals of probablty theory 2. Emergence of communcaton traffc 3. Stochastc & Markovan Processes

More information

### PSYCHOLOGICAL RESEARCH (PYC 304-C) Lecture 12

14 The Ch-squared dstrbuton PSYCHOLOGICAL RESEARCH (PYC 304-C) Lecture 1 If a normal varable X, havng mean µ and varance σ, s standardsed, the new varable Z has a mean 0 and varance 1. When ths standardsed

More information

### Statistical Methods to Develop Rating Models

Statstcal Methods to Develop Ratng Models [Evelyn Hayden and Danel Porath, Österrechsche Natonalbank and Unversty of Appled Scences at Manz] Source: The Basel II Rsk Parameters Estmaton, Valdaton, and

More information

### Implementation of Deutsch's Algorithm Using Mathcad

Implementaton of Deutsch's Algorthm Usng Mathcad Frank Roux The followng s a Mathcad mplementaton of Davd Deutsch's quantum computer prototype as presented on pages - n "Machnes, Logc and Quantum Physcs"

More information

### Logistic Regression. Steve Kroon

Logstc Regresson Steve Kroon Course notes sectons: 24.3-24.4 Dsclamer: these notes do not explctly ndcate whether values are vectors or scalars, but expects the reader to dscern ths from the context. Scenaro

More information

### The OC Curve of Attribute Acceptance Plans

The OC Curve of Attrbute Acceptance Plans The Operatng Characterstc (OC) curve descrbes the probablty of acceptng a lot as a functon of the lot s qualty. Fgure 1 shows a typcal OC Curve. 10 8 6 4 1 3 4

More information

### Formula of Total Probability, Bayes Rule, and Applications

1 Formula of Total Probablty, Bayes Rule, and Applcatons Recall that for any event A, the par of events A and A has an ntersecton that s empty, whereas the unon A A represents the total populaton of nterest.

More information

### Probabilistic Linear Classifier: Logistic Regression. CS534-Machine Learning

robablstc Lnear Classfer: Logstc Regresson CS534-Machne Learnng Three Man Approaches to learnng a Classfer Learn a classfer: a functon f, ŷ f Learn a probablstc dscrmnatve model,.e., the condtonal dstrbuton

More information

### ) of the Cell class is created containing information about events associated with the cell. Events are added to the Cell instance

Calbraton Method Instances of the Cell class (one nstance for each FMS cell) contan ADC raw data and methods assocated wth each partcular FMS cell. The calbraton method ncludes event selecton (Class Cell

More information

### 2.4 Bivariate distributions

page 28 2.4 Bvarate dstrbutons 2.4.1 Defntons Let X and Y be dscrete r.v.s defned on the same probablty space (S, F, P). Instead of treatng them separately, t s often necessary to thnk of them actng together

More information

### MAPP. MERIS level 3 cloud and water vapour products. Issue: 1. Revision: 0. Date: 9.12.1998. Function Name Organisation Signature Date

Ttel: Project: Doc. No.: MERIS level 3 cloud and water vapour products MAPP MAPP-ATBD-ClWVL3 Issue: 1 Revson: 0 Date: 9.12.1998 Functon Name Organsaton Sgnature Date Author: Bennartz FUB Preusker FUB Schüller

More information

### Exhaustive Regression. An Exploration of Regression-Based Data Mining Techniques Using Super Computation

Exhaustve Regresson An Exploraton of Regresson-Based Data Mnng Technques Usng Super Computaton Antony Daves, Ph.D. Assocate Professor of Economcs Duquesne Unversty Pttsburgh, PA 58 Research Fellow The

More information

### Single and multiple stage classifiers implementing logistic discrimination

Sngle and multple stage classfers mplementng logstc dscrmnaton Hélo Radke Bttencourt 1 Dens Alter de Olvera Moraes 2 Vctor Haertel 2 1 Pontfíca Unversdade Católca do Ro Grande do Sul - PUCRS Av. Ipranga,

More information

### Analysis of Energy-Conserving Access Protocols for Wireless Identification Networks

From the Proceedngs of Internatonal Conference on Telecommuncaton Systems (ITC-97), March 2-23, 1997. 1 Analyss of Energy-Conservng Access Protocols for Wreless Identfcaton etworks Imrch Chlamtac a, Chara

More information

### + + + - - This circuit than can be reduced to a planar circuit

MeshCurrent Method The meshcurrent s analog of the nodeoltage method. We sole for a new set of arables, mesh currents, that automatcally satsfy KCLs. As such, meshcurrent method reduces crcut soluton to

More information

### Forecasting the Demand of Emergency Supplies: Based on the CBR Theory and BP Neural Network

700 Proceedngs of the 8th Internatonal Conference on Innovaton & Management Forecastng the Demand of Emergency Supples: Based on the CBR Theory and BP Neural Network Fu Deqang, Lu Yun, L Changbng School

More information

### How Sets of Coherent Probabilities May Serve as Models for Degrees of Incoherence

1 st Internatonal Symposum on Imprecse Probabltes and Ther Applcatons, Ghent, Belgum, 29 June 2 July 1999 How Sets of Coherent Probabltes May Serve as Models for Degrees of Incoherence Mar J. Schervsh

More information

### Vision Mouse. Saurabh Sarkar a* University of Cincinnati, Cincinnati, USA ABSTRACT 1. INTRODUCTION

Vson Mouse Saurabh Sarkar a* a Unversty of Cncnnat, Cncnnat, USA ABSTRACT The report dscusses a vson based approach towards trackng of eyes and fngers. The report descrbes the process of locatng the possble

More information

### Efficient Project Portfolio as a tool for Enterprise Risk Management

Effcent Proect Portfolo as a tool for Enterprse Rsk Management Valentn O. Nkonov Ural State Techncal Unversty Growth Traectory Consultng Company January 5, 27 Effcent Proect Portfolo as a tool for Enterprse

More information

### Module 2 LOSSLESS IMAGE COMPRESSION SYSTEMS. Version 2 ECE IIT, Kharagpur

Module LOSSLESS IMAGE COMPRESSION SYSTEMS Lesson 3 Lossless Compresson: Huffman Codng Instructonal Objectves At the end of ths lesson, the students should be able to:. Defne and measure source entropy..

More information

### NPAR TESTS. One-Sample Chi-Square Test. Cell Specification. Observed Frequencies 1O i 6. Expected Frequencies 1EXP i 6

PAR TESTS If a WEIGHT varable s specfed, t s used to replcate a case as many tmes as ndcated by the weght value rounded to the nearest nteger. If the workspace requrements are exceeded and samplng has

More information

### Traffic State Estimation in the Traffic Management Center of Berlin

Traffc State Estmaton n the Traffc Management Center of Berln Authors: Peter Vortsch, PTV AG, Stumpfstrasse, D-763 Karlsruhe, Germany phone ++49/72/965/35, emal peter.vortsch@ptv.de Peter Möhl, PTV AG,

More information

### Probabilities and Probabilistic Models

Probabltes and Probablstc Models Probablstc models A model means a system that smulates an obect under consderaton. A probablstc model s a model that produces dfferent outcomes wth dfferent probabltes

More information

### Causal, Explanatory Forecasting. Analysis. Regression Analysis. Simple Linear Regression. Which is Independent? Forecasting

Causal, Explanatory Forecastng Assumes cause-and-effect relatonshp between system nputs and ts output Forecastng wth Regresson Analyss Rchard S. Barr Inputs System Cause + Effect Relatonshp The job of

More information

### Quantization Effects in Digital Filters

Quantzaton Effects n Dgtal Flters Dstrbuton of Truncaton Errors In two's complement representaton an exact number would have nfntely many bts (n general). When we lmt the number of bts to some fnte value

More information

### CHOLESTEROL REFERENCE METHOD LABORATORY NETWORK. Sample Stability Protocol

CHOLESTEROL REFERENCE METHOD LABORATORY NETWORK Sample Stablty Protocol Background The Cholesterol Reference Method Laboratory Network (CRMLN) developed certfcaton protocols for total cholesterol, HDL

More information

### Lecture 5,6 Linear Methods for Classification. Summary

Lecture 5,6 Lnear Methods for Classfcaton Rce ELEC 697 Farnaz Koushanfar Fall 2006 Summary Bayes Classfers Lnear Classfers Lnear regresson of an ndcator matrx Lnear dscrmnant analyss (LDA) Logstc regresson

More information

### 1. Measuring association using correlation and regression

How to measure assocaton I: Correlaton. 1. Measurng assocaton usng correlaton and regresson We often would lke to know how one varable, such as a mother's weght, s related to another varable, such as a

More information

### Extending Probabilistic Dynamic Epistemic Logic

Extendng Probablstc Dynamc Epstemc Logc Joshua Sack May 29, 2008 Probablty Space Defnton A probablty space s a tuple (S, A, µ), where 1 S s a set called the sample space. 2 A P(S) s a σ-algebra: a set

More information

### greatest common divisor

4. GCD 1 The greatest common dvsor of two ntegers a and b (not both zero) s the largest nteger whch s a common factor of both a and b. We denote ths number by gcd(a, b), or smply (a, b) when there s no

More information

### Can Auto Liability Insurance Purchases Signal Risk Attitude?

Internatonal Journal of Busness and Economcs, 2011, Vol. 10, No. 2, 159-164 Can Auto Lablty Insurance Purchases Sgnal Rsk Atttude? Chu-Shu L Department of Internatonal Busness, Asa Unversty, Tawan Sheng-Chang

More information

### 8.5 UNITARY AND HERMITIAN MATRICES. The conjugate transpose of a complex matrix A, denoted by A*, is given by

6 CHAPTER 8 COMPLEX VECTOR SPACES 5. Fnd the kernel of the lnear transformaton gven n Exercse 5. In Exercses 55 and 56, fnd the mage of v, for the ndcated composton, where and are gven by the followng

More information

### DEFINING %COMPLETE IN MICROSOFT PROJECT

CelersSystems DEFINING %COMPLETE IN MICROSOFT PROJECT PREPARED BY James E Aksel, PMP, PMI-SP, MVP For Addtonal Informaton about Earned Value Management Systems and reportng, please contact: CelersSystems,

More information

### CHAPTER 14 MORE ABOUT REGRESSION

CHAPTER 14 MORE ABOUT REGRESSION We learned n Chapter 5 that often a straght lne descrbes the pattern of a relatonshp between two quanttatve varables. For nstance, n Example 5.1 we explored the relatonshp

More information

### Communication Networks II Contents

8 / 1 -- Communcaton Networs II (Görg) -- www.comnets.un-bremen.de Communcaton Networs II Contents 1 Fundamentals of probablty theory 2 Traffc n communcaton networs 3 Stochastc & Marovan Processes (SP

More information

### Detecting Credit Card Fraud using Periodic Features

Detectng Credt Card Fraud usng Perodc Features Alejandro Correa Bahnsen, Djamla Aouada, Aleksandar Stojanovc and Björn Ottersten Interdscplnary Centre for Securty, Relablty and Trust Unversty of Luxembourg,

More information

### Learning from Large Distributed Data: A Scaling Down Sampling Scheme for Efficient Data Processing

Internatonal Journal of Machne Learnng and Computng, Vol. 4, No. 3, June 04 Learnng from Large Dstrbuted Data: A Scalng Down Samplng Scheme for Effcent Data Processng Che Ngufor and Janusz Wojtusak part

More information

### Distributed Column Subset Selection on MapReduce

Dstrbuted Column Subset Selecton on MapReduce Ahmed K. arahat Ahmed Elgohary Al Ghods Mohamed S. Kamel Unversty of Waterloo Waterloo, Ontaro, Canada N2L 3G1 Emal: {afarahat, aelgohary, aghodsb, mkamel}@uwaterloo.ca

More information

### Loop Parallelization

- - Loop Parallelzaton C-52 Complaton steps: nested loops operatng on arrays, sequentell executon of teraton space DECLARE B[..,..+] FOR I :=.. FOR J :=.. I B[I,J] := B[I-,J]+B[I-,J-] ED FOR ED FOR analyze

More information

### Simple Interest Loans (Section 5.1) :

Chapter 5 Fnance The frst part of ths revew wll explan the dfferent nterest and nvestment equatons you learned n secton 5.1 through 5.4 of your textbook and go through several examples. The second part

More information

### Traffic-light a stress test for life insurance provisions

MEMORANDUM Date 006-09-7 Authors Bengt von Bahr, Göran Ronge Traffc-lght a stress test for lfe nsurance provsons Fnansnspetonen P.O. Box 6750 SE-113 85 Stocholm [Sveavägen 167] Tel +46 8 787 80 00 Fax

More information

### Enterprise Master Patient Index

Enterprse Master Patent Index Healthcare data are captured n many dfferent settngs such as hosptals, clncs, labs, and physcan offces. Accordng to a report by the CDC, patents n the Unted States made an

More information

### All Roads Lead to Rome: Optimistic Recovery for Distributed Iterative Data Processing

All Roads Lead to Rome: Optmstc Recovery for Dstrbuted Iteratve Data Processng Sebastan Schelter Stephan Ewen Kostas Tzoumas Volker Markl Technsche Unverstät Berln, Germany frstname.lastname@tu-berln.de

More information

### A Probabilistic Theory of Coherence

A Probablstc Theory of Coherence BRANDEN FITELSON. The Coherence Measure C Let E be a set of n propostons E,..., E n. We seek a probablstc measure C(E) of the degree of coherence of E. Intutvely, we want

More information

### New bounds in Balog-Szemerédi-Gowers theorem

New bounds n Balog-Szemeréd-Gowers theorem By Tomasz Schoen Abstract We prove, n partcular, that every fnte subset A of an abelan group wth the addtve energy κ A 3 contans a set A such that A κ A and A

More information

### The covariance is the two variable analog to the variance. The formula for the covariance between two variables is

Regresson Lectures So far we have talked only about statstcs that descrbe one varable. What we are gong to be dscussng for much of the remander of the course s relatonshps between two or more varables.

More information

### We assume your students are learning about self-regulation (how to change how alert they feel) through the Alert Program with its three stages:

Welcome to ALERT BINGO, a fun-flled and educatonal way to learn the fve ways to change engnes levels (Put somethng n your Mouth, Move, Touch, Look, and Lsten) as descrbed n the How Does Your Engne Run?

More information

### ErrorPropagation.nb 1. Error Propagation

ErrorPropagaton.nb Error Propagaton Suppose that we make observatons of a quantty x that s subject to random fluctuatons or measurement errors. Our best estmate of the true value for ths quantty s then

More information

### Realistic Image Synthesis

Realstc Image Synthess - Combned Samplng and Path Tracng - Phlpp Slusallek Karol Myszkowsk Vncent Pegoraro Overvew: Today Combned Samplng (Multple Importance Samplng) Renderng and Measurng Equaton Random

More information

### Prediction of Disability Frequencies in Life Insurance

Predcton of Dsablty Frequences n Lfe Insurance Bernhard Köng Fran Weber Maro V. Wüthrch October 28, 2011 Abstract For the predcton of dsablty frequences, not only the observed, but also the ncurred but

More information

### The Magnetic Field. Concepts and Principles. Moving Charges. Permanent Magnets

. The Magnetc Feld Concepts and Prncples Movng Charges All charged partcles create electrc felds, and these felds can be detected by other charged partcles resultng n electrc force. However, a completely

More information

### The Greedy Method. Introduction. 0/1 Knapsack Problem

The Greedy Method Introducton We have completed data structures. We now are gong to look at algorthm desgn methods. Often we are lookng at optmzaton problems whose performance s exponental. For an optmzaton

More information

### Prediction of Disability Frequencies in Life Insurance

1 Predcton of Dsablty Frequences n Lfe Insurance Bernhard Köng 1, Fran Weber 1, Maro V. Wüthrch 2 Abstract: For the predcton of dsablty frequences, not only the observed, but also the ncurred but not yet

More information

### THE DISTRIBUTION OF LOAN PORTFOLIO VALUE * Oldrich Alfons Vasicek

HE DISRIBUION OF LOAN PORFOLIO VALUE * Oldrch Alfons Vascek he amount of captal necessary to support a portfolo of debt securtes depends on the probablty dstrbuton of the portfolo loss. Consder a portfolo

More information

### Inequality and The Accounting Period. Quentin Wodon and Shlomo Yitzhaki. World Bank and Hebrew University. September 2001.

Inequalty and The Accountng Perod Quentn Wodon and Shlomo Ytzha World Ban and Hebrew Unversty September Abstract Income nequalty typcally declnes wth the length of tme taen nto account for measurement.

More information

### Multiplication Algorithms for Radix-2 RN-Codings and Two s Complement Numbers

Multplcaton Algorthms for Radx- RN-Codngs and Two s Complement Numbers Jean-Luc Beuchat Projet Arénare, LIP, ENS Lyon 46, Allée d Itale F 69364 Lyon Cedex 07 jean-luc.beuchat@ens-lyon.fr Jean-Mchel Muller

More information

### x f(x) 1 0.25 1 0.75 x 1 0 1 1 0.04 0.01 0.20 1 0.12 0.03 0.60

BIVARIATE DISTRIBUTIONS Let be a varable that assumes the values { 1,,..., n }. Then, a functon that epresses the relatve frequenc of these values s called a unvarate frequenc functon. It must be true

More information

### On the Optimal Control of a Cascade of Hydro-Electric Power Stations

On the Optmal Control of a Cascade of Hydro-Electrc Power Statons M.C.M. Guedes a, A.F. Rbero a, G.V. Smrnov b and S. Vlela c a Department of Mathematcs, School of Scences, Unversty of Porto, Portugal;

More information

### Calculating the high frequency transmission line parameters of power cables

< ' Calculatng the hgh frequency transmsson lne parameters of power cables Authors: Dr. John Dcknson, Laboratory Servces Manager, N 0 RW E B Communcatons Mr. Peter J. Ncholson, Project Assgnment Manager,

More information

### CS 2750 Machine Learning. Lecture 3. Density estimation. CS 2750 Machine Learning. Announcements

Lecture 3 Densty estmaton Mlos Hauskrecht mlos@cs.ptt.edu 5329 Sennott Square Next lecture: Matlab tutoral Announcements Rules for attendng the class: Regstered for credt Regstered for audt (only f there

More information

### High Correlation between Net Promoter Score and the Development of Consumers' Willingness to Pay (Empirical Evidence from European Mobile Markets)

Hgh Correlaton between et Promoter Score and the Development of Consumers' Wllngness to Pay (Emprcal Evdence from European Moble Marets Ths paper shows that the correlaton between the et Promoter Score

More information

### Control Charts for Means (Simulation)

Chapter 290 Control Charts for Means (Smulaton) Introducton Ths procedure allows you to study the run length dstrbuton of Shewhart (Xbar), Cusum, FIR Cusum, and EWMA process control charts for means usng

More information

### PRIVATE SCHOOL CHOICE: THE EFFECTS OF RELIGIOUS AFFILIATION AND PARTICIPATION

PRIVATE SCHOOL CHOICE: THE EFFECTS OF RELIIOUS AFFILIATION AND PARTICIPATION Danny Cohen-Zada Department of Economcs, Ben-uron Unversty, Beer-Sheva 84105, Israel Wllam Sander Department of Economcs, DePaul

More information

### On Mean Squared Error of Hierarchical Estimator

S C H E D A E I N F O R M A T I C A E VOLUME 0 0 On Mean Squared Error of Herarchcal Estmator Stans law Brodowsk Faculty of Physcs, Astronomy, and Appled Computer Scence, Jagellonan Unversty, Reymonta

More information

### THE METHOD OF LEAST SQUARES THE METHOD OF LEAST SQUARES

The goal: to measure (determne) an unknown quantty x (the value of a RV X) Realsaton: n results: y 1, y 2,..., y j,..., y n, (the measured values of Y 1, Y 2,..., Y j,..., Y n ) every result s encumbered

More information

### STATISTICAL DATA ANALYSIS IN EXCEL

Mcroarray Center STATISTICAL DATA ANALYSIS IN EXCEL Lecture 6 Some Advanced Topcs Dr. Petr Nazarov 14-01-013 petr.nazarov@crp-sante.lu Statstcal data analyss n Ecel. 6. Some advanced topcs Correcton for

More information

### ActiveClean: Interactive Data Cleaning While Learning Convex Loss Models

ActveClean: Interactve Data Cleanng Whle Learnng Convex Loss Models Sanjay Krshnan, Jannan Wang, Eugene Wu, Mchael J. Frankln, Ken Goldberg UC Berkeley, Columba Unversty {sanjaykrshnan, jnwang, frankln,

More information

### Face Verification Problem. Face Recognition Problem. Application: Access Control. Biometric Authentication. Face Verification (1:1 matching)

Face Recognton Problem Face Verfcaton Problem Face Verfcaton (1:1 matchng) Querymage face query Face Recognton (1:N matchng) database Applcaton: Access Control www.vsage.com www.vsoncs.com Bometrc Authentcaton

More information

### HÜCKEL MOLECULAR ORBITAL THEORY

1 HÜCKEL MOLECULAR ORBITAL THEORY In general, the vast maorty polyatomc molecules can be thought of as consstng of a collecton of two electron bonds between pars of atoms. So the qualtatve pcture of σ

More information

### Properties of Indoor Received Signal Strength for WLAN Location Fingerprinting

Propertes of Indoor Receved Sgnal Strength for WLAN Locaton Fngerprntng Kamol Kaemarungs and Prashant Krshnamurthy Telecommuncatons Program, School of Informaton Scences, Unversty of Pttsburgh E-mal: kakst2,prashk@ptt.edu

More information

### SUPPLIER FINANCING AND STOCK MANAGEMENT. A JOINT VIEW.

SUPPLIER FINANCING AND STOCK MANAGEMENT. A JOINT VIEW. Lucía Isabel García Cebrán Departamento de Economía y Dreccón de Empresas Unversdad de Zaragoza Gran Vía, 2 50.005 Zaragoza (Span) Phone: 976-76-10-00

More information

### Section C2: BJT Structure and Operational Modes

Secton 2: JT Structure and Operatonal Modes Recall that the semconductor dode s smply a pn juncton. Dependng on how the juncton s based, current may easly flow between the dode termnals (forward bas, v

More information

### Calculation of Sampling Weights

Perre Foy Statstcs Canada 4 Calculaton of Samplng Weghts 4.1 OVERVIEW The basc sample desgn used n TIMSS Populatons 1 and 2 was a two-stage stratfed cluster desgn. 1 The frst stage conssted of a sample

More information

### Lecture 3: Force of Interest, Real Interest Rate, Annuity

Lecture 3: Force of Interest, Real Interest Rate, Annuty Goals: Study contnuous compoundng and force of nterest Dscuss real nterest rate Learn annuty-mmedate, and ts present value Study annuty-due, and

More information

### Ring structure of splines on triangulations

www.oeaw.ac.at Rng structure of splnes on trangulatons N. Vllamzar RICAM-Report 2014-48 www.rcam.oeaw.ac.at RING STRUCTURE OF SPLINES ON TRIANGULATIONS NELLY VILLAMIZAR Introducton For a trangulated regon

More information

### SPEE Recommended Evaluation Practice #6 Definition of Decline Curve Parameters Background:

SPEE Recommended Evaluaton Practce #6 efnton of eclne Curve Parameters Background: The producton hstores of ol and gas wells can be analyzed to estmate reserves and future ol and gas producton rates and

More information

### L10: Linear discriminants analysis

L0: Lnear dscrmnants analyss Lnear dscrmnant analyss, two classes Lnear dscrmnant analyss, C classes LDA vs. PCA Lmtatons of LDA Varants of LDA Other dmensonalty reducton methods CSCE 666 Pattern Analyss

More information

### Support Vector Machines

Support Vector Machnes Max Wellng Department of Computer Scence Unversty of Toronto 10 Kng s College Road Toronto, M5S 3G5 Canada wellng@cs.toronto.edu Abstract Ths s a note to explan support vector machnes.

More information

### Frequency Selective IQ Phase and IQ Amplitude Imbalance Adjustments for OFDM Direct Conversion Transmitters

Frequency Selectve IQ Phase and IQ Ampltude Imbalance Adjustments for OFDM Drect Converson ransmtters Edmund Coersmeer, Ernst Zelnsk Noka, Meesmannstrasse 103, 44807 Bochum, Germany edmund.coersmeer@noka.com,

More information

### The Mathematical Derivation of Least Squares

Pscholog 885 Prof. Federco The Mathematcal Dervaton of Least Squares Back when the powers that e forced ou to learn matr algera and calculus, I et ou all asked ourself the age-old queston: When the hell

More information

### Web Spam Detection Using Machine Learning in Specific Domain Features

Journal of Informaton Assurance and Securty 3 (2008) 220-229 Web Spam Detecton Usng Machne Learnng n Specfc Doman Features Hassan Najadat 1, Ismal Hmed 2 Department of Computer Informaton Systems Faculty

More information

### RequIn, a tool for fast web traffic inference

RequIn, a tool for fast web traffc nference Olver aul, Jean Etenne Kba GET/INT, LOR Department 9 rue Charles Fourer 90 Evry, France Olver.aul@nt-evry.fr, Jean-Etenne.Kba@nt-evry.fr Abstract As networked

More information

### Complete Fairness in Secure Two-Party Computation

Complete Farness n Secure Two-Party Computaton S. Dov Gordon Carmt Hazay Jonathan Katz Yehuda Lndell Abstract In the settng of secure two-party computaton, two mutually dstrustng partes wsh to compute

More information

### Mining Multiple Large Data Sources

The Internatonal Arab Journal of Informaton Technology, Vol. 7, No. 3, July 2 24 Mnng Multple Large Data Sources Anmesh Adhkar, Pralhad Ramachandrarao 2, Bhanu Prasad 3, and Jhml Adhkar 4 Department of

More information

### Texas Instruments 30X IIS Calculator

Texas Instruments 30X IIS Calculator Keystrokes for the TI-30X IIS are shown for a few topcs n whch keystrokes are unque. Start by readng the Quk Start secton. Then, before begnnng a specfc unt of the

More information

### QUANTUM MECHANICS, BRAS AND KETS

PH575 SPRING QUANTUM MECHANICS, BRAS AND KETS The followng summares the man relatons and defntons from quantum mechancs that we wll be usng. State of a phscal sstem: The state of a phscal sstem s represented

More information

### Learning from Multiple Outlooks

Learnng from Multple Outlooks Maayan Harel Department of Electrcal Engneerng, Technon, Hafa, Israel She Mannor Department of Electrcal Engneerng, Technon, Hafa, Israel maayanga@tx.technon.ac.l she@ee.technon.ac.l

More information

### A Hierarchical Anomaly Network Intrusion Detection System using Neural Network Classification

IDC IDC A Herarchcal Anomaly Network Intruson Detecton System usng Neural Network Classfcaton ZHENG ZHANG, JUN LI, C. N. MANIKOPOULOS, JAY JORGENSON and JOSE UCLES ECE Department, New Jersey Inst. of Tech.,

More information

### Intelligent stock trading system by turning point confirming and probabilistic reasoning

Expert Systems wth Applcatons Expert Systems wth Applcatons 34 (2008) 620 627 www.elsever.com/locate/eswa Intellgent stock tradng system by turnng pont confrmng and probablstc reasonng Depe Bao *, Zehong

More information