Class: Backoff (sections 4.6 and 4.7)

Size: px
Start display at page:

Download "Class: Backoff (sections 4.6 and 4.7)"

Transcription

1 Class: Backoff (sections 4.6 and 4.7) October 1, 2012 Admistrivia Next week Adi will talk on varaible length markov chains, and Jordan will talk on HMM s. I ll be setting up both of their talks this week. Good prediction means good compression means good models Three views: Probability: build a good model Information theory: construct a short code Statistics: predict 1

2 All three are basically the same. In particular, if we are using log loss. Probability: No loss just build a good model. Requires statistics to say some models are better than others. Information theory: loss = code length Statistics: loss = log(p(observed)) Arithmetic coding How to convert a probability to a good code. Very easy: equally likely It is easy to convert a equal probabilty to a code. Spot ourselfs an extra bit or two. List the events in order and just assign bit strings to them Almost as easy, diadic code books Try same trick with unequal probabilities Now some events have more than one code Killing off these redundent codes leads to: Huffman codes (if you sort by probabilities) Arithmetic codes (if you don t sort) 2

3 Information theory Good prediction implies good codes Primative form: Huffman codes Fancy form: Arithmetic codes Idea: random code book We have probabilty for each string Sample a string, call it string 1 Sample another string, call it string 2... (skip repeats if you like) Tell someone the index of when the true string is sampled Theorem: about 1/p, hence about log(p) bits. Markov chains for good prediction Nice model is a k-state Markov chain Theorem: LZ compression is as well as any k-state Markov chain. Variable length Markov chains are nice models Theorem: LZ compresses as well a variable length Markov chain. 3

4 Good compression implies good prediction Kraft identity: 2 code length 1 So treat 2 code length as a probability Allows converting any compression scheme into a probability forecasting scheme. Bob and I realized we could use this idea plus LZ to come up with a universal prediction algorithm. We started writing up a paper only to find that it had already won an award for best paper the year before. Their only error was the list of authors we weren t on it. So, yet another simutanious discovery. Backoff Problem: Some contexts are full, others empty. empty contexts? How to deal with Naive solution: Use best context available. If you have 5 grams, use them, otherwise If you have 4 grams, use them, otherwise... With nothing: Use base rate Example: We can do this will have lots of examples 5 grams. 4

5 Example: Twas brillig, and the slithy toves was brillig, and the slithy toves Did gyre and gimble in the wabe: All mimsy were the borogoves, And the mome raths outgrabe. So back off to base rate. Hence we guess very likely words next. like A, the,..., did Interpolation 0 / 1 context seems stupid statistically Better would be to have smooth rounding General method: ˆP (wn w n 1, ) = λ 0 (w s) ˆP (w n )+λ 1 (w s) ˆP (w n w n 1 )+ Begs the question as to how to pick the λ s. The books solution is cross validation. I.e. DUCK! But a strong statement is that the λ s depend only on the history not the current word. Is this right? Katz says no. Katz backoff Simple improvement: Use Good / Turing for the really low counts. 5

6 Not real helpful for zero counts though, since it basically gives base rates (Could this be improved by thinking of different models for each context?) So backoff until you have some data to actually use. But since we are now using the word to be forecast we have to worry about probabilities summing to 1 or not. Example. Consider the words x,y,z in order: { P ˆP (z x, y) if C(x, y, z) > 0 2 (z x, y) = α 2 (x, y) ˆP 1 (z y) otherwise { P ˆP (z y) if C(y, z) > 0 1 (z y) = α 1 (y) ˆP 0 (z) otherwise ˆP 0 (z) = P (z) (Note: There are slight changes from the book. I m using what should be done, not necessarilly what either Katz or others have said to do.) The P () are defined using Good / Turing. That is a proper probabilitic model. It assigns intellegent probabilities to unseen events. We can sum over all these unseen events and call that total α(). This keeps the probabilities summing to 1. P is now the good / turing probability. 6

7 Normalization Bayesian model generates proper predictions for zero events. This is just Good / Turing. But it uses a constant to fill in. What Katz does is refill-in these zeros by using the weaker contexts. So α is the total probability estimate of getting any zero count item. Alternative models for backoff Lots of similar results in information theory Variable length Markov chain Context tree algorithm LZW Other compression algorithms Let me describe a few of them A more principled solution? Backoff is our first real statistics problem 7

8 Room for lots of real statistical thinking here Thesis anyone? 8

Causal Semantics in Physics. exploring extra-mathematical constraints on physics equations

Causal Semantics in Physics. exploring extra-mathematical constraints on physics equations Causal Semantics in Physics exploring extra-mathematical constraints on physics equations 1 Act 1: Debating the role of causation in science Act 2: physics equations have syntax & semantics Act 3: causation

More information

Olivia chimed in. Do you think we can try using ratios so that we remember when to multiply and when to divide? Let s list out everything we know.

Olivia chimed in. Do you think we can try using ratios so that we remember when to multiply and when to divide? Let s list out everything we know. Hide Toolbars] In 1999, NASA s Mars Climate Orbiter, designed to monitor the Red Planet s atmosphere, was lost in space because engineers failed to make a simple conversion from English units to metric

More information

MA 3280 Lecture 05 - Generalized Echelon Form and Free Variables. Friday, January 31, 2014.

MA 3280 Lecture 05 - Generalized Echelon Form and Free Variables. Friday, January 31, 2014. MA 3280 Lecture 05 - Generalized Echelon Form and Free Variables Friday, January 31, 2014. Objectives: Generalize echelon form, and introduce free variables. Material from Section 3.5 starting on page

More information

Solving with Absolute Value

Solving with Absolute Value Solving with Absolute Value Who knew two little lines could cause so much trouble? Ask someone to solve the equation 3x 2 = 7 and they ll say No problem! Add just two little lines, and ask them to solve

More information

Math 31 Lesson Plan. Day 2: Sets; Binary Operations. Elizabeth Gillaspy. September 23, 2011

Math 31 Lesson Plan. Day 2: Sets; Binary Operations. Elizabeth Gillaspy. September 23, 2011 Math 31 Lesson Plan Day 2: Sets; Binary Operations Elizabeth Gillaspy September 23, 2011 Supplies needed: 30 worksheets. Scratch paper? Sign in sheet Goals for myself: Tell them what you re going to tell

More information

I bully myself cause I make me do what I put my mind to. -Marshall Mathers

I bully myself cause I make me do what I put my mind to. -Marshall Mathers I bully myself cause I make me do what I put my mind to. -Marshall Mathers Solve the following problem: `Twas 22 brillig, and the 5.6 slithy toves Did Gyre and 3.2 gimble in the ½ wabe Use any of the following

More information

Lecture 1: Shannon s Theorem

Lecture 1: Shannon s Theorem Lecture 1: Shannon s Theorem Lecturer: Travis Gagie January 13th, 2015 Welcome to Data Compression! I m Travis and I ll be your instructor this week. If you haven t registered yet, don t worry, we ll work

More information

Math101, Sections 2 and 3, Spring 2008 Review Sheet for Exam #2:

Math101, Sections 2 and 3, Spring 2008 Review Sheet for Exam #2: Math101, Sections 2 and 3, Spring 2008 Review Sheet for Exam #2: 03 17 08 3 All about lines 3.1 The Rectangular Coordinate System Know how to plot points in the rectangular coordinate system. Know the

More information

NP-Completeness I. Lecture Overview Introduction: Reduction and Expressiveness

NP-Completeness I. Lecture Overview Introduction: Reduction and Expressiveness Lecture 19 NP-Completeness I 19.1 Overview In the past few lectures we have looked at increasingly more expressive problems that we were able to solve using efficient algorithms. In this lecture we introduce

More information

Midterm sample questions

Midterm sample questions Midterm sample questions CS 585, Brendan O Connor and David Belanger October 12, 2014 1 Topics on the midterm Language concepts Translation issues: word order, multiword translations Human evaluation Parts

More information

PHYSICS 15a, Fall 2006 SPEED OF SOUND LAB Due: Tuesday, November 14

PHYSICS 15a, Fall 2006 SPEED OF SOUND LAB Due: Tuesday, November 14 PHYSICS 15a, Fall 2006 SPEED OF SOUND LAB Due: Tuesday, November 14 GENERAL INFO The goal of this lab is to determine the speed of sound in air, by making measurements and taking into consideration the

More information

Natural Language Processing. Statistical Inference: n-grams

Natural Language Processing. Statistical Inference: n-grams Natural Language Processing Statistical Inference: n-grams Updated 3/2009 Statistical Inference Statistical Inference consists of taking some data (generated in accordance with some unknown probability

More information

Entropy as a measure of surprise

Entropy as a measure of surprise Entropy as a measure of surprise Lecture 5: Sam Roweis September 26, 25 What does information do? It removes uncertainty. Information Conveyed = Uncertainty Removed = Surprise Yielded. How should we quantify

More information

Week 13: Language Modeling II Smoothing in Language Modeling. Irina Sergienya

Week 13: Language Modeling II Smoothing in Language Modeling. Irina Sergienya Week 13: Language Modeling II Smoothing in Language Modeling Irina Sergienya 07.07.2015 Couple of words first... There are much more smoothing techniques, [e.g. Katz back-off, Jelinek-Mercer,...] and techniques

More information

Lecture 10: Powers of Matrices, Difference Equations

Lecture 10: Powers of Matrices, Difference Equations Lecture 10: Powers of Matrices, Difference Equations Difference Equations A difference equation, also sometimes called a recurrence equation is an equation that defines a sequence recursively, i.e. each

More information

Please bring the task to your first physics lesson and hand it to the teacher.

Please bring the task to your first physics lesson and hand it to the teacher. Pre-enrolment task for 2014 entry Physics Why do I need to complete a pre-enrolment task? This bridging pack serves a number of purposes. It gives you practice in some of the important skills you will

More information

Problem Solving in Math (Math 43900) Fall 2013

Problem Solving in Math (Math 43900) Fall 2013 Problem Solving in Math (Math 43900) Fall 203 Week six (October ) problems recurrences Instructor: David Galvin Definition of a recurrence relation We met recurrences in the induction hand-out. Sometimes

More information

3F1 Information Theory, Lecture 3

3F1 Information Theory, Lecture 3 3F1 Information Theory, Lecture 3 Jossy Sayir Department of Engineering Michaelmas 2013, 29 November 2013 Memoryless Sources Arithmetic Coding Sources with Memory Markov Example 2 / 21 Encoding the output

More information

Chapter 1 Review of Equations and Inequalities

Chapter 1 Review of Equations and Inequalities Chapter 1 Review of Equations and Inequalities Part I Review of Basic Equations Recall that an equation is an expression with an equal sign in the middle. Also recall that, if a question asks you to solve

More information

We set up the basic model of two-sided, one-to-one matching

We set up the basic model of two-sided, one-to-one matching Econ 805 Advanced Micro Theory I Dan Quint Fall 2009 Lecture 18 To recap Tuesday: We set up the basic model of two-sided, one-to-one matching Two finite populations, call them Men and Women, who want to

More information

Sequences & Functions

Sequences & Functions Ch. 5 Sec. 1 Sequences & Functions Skip Counting to Arithmetic Sequences When you skipped counted as a child, you were introduced to arithmetic sequences. Example 1: 2, 4, 6, 8, adding 2 Example 2: 10,

More information

/633 Introduction to Algorithms Lecturer: Michael Dinitz Topic: Matroids and Greedy Algorithms Date: 10/31/16

/633 Introduction to Algorithms Lecturer: Michael Dinitz Topic: Matroids and Greedy Algorithms Date: 10/31/16 60.433/633 Introduction to Algorithms Lecturer: Michael Dinitz Topic: Matroids and Greedy Algorithms Date: 0/3/6 6. Introduction We talked a lot the last lecture about greedy algorithms. While both Prim

More information

CALCULUS III. Paul Dawkins

CALCULUS III. Paul Dawkins CALCULUS III Paul Dawkins Table of Contents Preface... iii Outline... iv Three Dimensional Space... Introduction... The -D Coordinate System... Equations of Lines... 9 Equations of Planes... 5 Quadric

More information

CS1800: Strong Induction. Professor Kevin Gold

CS1800: Strong Induction. Professor Kevin Gold CS1800: Strong Induction Professor Kevin Gold Mini-Primer/Refresher on Unrelated Topic: Limits This is meant to be a problem about reasoning about quantifiers, with a little practice of other skills, too

More information

Calculus II. Calculus II tends to be a very difficult course for many students. There are many reasons for this.

Calculus II. Calculus II tends to be a very difficult course for many students. There are many reasons for this. Preface Here are my online notes for my Calculus II course that I teach here at Lamar University. Despite the fact that these are my class notes they should be accessible to anyone wanting to learn Calculus

More information

2 Analogies between addition and multiplication

2 Analogies between addition and multiplication Problem Analysis The problem Start out with 99% water. Some of the water evaporates, end up with 98% water. How much of the water evaporates? Guesses Solution: Guesses: Not %. 2%. 5%. Not 00%. 3%..0%..5%.

More information

Preface.

Preface. This document was written and copyrighted by Paul Dawkins. Use of this document and its online version is governed by the Terms and Conditions of Use located at. The online version of this document is

More information

Hypothesis testing I. - In particular, we are talking about statistical hypotheses. [get everyone s finger length!] n =

Hypothesis testing I. - In particular, we are talking about statistical hypotheses. [get everyone s finger length!] n = Hypothesis testing I I. What is hypothesis testing? [Note we re temporarily bouncing around in the book a lot! Things will settle down again in a week or so] - Exactly what it says. We develop a hypothesis,

More information

Lecture 17: Floyd-Hoare Logic for Partial Correctness

Lecture 17: Floyd-Hoare Logic for Partial Correctness Lecture 17: Floyd-Hoare Logic for Partial Correctness Aims: To look at the following inference rules Page 1 of 9 sequence; assignment and consequence. 17.1. The Deduction System for Partial Correctness

More information

CH 73 THE QUADRATIC FORMULA, PART II

CH 73 THE QUADRATIC FORMULA, PART II 1 CH THE QUADRATIC FORMULA, PART II INTRODUCTION W ay back in Chapter 55 we used the Quadratic Formula to solve quadratic equations like 6x + 1x + 0 0, whose solutions are 5 and 8. In fact, all of the

More information

DIFFERENTIAL EQUATIONS

DIFFERENTIAL EQUATIONS DIFFERENTIAL EQUATIONS Basic Concepts Paul Dawkins Table of Contents Preface... Basic Concepts... 1 Introduction... 1 Definitions... Direction Fields... 8 Final Thoughts...19 007 Paul Dawkins i http://tutorial.math.lamar.edu/terms.aspx

More information

Hidden Markov Models: All the Glorious Gory Details

Hidden Markov Models: All the Glorious Gory Details Hidden Markov Models: All the Glorious Gory Details Noah A. Smith Department of Computer Science Johns Hopkins University nasmith@cs.jhu.edu 18 October 2004 1 Introduction Hidden Markov models (HMMs, hereafter)

More information

Lecture 6: Introducing Complexity

Lecture 6: Introducing Complexity COMP26120: Algorithms and Imperative Programming Lecture 6: Introducing Complexity Ian Pratt-Hartmann Room KB2.38: email: ipratt@cs.man.ac.uk 2015 16 You need this book: Make sure you use the up-to-date

More information

Chemistry Chapter 6 Practice Problems. To avoid confusion, there is no Iodine on this set. If you see Cl, it is Chlorine!

Chemistry Chapter 6 Practice Problems. To avoid confusion, there is no Iodine on this set. If you see Cl, it is Chlorine! Chemistry Chapter 6 Practice Problems The more you do, the better you ll be prepared Some contain some little tricks to keep you thinking, but generally they look very much like the test will. The test

More information

You separate binary numbers into columns in a similar fashion. 2 5 = 32

You separate binary numbers into columns in a similar fashion. 2 5 = 32 RSA Encryption 2 At the end of Part I of this article, we stated that RSA encryption works because it s impractical to factor n, which determines P 1 and P 2, which determines our private key, d, which

More information

Lecture 4: Applications of Orthogonality: QR Decompositions

Lecture 4: Applications of Orthogonality: QR Decompositions Math 08B Professor: Padraic Bartlett Lecture 4: Applications of Orthogonality: QR Decompositions Week 4 UCSB 204 In our last class, we described the following method for creating orthonormal bases, known

More information

3F1 Information Theory, Lecture 3

3F1 Information Theory, Lecture 3 3F1 Information Theory, Lecture 3 Jossy Sayir Department of Engineering Michaelmas 2011, 28 November 2011 Memoryless Sources Arithmetic Coding Sources with Memory 2 / 19 Summary of last lecture Prefix-free

More information

Learning Objectives

Learning Objectives Learning Objectives Learn about recurrence relations Learn the relationship between sequences and recurrence relations Explore how to solve recurrence relations by iteration Learn about linear homogeneous

More information

CISC 4090: Theory of Computation Chapter 1 Regular Languages. Section 1.1: Finite Automata. What is a computer? Finite automata

CISC 4090: Theory of Computation Chapter 1 Regular Languages. Section 1.1: Finite Automata. What is a computer? Finite automata CISC 4090: Theory of Computation Chapter Regular Languages Xiaolan Zhang, adapted from slides by Prof. Werschulz Section.: Finite Automata Fordham University Department of Computer and Information Sciences

More information

CS1800: Mathematical Induction. Professor Kevin Gold

CS1800: Mathematical Induction. Professor Kevin Gold CS1800: Mathematical Induction Professor Kevin Gold Induction: Used to Prove Patterns Just Keep Going For an algorithm, we may want to prove that it just keeps working, no matter how big the input size

More information

Logic. Propositional Logic: Syntax

Logic. Propositional Logic: Syntax Logic Propositional Logic: Syntax Logic is a tool for formalizing reasoning. There are lots of different logics: probabilistic logic: for reasoning about probability temporal logic: for reasoning about

More information

Cryptography and Network Security Prof. D. Mukhopadhyay Department of Computer Science and Engineering Indian Institute of Technology, Kharagpur

Cryptography and Network Security Prof. D. Mukhopadhyay Department of Computer Science and Engineering Indian Institute of Technology, Kharagpur Cryptography and Network Security Prof. D. Mukhopadhyay Department of Computer Science and Engineering Indian Institute of Technology, Kharagpur Module No. # 01 Lecture No. # 08 Shannon s Theory (Contd.)

More information

[Disclaimer: This is not a complete list of everything you need to know, just some of the topics that gave people difficulty.]

[Disclaimer: This is not a complete list of everything you need to know, just some of the topics that gave people difficulty.] Math 43 Review Notes [Disclaimer: This is not a complete list of everything you need to know, just some of the topics that gave people difficulty Dot Product If v (v, v, v 3 and w (w, w, w 3, then the

More information

Algebra. Here are a couple of warnings to my students who may be here to get a copy of what happened on a day that you missed.

Algebra. Here are a couple of warnings to my students who may be here to get a copy of what happened on a day that you missed. This document was written and copyrighted by Paul Dawkins. Use of this document and its online version is governed by the Terms and Conditions of Use located at. The online version of this document is

More information

MACHINE LEARNING INTRODUCTION: STRING CLASSIFICATION

MACHINE LEARNING INTRODUCTION: STRING CLASSIFICATION MACHINE LEARNING INTRODUCTION: STRING CLASSIFICATION THOMAS MAILUND Machine learning means different things to different people, and there is no general agreed upon core set of algorithms that must be

More information

An Intuitive Introduction to Motivic Homotopy Theory Vladimir Voevodsky

An Intuitive Introduction to Motivic Homotopy Theory Vladimir Voevodsky What follows is Vladimir Voevodsky s snapshot of his Fields Medal work on motivic homotopy, plus a little philosophy and from my point of view the main fun of doing mathematics Voevodsky (2002). Voevodsky

More information

Chapter 2: Source coding

Chapter 2: Source coding Chapter 2: meghdadi@ensil.unilim.fr University of Limoges Chapter 2: Entropy of Markov Source Chapter 2: Entropy of Markov Source Markov model for information sources Given the present, the future is independent

More information

Basic Principles of Lossless Coding. Universal Lossless coding. Lempel-Ziv Coding. 2. Exploit dependences between successive symbols.

Basic Principles of Lossless Coding. Universal Lossless coding. Lempel-Ziv Coding. 2. Exploit dependences between successive symbols. Universal Lossless coding Lempel-Ziv Coding Basic principles of lossless compression Historical review Variable-length-to-block coding Lempel-Ziv coding 1 Basic Principles of Lossless Coding 1. Exploit

More information

AN ALGEBRA PRIMER WITH A VIEW TOWARD CURVES OVER FINITE FIELDS

AN ALGEBRA PRIMER WITH A VIEW TOWARD CURVES OVER FINITE FIELDS AN ALGEBRA PRIMER WITH A VIEW TOWARD CURVES OVER FINITE FIELDS The integers are the set 1. Groups, Rings, and Fields: Basic Examples Z := {..., 3, 2, 1, 0, 1, 2, 3,...}, and we can add, subtract, and multiply

More information

The Gram-Schmidt Process

The Gram-Schmidt Process The Gram-Schmidt Process How and Why it Works This is intended as a complement to 5.4 in our textbook. I assume you have read that section, so I will not repeat the definitions it gives. Our goal is to

More information

Logic. Propositional Logic: Syntax. Wffs

Logic. Propositional Logic: Syntax. Wffs Logic Propositional Logic: Syntax Logic is a tool for formalizing reasoning. There are lots of different logics: probabilistic logic: for reasoning about probability temporal logic: for reasoning about

More information

Some Review Problems for Exam 2: Solutions

Some Review Problems for Exam 2: Solutions Math 5366 Fall 017 Some Review Problems for Exam : Solutions 1 Find the coefficient of x 15 in each of the following: 1 (a) (1 x) 6 Solution: 1 (1 x) = ( ) k + 5 x k 6 k ( ) ( ) 0 0 so the coefficient

More information

CHAPTER 7: TECHNIQUES OF INTEGRATION

CHAPTER 7: TECHNIQUES OF INTEGRATION CHAPTER 7: TECHNIQUES OF INTEGRATION DAVID GLICKENSTEIN. Introduction This semester we will be looking deep into the recesses of calculus. Some of the main topics will be: Integration: we will learn how

More information

Knots, Coloring and Applications

Knots, Coloring and Applications Knots, Coloring and Applications Ben Webster University of Virginia March 10, 2015 Ben Webster (UVA) Knots, Coloring and Applications March 10, 2015 1 / 14 This talk is online at http://people.virginia.edu/~btw4e/knots.pdf

More information

Axiomatic systems. Revisiting the rules of inference. Example: A theorem and its proof in an abstract axiomatic system:

Axiomatic systems. Revisiting the rules of inference. Example: A theorem and its proof in an abstract axiomatic system: Axiomatic systems Revisiting the rules of inference Material for this section references College Geometry: A Discovery Approach, 2/e, David C. Kay, Addison Wesley, 2001. In particular, see section 2.1,

More information

Source Coding. Master Universitario en Ingeniería de Telecomunicación. I. Santamaría Universidad de Cantabria

Source Coding. Master Universitario en Ingeniería de Telecomunicación. I. Santamaría Universidad de Cantabria Source Coding Master Universitario en Ingeniería de Telecomunicación I. Santamaría Universidad de Cantabria Contents Introduction Asymptotic Equipartition Property Optimal Codes (Huffman Coding) Universal

More information

Image and Multidimensional Signal Processing

Image and Multidimensional Signal Processing Image and Multidimensional Signal Processing Professor William Hoff Dept of Electrical Engineering &Computer Science http://inside.mines.edu/~whoff/ Image Compression 2 Image Compression Goal: Reduce amount

More information

Do students sleep the recommended 8 hours a night on average?

Do students sleep the recommended 8 hours a night on average? BIEB100. Professor Rifkin. Notes on Section 2.2, lecture of 27 January 2014. Do students sleep the recommended 8 hours a night on average? We first set up our null and alternative hypotheses: H0: μ= 8

More information

Lecture 8: Conditional probability I: definition, independence, the tree method, sampling, chain rule for independent events

Lecture 8: Conditional probability I: definition, independence, the tree method, sampling, chain rule for independent events Lecture 8: Conditional probability I: definition, independence, the tree method, sampling, chain rule for independent events Discrete Structures II (Summer 2018) Rutgers University Instructor: Abhishek

More information

Grades 7 & 8, Math Circles 10/11/12 October, Series & Polygonal Numbers

Grades 7 & 8, Math Circles 10/11/12 October, Series & Polygonal Numbers Faculty of Mathematics Waterloo, Ontario N2L G Centre for Education in Mathematics and Computing Introduction Grades 7 & 8, Math Circles 0//2 October, 207 Series & Polygonal Numbers Mathematicians are

More information

Logic: The Big Picture

Logic: The Big Picture Logic: The Big Picture A typical logic is described in terms of syntax: what are the legitimate formulas semantics: under what circumstances is a formula true proof theory/ axiomatization: rules for proving

More information

Language Models. Data Science: Jordan Boyd-Graber University of Maryland SLIDES ADAPTED FROM PHILIP KOEHN

Language Models. Data Science: Jordan Boyd-Graber University of Maryland SLIDES ADAPTED FROM PHILIP KOEHN Language Models Data Science: Jordan Boyd-Graber University of Maryland SLIDES ADAPTED FROM PHILIP KOEHN Data Science: Jordan Boyd-Graber UMD Language Models 1 / 8 Language models Language models answer

More information

ASTRO 114 Lecture Okay. We re now gonna continue discussing and conclude discussing the entire

ASTRO 114 Lecture Okay. We re now gonna continue discussing and conclude discussing the entire ASTRO 114 Lecture 55 1 Okay. We re now gonna continue discussing and conclude discussing the entire universe. So today we re gonna learn about everything, everything that we know of. There s still a lot

More information

What disease has been tamed to some degree through the help of chemical bonding theory?

What disease has been tamed to some degree through the help of chemical bonding theory? Chapter 9 Reading Guide Bonding I Dr. Baxley Tro 4 th edition 1 See end of this document for suggested (strongly urged) problems. Section 9.1, Why is chemical bonding important? What disease has been tamed

More information

Error Correcting Codes Prof. Dr. P. Vijay Kumar Department of Electrical Communication Engineering Indian Institute of Science, Bangalore

Error Correcting Codes Prof. Dr. P. Vijay Kumar Department of Electrical Communication Engineering Indian Institute of Science, Bangalore (Refer Slide Time: 00:15) Error Correcting Codes Prof. Dr. P. Vijay Kumar Department of Electrical Communication Engineering Indian Institute of Science, Bangalore Lecture No. # 03 Mathematical Preliminaries:

More information

Recap: Language models. Foundations of Natural Language Processing Lecture 4 Language Models: Evaluation and Smoothing. Two types of evaluation in NLP

Recap: Language models. Foundations of Natural Language Processing Lecture 4 Language Models: Evaluation and Smoothing. Two types of evaluation in NLP Recap: Language models Foundations of atural Language Processing Lecture 4 Language Models: Evaluation and Smoothing Alex Lascarides (Slides based on those from Alex Lascarides, Sharon Goldwater and Philipp

More information

Descriptive Statistics (And a little bit on rounding and significant digits)

Descriptive Statistics (And a little bit on rounding and significant digits) Descriptive Statistics (And a little bit on rounding and significant digits) Now that we know what our data look like, we d like to be able to describe it numerically. In other words, how can we represent

More information

Discrete Mathematics and Probability Theory Fall 2014 Anant Sahai Note 15. Random Variables: Distributions, Independence, and Expectations

Discrete Mathematics and Probability Theory Fall 2014 Anant Sahai Note 15. Random Variables: Distributions, Independence, and Expectations EECS 70 Discrete Mathematics and Probability Theory Fall 204 Anant Sahai Note 5 Random Variables: Distributions, Independence, and Expectations In the last note, we saw how useful it is to have a way of

More information

TheFourierTransformAndItsApplications-Lecture28

TheFourierTransformAndItsApplications-Lecture28 TheFourierTransformAndItsApplications-Lecture28 Instructor (Brad Osgood):All right. Let me remind you of the exam information as I said last time. I also sent out an announcement to the class this morning

More information

P (E) = P (A 1 )P (A 2 )... P (A n ).

P (E) = P (A 1 )P (A 2 )... P (A n ). Lecture 9: Conditional probability II: breaking complex events into smaller events, methods to solve probability problems, Bayes rule, law of total probability, Bayes theorem Discrete Structures II (Summer

More information

Secret Sharing: Four People, Need Three

Secret Sharing: Four People, Need Three Secret Sharing A secret is an n-bit string. Throughout this talk assume that Zelda has a secret s {0, 1} n. She will want to give shares of the secret to various people. Applications Rumor: Secret Sharing

More information

Where I is the current in Amperes, q is the charge in coulombs, and t is the time in seconds.

Where I is the current in Amperes, q is the charge in coulombs, and t is the time in seconds. CURRENT ELECTRICITY What is current? All of use like to ask this little question, which is quite easily answered, but not so easily understood. Current is the flow of charges. In other words, the flow,

More information

Lecture 22: Quantum computational complexity

Lecture 22: Quantum computational complexity CPSC 519/619: Quantum Computation John Watrous, University of Calgary Lecture 22: Quantum computational complexity April 11, 2006 This will be the last lecture of the course I hope you have enjoyed the

More information

As in the book. Each of the other thermopotentials yield similar and useful relations. Let s just do the Legendre Transformations and list them all:

As in the book. Each of the other thermopotentials yield similar and useful relations. Let s just do the Legendre Transformations and list them all: hysics 430 Solutions to roblem Set 9 1. Schroeder roblem 5.1. Mixed derivatives are fun and useful. One of the most important things that we have yet to see is that thermodynamics INSISS that certain things,

More information

8.1 Solutions of homogeneous linear differential equations

8.1 Solutions of homogeneous linear differential equations Math 21 - Spring 2014 Classnotes, Week 8 This week we will talk about solutions of homogeneous linear differential equations. This material doubles as an introduction to linear algebra, which is the subject

More information

Answers in blue. If you have questions or spot an error, let me know. 1. Find all matrices that commute with A =. 4 3

Answers in blue. If you have questions or spot an error, let me know. 1. Find all matrices that commute with A =. 4 3 Answers in blue. If you have questions or spot an error, let me know. 3 4. Find all matrices that commute with A =. 4 3 a b If we set B = and set AB = BA, we see that 3a + 4b = 3a 4c, 4a + 3b = 3b 4d,

More information

Final Review Sheet. B = (1, 1 + 3x, 1 + x 2 ) then 2 + 3x + 6x 2

Final Review Sheet. B = (1, 1 + 3x, 1 + x 2 ) then 2 + 3x + 6x 2 Final Review Sheet The final will cover Sections Chapters 1,2,3 and 4, as well as sections 5.1-5.4, 6.1-6.2 and 7.1-7.3 from chapters 5,6 and 7. This is essentially all material covered this term. Watch

More information

Any questions? Anything on anyone s mind? Other than that, any other corrections?

Any questions? Anything on anyone s mind? Other than that, any other corrections? TheFourierTransformAndItsApplications-Lecture06 Instructor (Brad Osgood):Okay, we re on. First thing I want to say is that I made a little mistake last time in lecture. It was gently pointed out to me.

More information

Introduction to Algebra: The First Week

Introduction to Algebra: The First Week Introduction to Algebra: The First Week Background: According to the thermostat on the wall, the temperature in the classroom right now is 72 degrees Fahrenheit. I want to write to my friend in Europe,

More information

MAT 211, Spring 2015, Introduction to Linear Algebra.

MAT 211, Spring 2015, Introduction to Linear Algebra. MAT 211, Spring 2015, Introduction to Linear Algebra. Lecture 04, 53103: MWF 10-10:53 AM. Location: Library W4535 Contact: mtehrani@scgp.stonybrook.edu Final Exam: Monday 5/18/15 8:00 AM-10:45 AM The aim

More information

Kolmogorov complexity ; induction, prediction and compression

Kolmogorov complexity ; induction, prediction and compression Kolmogorov complexity ; induction, prediction and compression Contents 1 Motivation for Kolmogorov complexity 1 2 Formal Definition 2 3 Trying to compute Kolmogorov complexity 3 4 Standard upper bounds

More information

Differential Equations

Differential Equations This document was written and copyrighted by Paul Dawkins. Use of this document and its online version is governed by the Terms and Conditions of Use located at. The online version of this document is

More information

Statistical Methods for NLP

Statistical Methods for NLP Statistical Methods for NLP Language Models, Graphical Models Sameer Maskey Week 13, April 13, 2010 Some slides provided by Stanley Chen and from Bishop Book Resources 1 Announcements Final Project Due,

More information

1 Maintaining a Dictionary

1 Maintaining a Dictionary 15-451/651: Design & Analysis of Algorithms February 1, 2016 Lecture #7: Hashing last changed: January 29, 2016 Hashing is a great practical tool, with an interesting and subtle theory too. In addition

More information

Math 308 Midterm Answers and Comments July 18, Part A. Short answer questions

Math 308 Midterm Answers and Comments July 18, Part A. Short answer questions Math 308 Midterm Answers and Comments July 18, 2011 Part A. Short answer questions (1) Compute the determinant of the matrix a 3 3 1 1 2. 1 a 3 The determinant is 2a 2 12. Comments: Everyone seemed to

More information

Computability Crib Sheet

Computability Crib Sheet Computer Science and Engineering, UCSD Winter 10 CSE 200: Computability and Complexity Instructor: Mihir Bellare Computability Crib Sheet January 3, 2010 Computability Crib Sheet This is a quick reference

More information

CS1800: Sequences & Sums. Professor Kevin Gold

CS1800: Sequences & Sums. Professor Kevin Gold CS1800: Sequences & Sums Professor Kevin Gold Moving Toward Analysis of Algorithms Today s tools help in the analysis of algorithms. We ll cover tools for deciding what equation best fits a sequence of

More information

Operation and Supply Chain Management Prof. G. Srinivasan Department of Management Studies Indian Institute of Technology, Madras

Operation and Supply Chain Management Prof. G. Srinivasan Department of Management Studies Indian Institute of Technology, Madras Operation and Supply Chain Management Prof. G. Srinivasan Department of Management Studies Indian Institute of Technology, Madras Lecture - 3 Forecasting Linear Models, Regression, Holt s, Seasonality

More information

Lecture 1: September 25, A quick reminder about random variables and convexity

Lecture 1: September 25, A quick reminder about random variables and convexity Information and Coding Theory Autumn 207 Lecturer: Madhur Tulsiani Lecture : September 25, 207 Administrivia This course will cover some basic concepts in information and coding theory, and their applications

More information

- a value calculated or derived from the data.

- a value calculated or derived from the data. Descriptive statistics: Note: I'm assuming you know some basics. If you don't, please read chapter 1 on your own. It's pretty easy material, and it gives you a good background as to why we need statistics.

More information

MAT2342 : Introduction to Applied Linear Algebra Mike Newman, fall Projections. introduction

MAT2342 : Introduction to Applied Linear Algebra Mike Newman, fall Projections. introduction MAT4 : Introduction to Applied Linear Algebra Mike Newman fall 7 9. Projections introduction One reason to consider projections is to understand approximate solutions to linear systems. A common example

More information

Lecture 4: Constructing the Integers, Rationals and Reals

Lecture 4: Constructing the Integers, Rationals and Reals Math/CS 20: Intro. to Math Professor: Padraic Bartlett Lecture 4: Constructing the Integers, Rationals and Reals Week 5 UCSB 204 The Integers Normally, using the natural numbers, you can easily define

More information

Hi, I'm Jocelyn, and we're going to go over Fall 2009, Exam 1, problem number 2.

Hi, I'm Jocelyn, and we're going to go over Fall 2009, Exam 1, problem number 2. MIT OpenCourseWare http://ocw.mit.edu 3.091SC Introduction to Solid State Chemistry, Fall 2010 Transcript Exam 1 Problem 2 The following content is provided under a Creative Commons license. Your support

More information

Stat 5421 Lecture Notes Proper Conjugate Priors for Exponential Families Charles J. Geyer March 28, 2016

Stat 5421 Lecture Notes Proper Conjugate Priors for Exponential Families Charles J. Geyer March 28, 2016 Stat 5421 Lecture Notes Proper Conjugate Priors for Exponential Families Charles J. Geyer March 28, 2016 1 Theory This section explains the theory of conjugate priors for exponential families of distributions,

More information

Linear Algebra, Summer 2011, pt. 2

Linear Algebra, Summer 2011, pt. 2 Linear Algebra, Summer 2, pt. 2 June 8, 2 Contents Inverses. 2 Vector Spaces. 3 2. Examples of vector spaces..................... 3 2.2 The column space......................... 6 2.3 The null space...........................

More information

MATH 308 COURSE SUMMARY

MATH 308 COURSE SUMMARY MATH 308 COURSE SUMMARY Approximately a third of the exam cover the material from the first two midterms, that is, chapter 6 and the first six sections of chapter 7. The rest of the exam will cover the

More information

CS4800: Algorithms & Data Jonathan Ullman

CS4800: Algorithms & Data Jonathan Ullman CS4800: Algorithms & Data Jonathan Ullman Lecture 22: Greedy Algorithms: Huffman Codes Data Compression and Entropy Apr 5, 2018 Data Compression How do we store strings of text compactly? A (binary) code

More information

6.080 / Great Ideas in Theoretical Computer Science Spring 2008

6.080 / Great Ideas in Theoretical Computer Science Spring 2008 MIT OpenCourseWare http://ocw.mit.edu 6.080 / 6.089 Great Ideas in Theoretical Computer Science Spring 2008 For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.

More information

Chapter 5: Data Compression

Chapter 5: Data Compression Chapter 5: Data Compression Definition. A source code C for a random variable X is a mapping from the range of X to the set of finite length strings of symbols from a D-ary alphabet. ˆX: source alphabet,

More information

MA 1125 Lecture 15 - The Standard Normal Distribution. Friday, October 6, Objectives: Introduce the standard normal distribution and table.

MA 1125 Lecture 15 - The Standard Normal Distribution. Friday, October 6, Objectives: Introduce the standard normal distribution and table. MA 1125 Lecture 15 - The Standard Normal Distribution Friday, October 6, 2017. Objectives: Introduce the standard normal distribution and table. 1. The Standard Normal Distribution We ve been looking at

More information