March 1, Florida State University. Concentration Inequalities: Martingale. Approach and Entropy Method. Lizhe Sun and Boning Yang.

Size: px
Start display at page:

Download "March 1, Florida State University. Concentration Inequalities: Martingale. Approach and Entropy Method. Lizhe Sun and Boning Yang."

Transcription

1 Florida State University March 1, 2018

2 Framework 1. (Lizhe) Basic inequalities Chernoff bounding Review for STA (Lizhe) Discrete-time martingales inequalities via martingale approach 3. (Boning)

3 Part I:

4 Why concentration inequalities? inequalities quantify how a random variable X deviates around its mean µ. They usually take the form of two-sides bounds for the tails of X µ, such as P( X µ t) something very small, t > 0. Based on CLT, we can get the asymptotic results when n. But in machine learning community and in high dimensional data analysis, we prefer to exploit the non-asymptotic properties for random variables.

5 Basic inequalities I First of all, we recall some basic tools and inequalities. For any nonnegative random variable X, E(X ) = 0 P(X t)dt. This implies Markov inequality: for any nonnegative random variable X, and t > 0, P(X t) E(X ). t

6 Basic inequalities II In general, if φ is a strictly monotonically increasing nonnegative-valued function, then for any random variable X and real number t, P(X t) = P{φ(X ) φ(t)} E(φ(X )). φ(t) For example, φ(x) = x 2 will induce Chebyshev s inequality: if X is an arbitrary random variable and t > 0, then P{ X EX t} = P{ X EX 2 t 2 } E X EX 2 t 2 = var{x } t 2

7 Chernoff bounding technique Taking φ(x) = exp(λx) where λ is an arbitrary positive number, for any random variable X, and any t > 0, we have P(X t) = P{exp(λX ) exp(λt)} exp( λt)e[exp(λx )] exp( λt + log E[exp(λX )]), λ > 0. Please note that if we want to bound the probability of the lower tail, P(X t), we follow the same steps, but with X rather than X. Now, we need to obtain tight uppers bound for exp( λt + log E[exp(λX )]).

8 Chernoff bounding technique: Hoeffding inequality I Theorem Let X be a random variable, such that X [a, b] a.s. for some finite a < b. Then, for any λ 0, ( ) E[exp(λ(X EX ))] exp λ 2 (b a) 2 /8 Proof. For any p [0, 1] and x R, let us define the function H p (s) = log[p exp(s(1 p)) + (1 p) exp( sp)]. Let ξ = X EX, where ξ [a EX, b EX ]. Using the convexity of the exponential function, we can write

9 Chernoff bounding technique: Hoeffding inequality II Proof. exp(λξ) = ( X a b X ) exp λ(b EX ) + b a b a t(a EX ) ( X a ) (b X ) exp(λ(b EX )) + exp(λ(a EX )). b a b a Taking expectations of both sides, we can get E[exp(λξ)] ( EX a ) (b EX ) exp(λ(b EX ))+ exp(λ(a EX )). b a b a Let p = EX a b a and s = λ(b a), we have exp(h p (s)) = ( EX a ) (b EX ) exp(λ(b EX ))+ exp(λ(a EX )). b a b a

10 Chernoff bounding technique: Hoeffding inequality III Proof. Using Taylor expansion, we can get the bound H p (s) s2 8 for all p [0, 1] and all s R, by using the above definitions of p and s, we can get the Hoeffding inequality. A disadvantage of Hoeffding inequality is that it ignores information about variance of the X. And the Bernstein inequality provide an improvement in this respect. Question: Why we use Chernoff bounding technique?

11 Example: bounding the random walk Symmetric Bernoulli distribution A random variable X has symmetric Bernoulli distribution (also called Rademacher distribution) if P(X = 1) = P(X = 1) = 1 2 Let X 1, X 2,, X n be independent symmetric Bernoulli random variables. Then for any t 0, we have P ( n X i t ) exp ( t2 ) 2n

12 Chernoff bounding technique: beyond Hoeffding inequality Bernstein s condition: Given a random variable X with EX = µ and var(x ) = σ 2, we say that the Bernstein s condition with parameter b holds if E[(X µ) k ] 1 2 k!σ2 b k 2, k = 3, 4, Theorem For any random variable X satisfying the Bernstein condition, we have E[exp(λ(X µ))] exp( λ2 σ 2 /2 1 b λ ), for all λ < 1 b, and moreover, we have the concentration inequality P( X µ t) 2 exp( 2(σ 2 ), for all t 0. + bt) t 2

13 Review for STA 6448

14 Part II:

15 Discrete-time martingales Definition Let (Ω, F, P) be a probability space. A sequence {X i, F i } n i=0, n N, where the X i are random variables and the F i are σ-algebras, is a martingale if the following conditions are satisifed: 1. The F i form a filteration, i.e., {, Ω} = F 0 F 1 F n = F. 2. X i L 1 (Ω, F i, P) for every i {0, 1,, n}. 3. For all i {1, 2,, n}, the equality E[X i F i 1 ] = X i 1 holds almost sure (a.s.). A martingale can be generated by the following procedure: given a r.v. X associated with a filtration {F i } n i=0, let X i = E[X F i ], i {1, 2, 3,, n}. Then, the sequence X 0, X 1, X 2,, X n forms a martingale.

16 decomposition Consider a r.v. X associated with a filtration {F i } n i=0, where F 0 = {, Ω} and F n = F, we have Consider E[X F 0 ] = EX and E[X F n ] = X. X EX = E[X F n ] E[X F 0 ] n = (E[X F i ] E[X F i 1 ]) = n ξ i, in which ξ i = E[X F i ] E[X F i 1 ]. Here, we call {ξ i } n martingale difference.

17 Review: Chernoff bounding technique Here, if we consider logarithmic moment generating function again, we have the following equality: log E[exp(λ(X EX ))] = log E[exp(λ n ξ i )] n = log E[ exp(λξ i )] Here, a intuitive idea is to bound each exp(λξ i ), i = 1, 2,, n.

18 Azuma inequality I Theorem Let {X i, F i } n i=0 be a real-valued martingale sequence. Suppose that there exist nonnegative real number d 1, d 2,, d n, such that X i X i 1 d i a.s. for all i {1, 2,, n}. Then, for every t > 0, P( X n X 0 t) 2 exp( 2 n t 2 d i 2 ) Proof. Here, we just consider P(X n X 0 t), t > 0. Let ξ i = X i X i 1 for i = 1, 2,, n denote the martingale difference. According to the assumption, we have ξ i d i and E[X i F i 1 ] = 0 a.s. for every i {1, 2,, n}.

19 Azuma inequality II Proof. Now we use the Chernoff technique: P(X n X 0 t) = P( n ξ i t) exp( λt)e[exp(λ n ξ i )], λ 0. Furthermore, E[exp(λ n [ ξ i )] = E E [ exp(λ n ξ i ) ] ] Fn 1 [ n 1 ] = E exp(λ ξ i )E[exp(λξ i ) F n 1 ]

20 Azuma inequality III Proof. The last equality holds since exp(λ n 1 ξ i) is F n 1 -measurable. And we can apply Hoeffding inequality to ξ i conditioned on F n 1. Because we know that E[ξ n F n 1 ] = 0 and ξ n [ d n, d n ] a.s., According to the Hoeffding inequality, we have E[exp(λξ n ) F n 1 ] exp ( λ 2 d 2 n 2 By continuing recursively the above inequality, we can bound the inequality by E [ exp(λ n ξ i ) ] n exp ( λ 2 di 2 ) (λ 2 = exp 2 2 ) n di 2 )

21 Azuma inequality IV Proof. Plugging the above bound into the inequality, we have P(X n X 0 t) exp ( λt + λ2 2 n di 2 ) t 0. After minimizing the right side of the above inequality, we get Above all, we have P(X n X 0 t) exp( 2 n t 2 P( X n X 0 t) 2 exp( 2 n d i 2 t 2 ) d i 2 )

22 McDiarmid inequality Bounded difference assumption Let f : R n R be a function that satisfies the bounded difference assumption sup f (x 1,, x i,, x n ) f (x 1,, x x 1,x 2,,x n,x i R i,, x n ) d i for every 1 i n, where d 1,, d n are some nonnegative real constants. Theorem Suppose that a measurable function f satisfies the bounded difference assumption with parameters (d 1, d 2,, d n ) and let X i n be independent (not necessary i.i.d) in some measurable space. Then P( f (X 1, X 2,, X n ) E[f (X 1, X 2,, X n )] t) 2 exp ( 2t2 ) n d i 2

23 McDiarmid inequality: the outline of the proof Construct the martingale difference ξ i = E[f (X 1, X 2,, X n ) F i ] E[f (X 1, X 2,, X n ) F i 1 ] Thus we get f (X 1, X 2,, X n ) E[f (X 1, X 2,, X n )] = n ξ i Bounded ξ i, construct r.v. A i and B i, prove A i ξ i B i and B i A i d i. Using Hoeffding inequality, similar to Azuma inequality.

24 A summary

25 Reference [1]. Lecture notes and related materials in STA [2]. Raginsky, Maxim, and Igal Sason. of measure inequalities in information theory, communications, and coding. Foundations and Trends in Communications and Information Theory (2013):

26 Thank you Thank you

STAT 200C: High-dimensional Statistics

STAT 200C: High-dimensional Statistics STAT 200C: High-dimensional Statistics Arash A. Amini May 30, 2018 1 / 59 Classical case: n d. Asymptotic assumption: d is fixed and n. Basic tools: LLN and CLT. High-dimensional setting: n d, e.g. n/d

More information

Machine Learning Theory Lecture 4

Machine Learning Theory Lecture 4 Machine Learning Theory Lecture 4 Nicholas Harvey October 9, 018 1 Basic Probability One of the first concentration bounds that you learn in probability theory is Markov s inequality. It bounds the right-tail

More information

Statistical Machine Learning

Statistical Machine Learning Statistical Machine Learning UoC Stats 37700, Winter quarter Lecture 2: Introduction to statistical learning theory. 1 / 22 Goals of statistical learning theory SLT aims at studying the performance of

More information

Lecture 2. We now introduce some fundamental tools in martingale theory, which are useful in controlling the fluctuation of martingales.

Lecture 2. We now introduce some fundamental tools in martingale theory, which are useful in controlling the fluctuation of martingales. Lecture 2 1 Martingales We now introduce some fundamental tools in martingale theory, which are useful in controlling the fluctuation of martingales. 1.1 Doob s inequality We have the following maximal

More information

STAT 200C: High-dimensional Statistics

STAT 200C: High-dimensional Statistics STAT 200C: High-dimensional Statistics Arash A. Amini April 27, 2018 1 / 80 Classical case: n d. Asymptotic assumption: d is fixed and n. Basic tools: LLN and CLT. High-dimensional setting: n d, e.g. n/d

More information

Concentration inequalities and the entropy method

Concentration inequalities and the entropy method Concentration inequalities and the entropy method Gábor Lugosi ICREA and Pompeu Fabra University Barcelona what is concentration? We are interested in bounding random fluctuations of functions of many

More information

Tail and Concentration Inequalities

Tail and Concentration Inequalities CSE 694: Probabilistic Analysis and Randomized Algorithms Lecturer: Hung Q. Ngo SUNY at Buffalo, Spring 2011 Last update: February 19, 2011 Tail and Concentration Ineualities From here on, we use 1 A to

More information

Lecture 1 Measure concentration

Lecture 1 Measure concentration CSE 29: Learning Theory Fall 2006 Lecture Measure concentration Lecturer: Sanjoy Dasgupta Scribe: Nakul Verma, Aaron Arvey, and Paul Ruvolo. Concentration of measure: examples We start with some examples

More information

P (A G) dp G P (A G)

P (A G) dp G P (A G) First homework assignment. Due at 12:15 on 22 September 2016. Homework 1. We roll two dices. X is the result of one of them and Z the sum of the results. Find E [X Z. Homework 2. Let X be a r.v.. Assume

More information

High Dimensional Probability

High Dimensional Probability High Dimensional Probability for Mathematicians and Data Scientists Roman Vershynin 1 1 University of Michigan. Webpage: www.umich.edu/~romanv ii Preface Who is this book for? This is a textbook in probability

More information

Concentration inequalities and tail bounds

Concentration inequalities and tail bounds Concentration inequalities and tail bounds John Duchi Outline I Basics and motivation 1 Law of large numbers 2 Markov inequality 3 Cherno bounds II Sub-Gaussian random variables 1 Definitions 2 Examples

More information

COMPSCI 240: Reasoning Under Uncertainty

COMPSCI 240: Reasoning Under Uncertainty COMPSCI 240: Reasoning Under Uncertainty Andrew Lan and Nic Herndon University of Massachusetts at Amherst Spring 2019 Lecture 20: Central limit theorem & The strong law of large numbers Markov and Chebyshev

More information

On the Concentration of the Crest Factor for OFDM Signals

On the Concentration of the Crest Factor for OFDM Signals On the Concentration of the Crest Factor for OFDM Signals Igal Sason Department of Electrical Engineering Technion - Israel Institute of Technology Haifa 32000, Israel The 2011 IEEE International Symposium

More information

Selected Exercises on Expectations and Some Probability Inequalities

Selected Exercises on Expectations and Some Probability Inequalities Selected Exercises on Expectations and Some Probability Inequalities # If E(X 2 ) = and E X a > 0, then P( X λa) ( λ) 2 a 2 for 0 < λ

More information

Hoeffding, Chernoff, Bennet, and Bernstein Bounds

Hoeffding, Chernoff, Bennet, and Bernstein Bounds Stat 928: Statistical Learning Theory Lecture: 6 Hoeffding, Chernoff, Bennet, Bernstein Bounds Instructor: Sham Kakade 1 Hoeffding s Bound We say X is a sub-gaussian rom variable if it has quadratically

More information

Introduction to Machine Learning CMU-10701

Introduction to Machine Learning CMU-10701 Introduction to Machine Learning CMU-10701 Stochastic Convergence Barnabás Póczos Motivation 2 What have we seen so far? Several algorithms that seem to work fine on training datasets: Linear regression

More information

Lecture Notes 3 Convergence (Chapter 5)

Lecture Notes 3 Convergence (Chapter 5) Lecture Notes 3 Convergence (Chapter 5) 1 Convergence of Random Variables Let X 1, X 2,... be a sequence of random variables and let X be another random variable. Let F n denote the cdf of X n and let

More information

AN INEQUALITY FOR TAIL PROBABILITIES OF MARTINGALES WITH BOUNDED DIFFERENCES

AN INEQUALITY FOR TAIL PROBABILITIES OF MARTINGALES WITH BOUNDED DIFFERENCES Lithuanian Mathematical Journal, Vol. 4, No. 3, 00 AN INEQUALITY FOR TAIL PROBABILITIES OF MARTINGALES WITH BOUNDED DIFFERENCES V. Bentkus Vilnius Institute of Mathematics and Informatics, Akademijos 4,

More information

18.175: Lecture 17 Poisson random variables

18.175: Lecture 17 Poisson random variables 18.175: Lecture 17 Poisson random variables Scott Sheffield MIT 1 Outline More on random walks and local CLT Poisson random variable convergence Extend CLT idea to stable random variables 2 Outline More

More information

Concentration of Measure with Applications in Information Theory, Communications and Coding

Concentration of Measure with Applications in Information Theory, Communications and Coding Concentration of Measure with Applications in Information Theory, Communications and Coding Maxim Raginsky UIUC Urbana, IL 61801, USA maxim@illinois.edu Igal Sason Technion Haifa 32000, Israel sason@ee.technion.ac.il

More information

Outline. Martingales. Piotr Wojciechowski 1. 1 Lane Department of Computer Science and Electrical Engineering West Virginia University.

Outline. Martingales. Piotr Wojciechowski 1. 1 Lane Department of Computer Science and Electrical Engineering West Virginia University. Outline Piotr 1 1 Lane Department of Computer Science and Electrical Engineering West Virginia University 8 April, 01 Outline Outline 1 Tail Inequalities Outline Outline 1 Tail Inequalities General Outline

More information

1. Stochastic Processes and filtrations

1. Stochastic Processes and filtrations 1. Stochastic Processes and 1. Stoch. pr., A stochastic process (X t ) t T is a collection of random variables on (Ω, F) with values in a measurable space (S, S), i.e., for all t, In our case X t : Ω S

More information

Lecture 4: September Reminder: convergence of sequences

Lecture 4: September Reminder: convergence of sequences 36-705: Intermediate Statistics Fall 2017 Lecturer: Siva Balakrishnan Lecture 4: September 6 In this lecture we discuss the convergence of random variables. At a high-level, our first few lectures focused

More information

Proving the central limit theorem

Proving the central limit theorem SOR3012: Stochastic Processes Proving the central limit theorem Gareth Tribello March 3, 2019 1 Purpose In the lectures and exercises we have learnt about the law of large numbers and the central limit

More information

Concentration Inequalities

Concentration Inequalities Chapter Concentration Inequalities I. Moment generating functions, the Chernoff method, and sub-gaussian and sub-exponential random variables a. Goal for this section: given a random variable X, how does

More information

MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 9 10/2/2013. Conditional expectations, filtration and martingales

MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 9 10/2/2013. Conditional expectations, filtration and martingales MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 9 10/2/2013 Conditional expectations, filtration and martingales Content. 1. Conditional expectations 2. Martingales, sub-martingales

More information

EXPONENTIAL INEQUALITIES IN NONPARAMETRIC ESTIMATION

EXPONENTIAL INEQUALITIES IN NONPARAMETRIC ESTIMATION EXPONENTIAL INEQUALITIES IN NONPARAMETRIC ESTIMATION Luc Devroye Division of Statistics University of California at Davis Davis, CA 95616 ABSTRACT We derive exponential inequalities for the oscillation

More information

Theory and Applications of Stochastic Systems Lecture Exponential Martingale for Random Walk

Theory and Applications of Stochastic Systems Lecture Exponential Martingale for Random Walk Instructor: Victor F. Araman December 4, 2003 Theory and Applications of Stochastic Systems Lecture 0 B60.432.0 Exponential Martingale for Random Walk Let (S n : n 0) be a random walk with i.i.d. increments

More information

Limiting Distributions

Limiting Distributions Limiting Distributions We introduce the mode of convergence for a sequence of random variables, and discuss the convergence in probability and in distribution. The concept of convergence leads us to the

More information

STA 711: Probability & Measure Theory Robert L. Wolpert

STA 711: Probability & Measure Theory Robert L. Wolpert STA 711: Probability & Measure Theory Robert L. Wolpert 6 Independence 6.1 Independent Events A collection of events {A i } F in a probability space (Ω,F,P) is called independent if P[ i I A i ] = P[A

More information

COMS 4771 Introduction to Machine Learning. Nakul Verma

COMS 4771 Introduction to Machine Learning. Nakul Verma COMS 4771 Introduction to Machine Learning Nakul Verma Announcements HW2 due now! Project proposal due on tomorrow Midterm next lecture! HW3 posted Last time Linear Regression Parametric vs Nonparametric

More information

Example continued. Math 425 Intro to Probability Lecture 37. Example continued. Example

Example continued. Math 425 Intro to Probability Lecture 37. Example continued. Example continued : Coin tossing Math 425 Intro to Probability Lecture 37 Kenneth Harris kaharri@umich.edu Department of Mathematics University of Michigan April 8, 2009 Consider a Bernoulli trials process with

More information

18.175: Lecture 3 Integration

18.175: Lecture 3 Integration 18.175: Lecture 3 Scott Sheffield MIT Outline Outline Recall definitions Probability space is triple (Ω, F, P) where Ω is sample space, F is set of events (the σ-algebra) and P : F [0, 1] is the probability

More information

Lecture 4. P r[x > ce[x]] 1/c. = ap r[x = a] + a>ce[x] P r[x = a]

Lecture 4. P r[x > ce[x]] 1/c. = ap r[x = a] + a>ce[x] P r[x = a] U.C. Berkeley CS273: Parallel and Distributed Theory Lecture 4 Professor Satish Rao September 7, 2010 Lecturer: Satish Rao Last revised September 13, 2010 Lecture 4 1 Deviation bounds. Deviation bounds

More information

Chapter 3 sections. SKIP: 3.10 Markov Chains. SKIP: pages Chapter 3 - continued

Chapter 3 sections. SKIP: 3.10 Markov Chains. SKIP: pages Chapter 3 - continued Chapter 3 sections 3.1 Random Variables and Discrete Distributions 3.2 Continuous Distributions 3.3 The Cumulative Distribution Function 3.4 Bivariate Distributions 3.5 Marginal Distributions 3.6 Conditional

More information

Computer Intensive Methods in Mathematical Statistics

Computer Intensive Methods in Mathematical Statistics Computer Intensive Methods in Mathematical Statistics Department of mathematics johawes@kth.se Lecture 5 Sequential Monte Carlo methods I 31 March 2017 Computer Intensive Methods (1) Plan of today s lecture

More information

Lecture Notes 5 Convergence and Limit Theorems. Convergence with Probability 1. Convergence in Mean Square. Convergence in Probability, WLLN

Lecture Notes 5 Convergence and Limit Theorems. Convergence with Probability 1. Convergence in Mean Square. Convergence in Probability, WLLN Lecture Notes 5 Convergence and Limit Theorems Motivation Convergence with Probability Convergence in Mean Square Convergence in Probability, WLLN Convergence in Distribution, CLT EE 278: Convergence and

More information

Disjointness and Additivity

Disjointness and Additivity Midterm 2: Format Midterm 2 Review CS70 Summer 2016 - Lecture 6D David Dinh 28 July 2016 UC Berkeley 8 questions, 190 points, 110 minutes (same as MT1). Two pages (one double-sided sheet) of handwritten

More information

Chapter 3 sections. SKIP: 3.10 Markov Chains. SKIP: pages Chapter 3 - continued

Chapter 3 sections. SKIP: 3.10 Markov Chains. SKIP: pages Chapter 3 - continued Chapter 3 sections Chapter 3 - continued 3.1 Random Variables and Discrete Distributions 3.2 Continuous Distributions 3.3 The Cumulative Distribution Function 3.4 Bivariate Distributions 3.5 Marginal Distributions

More information

Midterm 2 Review. CS70 Summer Lecture 6D. David Dinh 28 July UC Berkeley

Midterm 2 Review. CS70 Summer Lecture 6D. David Dinh 28 July UC Berkeley Midterm 2 Review CS70 Summer 2016 - Lecture 6D David Dinh 28 July 2016 UC Berkeley Midterm 2: Format 8 questions, 190 points, 110 minutes (same as MT1). Two pages (one double-sided sheet) of handwritten

More information

Chapter 6: Large Random Samples Sections

Chapter 6: Large Random Samples Sections Chapter 6: Large Random Samples Sections 6.1: Introduction 6.2: The Law of Large Numbers Skip p. 356-358 Skip p. 366-368 Skip 6.4: The correction for continuity Remember: The Midterm is October 25th in

More information

Probability inequalities 11

Probability inequalities 11 Paninski, Intro. Math. Stats., October 5, 2005 29 Probability inequalities 11 There is an adage in probability that says that behind every limit theorem lies a probability inequality (i.e., a bound on

More information

Bennett-type Generalization Bounds: Large-deviation Case and Faster Rate of Convergence

Bennett-type Generalization Bounds: Large-deviation Case and Faster Rate of Convergence Bennett-type Generalization Bounds: Large-deviation Case and Faster Rate of Convergence Chao Zhang The Biodesign Institute Arizona State University Tempe, AZ 8587, USA Abstract In this paper, we present

More information

Lecture 11. Probability Theory: an Overveiw

Lecture 11. Probability Theory: an Overveiw Math 408 - Mathematical Statistics Lecture 11. Probability Theory: an Overveiw February 11, 2013 Konstantin Zuev (USC) Math 408, Lecture 11 February 11, 2013 1 / 24 The starting point in developing the

More information

Essentials on the Analysis of Randomized Algorithms

Essentials on the Analysis of Randomized Algorithms Essentials on the Analysis of Randomized Algorithms Dimitris Diochnos Feb 0, 2009 Abstract These notes were written with Monte Carlo algorithms primarily in mind. Topics covered are basic (discrete) random

More information

Lecture 1: August 28

Lecture 1: August 28 36-705: Intermediate Statistics Fall 2017 Lecturer: Siva Balakrishnan Lecture 1: August 28 Our broad goal for the first few lectures is to try to understand the behaviour of sums of independent random

More information

Chapter 7. Basic Probability Theory

Chapter 7. Basic Probability Theory Chapter 7. Basic Probability Theory I-Liang Chern October 20, 2016 1 / 49 What s kind of matrices satisfying RIP Random matrices with iid Gaussian entries iid Bernoulli entries (+/ 1) iid subgaussian entries

More information

Stat 260/CS Learning in Sequential Decision Problems. Peter Bartlett

Stat 260/CS Learning in Sequential Decision Problems. Peter Bartlett Stat 260/CS 294-102. Learning in Sequential Decision Problems. Peter Bartlett 1. Multi-armed bandit algorithms. Concentration inequalities. P(X ǫ) exp( ψ (ǫ))). Cumulant generating function bounds. Hoeffding

More information

The Moment Method; Convex Duality; and Large/Medium/Small Deviations

The Moment Method; Convex Duality; and Large/Medium/Small Deviations Stat 928: Statistical Learning Theory Lecture: 5 The Moment Method; Convex Duality; and Large/Medium/Small Deviations Instructor: Sham Kakade The Exponential Inequality and Convex Duality The exponential

More information

Probability Theory I: Syllabus and Exercise

Probability Theory I: Syllabus and Exercise Probability Theory I: Syllabus and Exercise Narn-Rueih Shieh **Copyright Reserved** This course is suitable for those who have taken Basic Probability; some knowledge of Real Analysis is recommended( will

More information

Mathematical Statistics

Mathematical Statistics Mathematical Statistics Chapter Three. Point Estimation 3.4 Uniformly Minimum Variance Unbiased Estimator(UMVUE) Criteria for Best Estimators MSE Criterion Let F = {p(x; θ) : θ Θ} be a parametric distribution

More information

Uniform concentration inequalities, martingales, Rademacher complexity and symmetrization

Uniform concentration inequalities, martingales, Rademacher complexity and symmetrization Uniform concentration inequalities, martingales, Rademacher complexity and symmetrization John Duchi Outline I Motivation 1 Uniform laws of large numbers 2 Loss minimization and data dependence II Uniform

More information

Stochastic Optimization

Stochastic Optimization Introduction Related Work SGD Epoch-GD LM A DA NANJING UNIVERSITY Lijun Zhang Nanjing University, China May 26, 2017 Introduction Related Work SGD Epoch-GD Outline 1 Introduction 2 Related Work 3 Stochastic

More information

Advanced Probability Theory (Math541)

Advanced Probability Theory (Math541) Advanced Probability Theory (Math541) Instructor: Kani Chen (Classic)/Modern Probability Theory (1900-1960) Instructor: Kani Chen (HKUST) Advanced Probability Theory (Math541) 1 / 17 Primitive/Classic

More information

Lecture 3: Expected Value. These integrals are taken over all of Ω. If we wish to integrate over a measurable subset A Ω, we will write

Lecture 3: Expected Value. These integrals are taken over all of Ω. If we wish to integrate over a measurable subset A Ω, we will write Lecture 3: Expected Value 1.) Definitions. If X 0 is a random variable on (Ω, F, P), then we define its expected value to be EX = XdP. Notice that this quantity may be. For general X, we say that EX exists

More information

Modern Discrete Probability Branching processes

Modern Discrete Probability Branching processes Modern Discrete Probability IV - Branching processes Review Sébastien Roch UW Madison Mathematics November 15, 2014 1 Basic definitions 2 3 4 Galton-Watson branching processes I Definition A Galton-Watson

More information

Random Process Lecture 1. Fundamentals of Probability

Random Process Lecture 1. Fundamentals of Probability Random Process Lecture 1. Fundamentals of Probability Husheng Li Min Kao Department of Electrical Engineering and Computer Science University of Tennessee, Knoxville Spring, 2016 1/43 Outline 2/43 1 Syllabus

More information

Brownian motion. Samy Tindel. Purdue University. Probability Theory 2 - MA 539

Brownian motion. Samy Tindel. Purdue University. Probability Theory 2 - MA 539 Brownian motion Samy Tindel Purdue University Probability Theory 2 - MA 539 Mostly taken from Brownian Motion and Stochastic Calculus by I. Karatzas and S. Shreve Samy T. Brownian motion Probability Theory

More information

Appendix B: Inequalities Involving Random Variables and Their Expectations

Appendix B: Inequalities Involving Random Variables and Their Expectations Chapter Fourteen Appendix B: Inequalities Involving Random Variables and Their Expectations In this appendix we present specific properties of the expectation (additional to just the integral of measurable

More information

4 Expectation & the Lebesgue Theorems

4 Expectation & the Lebesgue Theorems STA 205: Probability & Measure Theory Robert L. Wolpert 4 Expectation & the Lebesgue Theorems Let X and {X n : n N} be random variables on a probability space (Ω,F,P). If X n (ω) X(ω) for each ω Ω, does

More information

Matrix concentration inequalities

Matrix concentration inequalities ELE 538B: Mathematics of High-Dimensional Data Matrix concentration inequalities Yuxin Chen Princeton University, Fall 2018 Recap: matrix Bernstein inequality Consider a sequence of independent random

More information

CSE 312 Final Review: Section AA

CSE 312 Final Review: Section AA CSE 312 TAs December 8, 2011 General Information General Information Comprehensive Midterm General Information Comprehensive Midterm Heavily weighted toward material after the midterm Pre-Midterm Material

More information

Lecture 2: Review of Basic Probability Theory

Lecture 2: Review of Basic Probability Theory ECE 830 Fall 2010 Statistical Signal Processing instructor: R. Nowak, scribe: R. Nowak Lecture 2: Review of Basic Probability Theory Probabilistic models will be used throughout the course to represent

More information

Part II Probability and Measure

Part II Probability and Measure Part II Probability and Measure Theorems Based on lectures by J. Miller Notes taken by Dexter Chua Michaelmas 2016 These notes are not endorsed by the lecturers, and I have modified them (often significantly)

More information

Part IA Probability. Theorems. Based on lectures by R. Weber Notes taken by Dexter Chua. Lent 2015

Part IA Probability. Theorems. Based on lectures by R. Weber Notes taken by Dexter Chua. Lent 2015 Part IA Probability Theorems Based on lectures by R. Weber Notes taken by Dexter Chua Lent 2015 These notes are not endorsed by the lecturers, and I have modified them (often significantly) after lectures.

More information

CSE 525 Randomized Algorithms & Probabilistic Analysis Spring Lecture 3: April 9

CSE 525 Randomized Algorithms & Probabilistic Analysis Spring Lecture 3: April 9 CSE 55 Randomized Algorithms & Probabilistic Analysis Spring 01 Lecture : April 9 Lecturer: Anna Karlin Scribe: Tyler Rigsby & John MacKinnon.1 Kinds of randomization in algorithms So far in our discussion

More information

18.175: Lecture 13 Infinite divisibility and Lévy processes

18.175: Lecture 13 Infinite divisibility and Lévy processes 18.175 Lecture 13 18.175: Lecture 13 Infinite divisibility and Lévy processes Scott Sheffield MIT Outline Poisson random variable convergence Extend CLT idea to stable random variables Infinite divisibility

More information

Let (Ω, F) be a measureable space. A filtration in discrete time is a sequence of. F s F t

Let (Ω, F) be a measureable space. A filtration in discrete time is a sequence of. F s F t 2.2 Filtrations Let (Ω, F) be a measureable space. A filtration in discrete time is a sequence of σ algebras {F t } such that F t F and F t F t+1 for all t = 0, 1,.... In continuous time, the second condition

More information

On Concentration of Martingales and Applications in Information Theory, Communication & Coding

On Concentration of Martingales and Applications in Information Theory, Communication & Coding On Concentration of Martingales and Applications in Information Theory, Communication & Coding Igal Sason Department of Electrical Engineering Technion - Israel Institute of Technology Haifa 32000, Israel

More information

Probability Background

Probability Background Probability Background Namrata Vaswani, Iowa State University August 24, 2015 Probability recap 1: EE 322 notes Quick test of concepts: Given random variables X 1, X 2,... X n. Compute the PDF of the second

More information

Lecture 11 : Asymptotic Sample Complexity

Lecture 11 : Asymptotic Sample Complexity Lecture 11 : Asymptotic Sample Complexity MATH285K - Spring 2010 Lecturer: Sebastien Roch References: [DMR09]. Previous class THM 11.1 (Strong Quartet Evidence) Let Q be a collection of quartet trees on

More information

Transforms. Convergence of probability generating functions. Convergence of characteristic functions functions

Transforms. Convergence of probability generating functions. Convergence of characteristic functions functions Transforms For non-negative integer value ranom variables, let the probability generating function g X : [0, 1] [0, 1] be efine by g X (t) = E(t X ). The moment generating function ψ X (t) = E(e tx ) is

More information

Lecture 5: Importance sampling and Hamilton-Jacobi equations

Lecture 5: Importance sampling and Hamilton-Jacobi equations Lecture 5: Importance sampling and Hamilton-Jacobi equations Henrik Hult Department of Mathematics KTH Royal Institute of Technology Sweden Summer School on Monte Carlo Methods and Rare Events Brown University,

More information

X = X X n, + X 2

X = X X n, + X 2 CS 70 Discrete Mathematics for CS Fall 2003 Wagner Lecture 22 Variance Question: At each time step, I flip a fair coin. If it comes up Heads, I walk one step to the right; if it comes up Tails, I walk

More information

Stochastic Models (Lecture #4)

Stochastic Models (Lecture #4) Stochastic Models (Lecture #4) Thomas Verdebout Université libre de Bruxelles (ULB) Today Today, our goal will be to discuss limits of sequences of rv, and to study famous limiting results. Convergence

More information

6.1 Moment Generating and Characteristic Functions

6.1 Moment Generating and Characteristic Functions Chapter 6 Limit Theorems The power statistics can mostly be seen when there is a large collection of data points and we are interested in understanding the macro state of the system, e.g., the average,

More information

3. Review of Probability and Statistics

3. Review of Probability and Statistics 3. Review of Probability and Statistics ECE 830, Spring 2014 Probabilistic models will be used throughout the course to represent noise, errors, and uncertainty in signal processing problems. This lecture

More information

18.175: Lecture 15 Characteristic functions and central limit theorem

18.175: Lecture 15 Characteristic functions and central limit theorem 18.175: Lecture 15 Characteristic functions and central limit theorem Scott Sheffield MIT Outline Characteristic functions Outline Characteristic functions Characteristic functions Let X be a random variable.

More information

Entropy and Ergodic Theory Lecture 15: A first look at concentration

Entropy and Ergodic Theory Lecture 15: A first look at concentration Entropy and Ergodic Theory Lecture 15: A first look at concentration 1 Introduction to concentration Let X 1, X 2,... be i.i.d. R-valued RVs with common distribution µ, and suppose for simplicity that

More information

Ergodic Theorems. Samy Tindel. Purdue University. Probability Theory 2 - MA 539. Taken from Probability: Theory and examples by R.

Ergodic Theorems. Samy Tindel. Purdue University. Probability Theory 2 - MA 539. Taken from Probability: Theory and examples by R. Ergodic Theorems Samy Tindel Purdue University Probability Theory 2 - MA 539 Taken from Probability: Theory and examples by R. Durrett Samy T. Ergodic theorems Probability Theory 1 / 92 Outline 1 Definitions

More information

On the convergence of sequences of random variables: A primer

On the convergence of sequences of random variables: A primer BCAM May 2012 1 On the convergence of sequences of random variables: A primer Armand M. Makowski ECE & ISR/HyNet University of Maryland at College Park armand@isr.umd.edu BCAM May 2012 2 A sequence a :

More information

Point Process Control

Point Process Control Point Process Control The following note is based on Chapters I, II and VII in Brémaud s book Point Processes and Queues (1981). 1 Basic Definitions Consider some probability space (Ω, F, P). A real-valued

More information

Introduction to Self-normalized Limit Theory

Introduction to Self-normalized Limit Theory Introduction to Self-normalized Limit Theory Qi-Man Shao The Chinese University of Hong Kong E-mail: qmshao@cuhk.edu.hk Outline What is the self-normalization? Why? Classical limit theorems Self-normalized

More information

Limiting Distributions

Limiting Distributions We introduce the mode of convergence for a sequence of random variables, and discuss the convergence in probability and in distribution. The concept of convergence leads us to the two fundamental results

More information

Mod-φ convergence I: examples and probabilistic estimates

Mod-φ convergence I: examples and probabilistic estimates Mod-φ convergence I: examples and probabilistic estimates Valentin Féray (joint work with Pierre-Loïc Méliot and Ashkan Nikeghbali) Institut für Mathematik, Universität Zürich Summer school in Villa Volpi,

More information

Anti-concentration Inequalities

Anti-concentration Inequalities Anti-concentration Inequalities Roman Vershynin Mark Rudelson University of California, Davis University of Missouri-Columbia Phenomena in High Dimensions Third Annual Conference Samos, Greece June 2007

More information

{X i } realize. n i=1 X i. Note that again X is a random variable. If we are to

{X i } realize. n i=1 X i. Note that again X is a random variable. If we are to 3 Convergence This topic will overview a variety of extremely powerful analysis results that span statistics, estimation theorem, and big data. It provides a framework to think about how to aggregate more

More information

6.207/14.15: Networks Lecture 3: Erdös-Renyi graphs and Branching processes

6.207/14.15: Networks Lecture 3: Erdös-Renyi graphs and Branching processes 6.207/14.15: Networks Lecture 3: Erdös-Renyi graphs and Branching processes Daron Acemoglu and Asu Ozdaglar MIT September 16, 2009 1 Outline Erdös-Renyi random graph model Branching processes Phase transitions

More information

Computational and Statistical Learning Theory

Computational and Statistical Learning Theory Computational and Statistical Learning Theory Problem set 1 Due: Monday, October 10th Please send your solutions to learning-submissions@ttic.edu Notation: Input space: X Label space: Y = {±1} Sample:

More information

On the Concentration of the Crest Factor for OFDM Signals

On the Concentration of the Crest Factor for OFDM Signals On the Concentration of the Crest Factor for OFDM Signals Igal Sason Department of Electrical Engineering Technion - Israel Institute of Technology, Haifa 3, Israel E-mail: sason@eetechnionacil Abstract

More information

Part IA Probability. Definitions. Based on lectures by R. Weber Notes taken by Dexter Chua. Lent 2015

Part IA Probability. Definitions. Based on lectures by R. Weber Notes taken by Dexter Chua. Lent 2015 Part IA Probability Definitions Based on lectures by R. Weber Notes taken by Dexter Chua Lent 2015 These notes are not endorsed by the lecturers, and I have modified them (often significantly) after lectures.

More information

Kousha Etessami. U. of Edinburgh, UK. Kousha Etessami (U. of Edinburgh, UK) Discrete Mathematics (Chapter 7) 1 / 13

Kousha Etessami. U. of Edinburgh, UK. Kousha Etessami (U. of Edinburgh, UK) Discrete Mathematics (Chapter 7) 1 / 13 Discrete Mathematics & Mathematical Reasoning Chapter 7 (continued): Markov and Chebyshev s Inequalities; and Examples in probability: the birthday problem Kousha Etessami U. of Edinburgh, UK Kousha Etessami

More information

Math 6810 (Probability) Fall Lecture notes

Math 6810 (Probability) Fall Lecture notes Math 6810 (Probability) Fall 2012 Lecture notes Pieter Allaart University of North Texas September 23, 2012 2 Text: Introduction to Stochastic Calculus with Applications, by Fima C. Klebaner (3rd edition),

More information

Concentration of Measures by Bounded Size Bias Couplings

Concentration of Measures by Bounded Size Bias Couplings Concentration of Measures by Bounded Size Bias Couplings Subhankar Ghosh, Larry Goldstein University of Southern California [arxiv:0906.3886] January 10 th, 2013 Concentration of Measure Distributional

More information

Lecture 7: Chapter 7. Sums of Random Variables and Long-Term Averages

Lecture 7: Chapter 7. Sums of Random Variables and Long-Term Averages Lecture 7: Chapter 7. Sums of Random Variables and Long-Term Averages ELEC206 Probability and Random Processes, Fall 2014 Gil-Jin Jang gjang@knu.ac.kr School of EE, KNU page 1 / 15 Chapter 7. Sums of Random

More information

Exercises Measure Theoretic Probability

Exercises Measure Theoretic Probability Exercises Measure Theoretic Probability 2002-2003 Week 1 1. Prove the folloing statements. (a) The intersection of an arbitrary family of d-systems is again a d- system. (b) The intersection of an arbitrary

More information

Lithuanian Mathematical Journal, 2006, No 1

Lithuanian Mathematical Journal, 2006, No 1 ON DOMINATION OF TAIL PROBABILITIES OF (SUPER)MARTINGALES: EXPLICIT BOUNDS V. Bentkus, 1,3 N. Kalosha, 2,3 M. van Zuijlen 2,3 Lithuanian Mathematical Journal, 2006, No 1 Abstract. Let X be a random variable

More information

Filtrations, Markov Processes and Martingales. Lectures on Lévy Processes and Stochastic Calculus, Braunschweig, Lecture 3: The Lévy-Itô Decomposition

Filtrations, Markov Processes and Martingales. Lectures on Lévy Processes and Stochastic Calculus, Braunschweig, Lecture 3: The Lévy-Itô Decomposition Filtrations, Markov Processes and Martingales Lectures on Lévy Processes and Stochastic Calculus, Braunschweig, Lecture 3: The Lévy-Itô Decomposition David pplebaum Probability and Statistics Department,

More information

Concentration function and other stuff

Concentration function and other stuff Concentration function and other stuff Sabrina Sixta Tuesday, June 16, 2014 Sabrina Sixta () Concentration function and other stuff Tuesday, June 16, 2014 1 / 13 Table of Contents Outline 1 Chernoff Bound

More information

1 Review of The Learning Setting

1 Review of The Learning Setting COS 5: Theoretical Machine Learning Lecturer: Rob Schapire Lecture #8 Scribe: Changyan Wang February 28, 208 Review of The Learning Setting Last class, we moved beyond the PAC model: in the PAC model we

More information