Amplification of stochastic advantage

Slides:



Advertisements
Similar presentations
Lecture 18: Temporal-Difference Learning
Advertisements

CSCI 3160 Design and Analysis of Algorithms Tutorial 4
General Statistics Ch En 475 Unit Operations. Quantifying variables (i.e. answering a question with a number) 1. Directly measure the variable. - referred.
Week11 Parameter, Statistic and Random Samples A parameter is a number that describes the population. It is a fixed number, but in practice we do not know.
POINT ESTIMATION AND INTERVAL ESTIMATION
Ch. 8 – Practical Examples of Confidence Intervals for z, t, p.
Randomized Algorithms Kyomin Jung KAIST Applied Algorithm Lab Jan 12, WSAC
Probability theory and average-case complexity. Review of probability theory.
Evaluation (practice). 2 Predicting performance  Assume the estimated error rate is 25%. How close is this to the true error rate?  Depends on the amount.
Evaluation.
Importance Sampling. What is Importance Sampling ? A simulation technique Used when we are interested in rare events Examples: Bit Error Rate on a channel,
Part 2b Parameter Estimation CSE717, FALL 2008 CUBS, Univ at Buffalo.
Evaluation.
Evaluating Hypotheses
Study Group Randomized Algorithms Jun 7, 2003 Jun 14, 2003.
Lehrstuhl für Informatik 2 Gabriella Kókai: Maschine Learning 1 Evaluating Hypotheses.
A) Transformation method (for continuous distributions) U(0,1) : uniform distribution f(x) : arbitrary distribution f(x) dx = U(0,1)(u) du When inverse.
Ka-fu Wong © 2007 ECON1003: Analysis of Economic Data Supplement8-1 Additional real-life examples (proportions) Supplement 8: Additional real-life examples.
The Binomial Distribution. Introduction # correct TallyFrequencyP(experiment)P(theory) Mix the cards, select one & guess the type. Repeat 3 times.
Chapter 7 Estimation: Single Population
Introduction to Data Analysis Probability Distributions.
Sullivan – Fundamentals of Statistics – 2 nd Edition – Chapter 9 Section 1 – Slide 1 of 39 Chapter 9 Section 1 The Logic in Constructing Confidence Intervals.
STA Lecture 161 STA 291 Lecture 16 Normal distributions: ( mean and SD ) use table or web page. The sampling distribution of and are both (approximately)
Populations, Samples, Standard errors, confidence intervals Dr. Omar Al Jadaan.
Confidence Intervals 1 Chapter 6. Chapter Outline Confidence Intervals for the Mean (Large Samples) 6.2 Confidence Intervals for the Mean (Small.
Chap 6-1 A Course In Business Statistics, 4th © 2006 Prentice-Hall, Inc. A Course In Business Statistics 4 th Edition Chapter 6 Introduction to Sampling.
Prabhas Chongstitvatana 1 Primality Testing Is a given odd integer prime or composite ? No known algorithm can solve this problem with certainty in a reasonable.
Chapter 5 Discrete Probability Distributions
Chapter 7 Point Estimation
10.1: Confidence Intervals – The Basics. Review Question!!! If the mean and the standard deviation of a continuous random variable that is normally distributed.
Estimation (Point Estimation)
General Statistics Ch En 475 Unit Operations. Quantifying variables (i.e. answering a question with a number) 1. Directly measure the variable. - referred.
Sampling Theory The procedure for drawing a random sample a distribution is that numbers 1, 2, … are assigned to the elements of the distribution and tables.
Statistics Presentation Ch En 475 Unit Operations.
Machine Learning Chapter 5. Evaluating Hypotheses
Prabhas Chongstitvatana1 Las Vegas algorithm The answer obtained is always correct but sometime no answer. Modified deterministic algorithm by using randomness.
The Markov Chain Monte Carlo Method Isabelle Stanton May 8, 2008 Theory Lunch.
Chapter 8: Confidence Intervals based on a Single Sample
Prabhas Chongstitvatana1 Monte Carlo integration It is a numerical probabilistic algorithm ab I/(b-a) f.
1 Mean Analysis. 2 Introduction l If we use sample mean (the mean of the sample) to approximate the population mean (the mean of the population), errors.
Section 5.2 Binomial Probabilities. 2 Features of a Binomial Experiment 1.There are a fixed number of trials, n 2.The n trials are independent and repeated.
ICS 353: Design and Analysis of Algorithms
Introduction to Probability – Experimental Probability.
Prabhas Chongstitvatana1 Numerical probabilistic The answer is always an approximation.
Chapter Outline 6.1 Confidence Intervals for the Mean (Large Samples) 6.2 Confidence Intervals for the Mean (Small Samples) 6.3 Confidence Intervals for.
Chapter 6 Large Random Samples Weiqi Luo ( 骆伟祺 ) School of Data & Computer Science Sun Yat-Sen University :
Ex St 801 Statistical Methods Inference about a Single Population Mean (CI)
Chapter 8 Estimation ©. Estimator and Estimate estimator estimate An estimator of a population parameter is a random variable that depends on the sample.
Evaluating Hypotheses. Outline Empirically evaluating the accuracy of hypotheses is fundamental to machine learning – How well does this estimate its.
Rodney Nielsen Many of these slides were adapted from: I. H. Witten, E. Frank and M. A. Hall Data Science Credibility: Evaluating What’s Been Learned Predicting.
Random Variables Lecture Lecturer : FATEN AL-HUSSAIN.
1. 2 At the end of the lesson, students will be able to (c)Understand the Binomial distribution B(n,p) (d) find the mean and variance of Binomial distribution.
© 2001 Prentice-Hall, Inc.Chap 8-1 BA 201 Lecture 12 Confidence Interval Estimation.
Evaluating Hypotheses. Outline Empirically evaluating the accuracy of hypotheses is fundamental to machine learning – How well does this estimate accuracy.
Chapter 19 Monte Carlo Valuation. © 2013 Pearson Education, Inc., publishing as Prentice Hall. All rights reserved.19-2 Monte Carlo Valuation Simulation.
usually unimportant in social surveys:
Statistical Estimation
Lecture 8 Randomized Algorithms
Prabhas Chongstitvatana
Numerical probabilistic
Chapter 6 Confidence Intervals.
Evaluating Hypotheses
Amplification of stochastic advantage
Sampling Distributions
Probabilistic algorithms
Monte Carlo integration
Probabilistic algorithms
Chapter 6 Confidence Intervals.
Chapter 8 Estimation.
Maximum Likelihood Estimation (MLE)
Presentation transcript:

Amplification of stochastic advantage Biased : known with certainty one of the possible answer is always correct. Error can be reduced by repeat the algorithm. Unbiased example coin flip for p-correct “advantage” is p - 1/2 Prabhas Chongstitvatana

Prabhas Chongstitvatana Let MC be a 3/4-correct unbiased What is the error prob. Of MC3 (mojority vote) ? MC3 is 27/32-correct ( > 84% ) Prabhas Chongstitvatana

Prabhas Chongstitvatana What is the error prob. of MC with advantage e > 0 in majority vote k times ? Let Xi = 1 if correct answer, 0 otherwise Pr[ Xi = 1 ] >= 1/2 + e assume for simplicity Pr[ Xi = 1 ] = 1/2 + e ; k is odd (no tie) E( Xi ) = 1/2 + e; Var(Xi ) = (1/2 + e) (1/2-e) = 1/4 - e2 Prabhas Chongstitvatana

Prabhas Chongstitvatana Let X is a random variable corresponds to the number of correct answer is k trials. E(X) = (1/2+e)k Var(X) = (1/4 - e2) k Prabhas Chongstitvatana

Prabhas Chongstitvatana error prob. Pr [ X <= k/2 ] can be calculated is normal distributed if k >= 30 Prabhas Chongstitvatana

Prabhas Chongstitvatana If need error < 5% Pr [ X < E(X) - 1.645 sqrt Var(X) ] ~ 5% (from the table of normal distribution) Pr [ X <= k/2 ] < 5% if k/2 < E(X) - 1.645 sqrt Var(X) k > 2.706 ( 1/(4e2) - 1 ) Prabhas Chongstitvatana

Prabhas Chongstitvatana Example e = 5%, which is 55%-correct unbiased Monte Carlo, k > 2.706 ( 1/(4e2) - 1 ) k > 267.894, majority vote repeat 269 times to obtain 95%-correct. Repetition turn 5% advantage into 5% error prob Prabhas Chongstitvatana

Prabhas Chongstitvatana It takes a large number of run for unbiased Compare to biased Run 55%-correct bias MC 4 times reduces the error prob. To 0.454 ~ 0.041 ( 4.1% ) Prabhas Chongstitvatana

Prabhas Chongstitvatana Not much more expensive to obtain more confidence. If we want 0.5% confidence (10 times more) Pr[X < E(X) - 2.576 sqrt Var(X) ] ~ 5% k > 6.636 (1/(4e2) - 1 ) This makes it 99.5%-correct with less than 2.5 times more expensive than 95%-correct. Prabhas Chongstitvatana

Prabhas Chongstitvatana To reduce error prob < del for an unbiased Monte Carlo with advantage e; the number of repetition is proportional to 1/e2, also to log 1/del Prabhas Chongstitvatana