The Multiplicative Weights Update Method Based on Arora, Hazan & Kale (2005) Mashor Housh Oded Cats Advanced simulation methods Prof. Rubinstein.

Slides:



Advertisements
Similar presentations
Algorithm Design Methods (I) Fall 2003 CSE, POSTECH.
Advertisements

Algorithm Design Methods Spring 2007 CSE, POSTECH.
Primal Dual Combinatorial Algorithms Qihui Zhu May 11, 2009.
1 LP Duality Lecture 13: Feb Min-Max Theorems In bipartite graph, Maximum matching = Minimum Vertex Cover In every graph, Maximum Flow = Minimum.
On-line learning and Boosting
Totally Unimodular Matrices
Lecturer: Moni Naor Algorithmic Game Theory Uri Feige Robi Krauthgamer Moni Naor Lecture 8: Regret Minimization.
APPENDIX An Alternative View of the Payoff Matrix n Assume total maximum profits of all oligopolists is constant at 200 units. n Alternative policies.
Introduction to Algorithms
1 EE5900 Advanced Embedded System For Smart Infrastructure Static Scheduling.
The Randomization Repertoire Rajmohan Rajaraman Northeastern University, Boston May 2012 Chennai Network Optimization WorkshopThe Randomization Repertoire1.
1 Chapter 4: Minimax Equilibrium in Zero Sum Game SCIT1003 Chapter 4: Minimax Equilibrium in Zero Sum Game Prof. Tsang.
MIT and James Orlin © Game Theory 2-person 0-sum (or constant sum) game theory 2-person game theory (e.g., prisoner’s dilemma)
Study Group Randomized Algorithms 21 st June 03. Topics Covered Game Tree Evaluation –its expected run time is better than the worst- case complexity.
Techniques for Dealing with Hard Problems Backtrack: –Systematically enumerates all potential solutions by continually trying to extend a partial solution.
How Bad is Selfish Routing? By Tim Roughgarden Eva Tardos Presented by Alex Kogan.
Online learning, minimizing regret, and combining expert advice
Part 3: The Minimax Theorem
Games of Prediction or Things get simpler as Yoav Freund Banter Inc.
1 EL736 Communications Networks II: Design and Algorithms Class8: Networks with Shortest-Path Routing Yong Liu 10/31/2007.
Oblivious Routing for the L p -norm Matthias Englert Harald Räcke 1.
Lecture 1 - Introduction 1.  Introduction to Game Theory  Basic Game Theory Examples  Strategic Games  More Game Theory Examples  Equilibrium  Mixed.
Totally Unimodular Matrices Lecture 11: Feb 23 Simplex Algorithm Elliposid Algorithm.
Duality Lecture 10: Feb 9. Min-Max theorems In bipartite graph, Maximum matching = Minimum Vertex Cover In every graph, Maximum Flow = Minimum Cut Both.
1 Computing Nash Equilibrium Presenter: Yishay Mansour.
Implicit Hitting Set Problems Richard M. Karp Harvard University August 29, 2011.
An Introduction to Game Theory Part III: Strictly Competitive Games Bernhard Nebel.
Ensemble Learning: An Introduction
UNIT II: The Basic Theory Zero-sum Games Nonzero-sum Games Nash Equilibrium: Properties and Problems Bargaining Games Bargaining and Negotiation Review.
Job Scheduling Lecture 19: March 19. Job Scheduling: Unrelated Multiple Machines There are n jobs, each job has: a processing time p(i,j) (the time to.
Introduction to Boosting Aristotelis Tsirigos SCLT seminar - NYU Computer Science.
Finite Mathematics & Its Applications, 10/e by Goldstein/Schneider/SiegelCopyright © 2010 Pearson Education, Inc. 1 of 68 Chapter 9 The Theory of Games.
Algorithms and Economics of Networks Abraham Flaxman and Vahab Mirrokni, Microsoft Research.
Linear Programming – Max Flow – Min Cut Orgad Keller.
Machine Learning: Ensemble Methods
Game Theory.
Experts and Boosting Algorithms. Experts: Motivation Given a set of experts –No prior information –No consistent behavior –Goal: Predict as the best expert.
UNIT II: The Basic Theory Zero-sum Games Nonzero-sum Games Nash Equilibrium: Properties and Problems Bargaining Games Bargaining and Negotiation Review.
Approximation Algorithms: Bristol Summer School 2008 Seffi Naor Computer Science Dept. Technion Haifa, Israel TexPoint fonts used in EMF. Read the TexPoint.
An Intro to Game Theory Avrim Blum 12/07/04.
Experts Learning and The Minimax Theorem for Zero-Sum Games Maria Florina Balcan December 8th 2011.
Game Theory.
1 Lecture 4 Maximal Flow Problems Set Covering Problems.
Multiplicative Weights Algorithms CompSci Instructor: Ashwin Machanavajjhala 1Lecture 13 : Fall 12.
CPS 170: Artificial Intelligence Game Theory Instructor: Vincent Conitzer.
MAKING COMPLEX DEClSlONS
Design Techniques for Approximation Algorithms and Approximation Classes.
1 Introduction to Approximation Algorithms. 2 NP-completeness Do your best then.
1 Multiplicative Weights Update Method Boaz Kaminer Andrey Dolgin Based on: Aurora S., Hazan E. and Kale S., “The Multiplicative Weights Update Method:
Benk Erika Kelemen Zsolt
Game theory & Linear Programming Steve Gu Mar 28, 2008.
Monte Carlo Methods Versatile methods for analyzing the behavior of some activity, plan or process that involves uncertainty.
Algorithmic Game Theory and Internet Computing Vijay V. Vazirani Georgia Tech Primal-Dual Algorithms for Rational Convex Programs II: Dealing with Infeasibility.
1 What is Game Theory About? r Analysis of situations where conflict of interests is present r Goal is to prescribe how conflicts can be resolved 2 2 r.
Strategic Behavior in Business and Econ Static Games of complete information: Dominant Strategies and Nash Equilibrium in pure and mixed strategies.
Shall we play a game? Game Theory and Computer Science Game Theory /06/05 - Zero-sum games - General-sum games.
Implicit Hitting Set Problems Richard M. Karp Erick Moreno Centeno DIMACS 20 th Anniversary.
1 a1a1 A1A1 a2a2 a3a3 A2A Mixed Strategies When there is no saddle point: We’ll think of playing the game repeatedly. We continue to assume that.
1 Slides by Yong Liu 1, Deep Medhi 2, and Michał Pióro 3 1 Polytechnic University, New York, USA 2 University of Missouri-Kansas City, USA 3 Warsaw University.
TU/e Algorithms (2IL15) – Lecture 12 1 Linear Programming.
Computacion Inteligente Least-Square Methods for System Identification.
Approximation Algorithms based on linear programming.
On-Line Algorithms in Machine Learning By: WALEED ABDULWAHAB YAHYA AL-GOBI MUHAMMAD BURHAN HAFEZ KIM HYEONGCHEOL HE RUIDAN SHANG XINDI.
TU/e Algorithms (2IL15) – Lecture 12 1 Linear Programming.
Now that we have set of pure strategies for each player, we need to find the payoffs to put the game in strategic form. Random payoffs. The actual outcome.
Non-additive Security Games
The
Lecture 20 Linear Program Duality
The Weighted Majority Algorithm
Chapter 1. Formulations.
Presentation transcript:

The Multiplicative Weights Update Method Based on Arora, Hazan & Kale (2005) Mashor Housh Oded Cats Advanced simulation methods Prof. Rubinstein

Outline Weighted Majority Algorithm Binary case Generalized Applications Game Theory Zero-Sum game Linear Programming Fractional Packing problem NP-hard problems Set Cover problem Artificial intelligence (Boosting)

WMA – Binary case N experts give their predictions Our decision rule is a weighted majority of the expert predictions Initially, all experts have the same weight on our decision rule The update rule for incorrect experts is:

WMA – Binary case This procedure will yield in gains/losses that are roughly as good as those of the best of these experts Theorem 1 – The algorithm results in the following bound: Where: - the number of mistakes that expert I after t steps - the number of mistakes that our algorithm made

WMA Binary case – Proof of Theorem 1 I.By induction: II.Define the ‘potential function’: III.Each time we make a mistake, at least half of the total weight decrease by a factor of, so: IV.By induction: V.Using for

WMA – Binary case : Example1 4 analysts give their prediction to the stock exchange: 3 are always wrong and the 4 th is always right 4321 Market Day Expert

WMA – Binary case : Example1 (Cont.) 4321 Market /10.75/11.5/13/4 Balance of powers User Day Expert

WMA – Binary case : Example1 (Cont.) Since our fourth analyst is never wrong:

WMA – Binary case : Example2 100 analysts give their prediction to the stock exchange: 99 predict up with probability 0.05 the 100 th expert predicts up with probability 0.99 The market goes up at 99% of the time.

WMA – Binary case : Example2 (Cont.)

Generalization of the WMA Set of events/outcomes (P) is not bounded is the penalty that expert pays when the outcome is is the distribution associated with the experts The probability to choose an expert is: At every round we choose an expert according to D and follow his advice

Generalization of the WMA (Cont.) The update rule is: The expected penalty of the randomized algorithm is not much worse than that of the best expert Theorem 2 - The algorithm results in: Generalization of the WMA

WMA – Comparison via example The stock market example, using randomized expert instead of majority vote Market

WMA – Comparison via example (Cont.) With penalty only

WMA – Comparison via example (Cont.)

With penalty and reward

Generalization of the WMA - Example 4 weather man give their forecast There are four possible weather conditions The payoff matrix is: SunnyCloudyRainySnowy

Generalization of the WMA – Example (Cont.) The actual weather is sunny and cloudy alternately

Generalization of the WMA – Example (Cont.) The actual weather varies on the four possible weather conditions alternately

Applications Define the following components in order to draw analogy: Experts Events Payoff matrix Weights Update rule

Applications Game theory Zero-Sum games Experts – pure strategies to row player Events – pure strategies to column player Payoff matrix – the payoff to the row player, when the row player plays strategy and the column player plays strategy A distribution on the experts represents a mixed row strategy The game value is (von Neumann’s MinMax theory and Nash equilibrium):

Applications Game theory 1) Initialize. Determine. 2) Random a row strategy according to 3) The column player choose the strategy that maximizes his revenues 4) Update 5) If stop. Otherwise – return to step 2. Algorithm for solving Zero-Sum game:

Applications Game theory – Example /2 1/31/41 1/2 1/31/42 1/2 1/31/43 1/21/31/4 The row player choose minimum of maximum penalty The column player choose maximum of minimum penalty

Applications Game theory – Example1 (Cont.)

Applications Game theory – Example2 (1) The row player chooses a strategy randomly (2) The column player chooses the strategy that yield maximum benefits for him (3) Updating the weighting over row strategies 321 1/3 1/41 3/ /3 1/32/33 1/300

Applications Game theory – Example2 (Cont.)

Applications Artificial Intelligence The objective is to learn an unknown function A sequence of training examples is given: is the fixed unknown distribution on the domain The learning algorithm results in an hypothesis The error is:

Applications Artificial Intelligence (Cont.) Strong learning algorithm – for every and, with probability -weak learning algorithm – for every and, with probability Boosting – combining several moderately accurate rules-of-thumb into a singly highly accurate prediction rule.

Applications Artificial Intelligence (Cont.) Experts – samples in the training set Events – set of all hypotheses that can be generated by the weak learning algorithm Payoff matrix – The final hypothesis is obtained via majority vote among

Applications Linear Programming Finding a feasible solution for a set of m constraints Experts – constraints Events – solution vectors Payoff matrix - the distance from satisfying the constraint: The final solution is Track cases that there is no feasible solution

1) Initialize, and the resulting. 2) Given an oracle which solves the following feasibility problem with a single constraint plus a set of easy constraints (Plotkin, Shmoys and Tardos): Where: If there is no feasible solution – break. Applications Linear Programming (Cont.) Algorithm for finding a feasible solution to a LP problem:

Applications Linear Programming (Cont.) 3) Update Where: 4) Update 5) If stop. Otherwise – return to step 2. Algorithm for finding a feasible solution to a LP problem (Cont.):

Applications Linear Programming - Example Finding a feasible solution to the following problem: Solution:

Applications Fractional Vertex Covering problem Finding a feasible solution for a set of m constraints Experts – constraints Events – solution vectors Payoff matrix - The final solution is Track cases that there is no feasible solution

1) Initialize, and the resulting. 2) Given an oracle which solves the following feasibility problem with a single constraint plus a set of easy constraints (Plotkin, Shmoys and Tardos): Where: If there is no feasible solution – break. Applications Fractional Vertex Covering problem (Cont.) Algorithm for finding a feasible solution to a Fractional Covering problem:

Applications Fractional Vertex Covering problem (Cont.) 3) Update Where: 4) Update 5) If stop. Otherwise – return to step 2. Algorithm for finding a feasible solution to a Fractional Covering problem:

Applications Flow problems Maximum multi-commodity flow problem A set of source-sink pairs and capacity restrained edges

Applications Flow problems (Cont.) Experts – edges Events – a flow of value on the path, where is the minimum capacity of an edge. Payoff matrix: Update rule Terminate rule:

Applications Set Cover problem Find the minimal number of subsets in collection that their union equals the universe Experts – elements in the universe Events – sets in the collection Payoff matrix –

Applications Set Cover problem (Cont.) Update rule: We would search for the set which maximizes (Greedy Set Cover Algorithm)

Applications Vertex covering problem - Example

Applications Vertex covering problem – Example (Cont.) Find the minimum number of nodes (subsets) that cover all of the edges (universal) E set of edges 1,2,41,2,4 2,3,52,3,5 1,31,3 4,5,64,5,6 6

Applications Vertex covering problem – Example (Cont.) The maximum subset which includes edge is: node =1 The selected nodes are : c =1 Iteration: The probability for edge i p = Choose edge following the distribution i = 3 The maximum subset which includes edge is: node = 2 The selected nodes are : c =1 2 Iteration: The probability for edge i p = Choose edge following the distribution i = 6 The maximum subset which includes edge is: node = 4 The selected nodes are : c = 1 2 4

Applications Vertex covering problem – Example (Cont.) Iteration: The probability for edge i p = Choose edge following the distribution i = 6 The maximum subset which includes edge is: node = 5 The selected nodes are : c = 5 Iteration: The probability for edge i p = Choose edge following the distribution i = 3 The maximum subset which includes edge is: node = 2 The selected nodes are : c = 5 2 Iteration: The probability for edge i p = Choose edge following the distribution i = 1 The maximum subset which includes edge is: node = 1 The selected nodes are : c = 5 2 1

Summary This paper presents a comprehensive meta-analysis on the weights update method Various fields developed independently methods that has a common ground which can be generalized into one conceptual procedure The procedure includes the determination of experts, events, penalty matrix, weights and an update rule Additional relevant input: error size, size of and number of iterations.