Geometry of Online Packing Linear Programs Marco Molinaro and R. Ravi Carnegie Mellon University.

Slides:



Advertisements
Similar presentations
A Primal-Dual Approach to Online Optimization Problems.
Advertisements

Explaining The Distance Formula By: Mario Guzman Geometry per.3.
Linear programming: lp_solve, max flow, dual CSC 282 Fall 2013.
1 Non-Linear and Smooth Regression Non-linear parametric models: There is a known functional form y=  x,  derived from known theory or from previous.
1 Learning with continuous experts using Drifting Games work with Robert E. Schapire Princeton University work with Robert E. Schapire Princeton University.
Recurrence Relations Reading Material –Chapter 2 as a whole, but in particular Section 2.8 –Chapter 4 from Cormen’s Book.
Recurrence Relations Reading Material –Chapter 2 as a whole, but in particular Section 2.8 –Chapter 4 from Cormen’s Book.
Ger man Aerospace Center Gothenburg, April, 2007 Coding Schemes for Crisscross Error Patterns Simon Plass, Gerd Richter, and A.J. Han Vinck.
1 Topic The Substitution Method. 2 Topic The Substitution Method California Standard: 9.0 Students solve a system of two linear equations.
EXERCISE R.4 1 R.4*Find the expected value of X in Exercise R.2. [R.2*A random variable X is defined to be the larger of the numbers when two dice are.
7.1 SOLVING SYSTEMS BY GRAPHING The students will be able to: Identify solutions of linear equations in two variables. Solve systems of linear equations.
Using Error-Correcting Codes For Text Classification Rayid Ghani Center for Automated Learning & Discovery, Carnegie Mellon University.
How the Experts Algorithm Can Help Solve LPs Online Marco Molinaro TU Delft Anupam Gupta Carnegie Mellon University.
Solving Linear Systems by Linear Combinations
1.3 “Solving Linear Equations” Steps: 1.Isolate the variable. 2.To solve when there is a fraction next to a variable, multiply both sides by the reciprocal.
Approximation Algorithms for Stochastic Combinatorial Optimization Part I: Multistage problems Anupam Gupta Carnegie Mellon University.
Experiences with Enumeration of Integer Projections of Parametric Polytopes Sven Verdoolaege, Kristof Beyls, Maurice Bruynooghe, Francky Catthoor Compiler.
Solving Linear Systems by Substitution O Chapter 7 Section 2.
Soham Uday Mehta. Linear Programming in 3 variables.
Column Generation By Soumitra Pal Under the guidance of Prof. A. G. Ranade.
BOĞAZİÇİ UNIVERSITY DEPARTMENT OF MANAGEMENT INFORMATION SYSTEMS MATLAB AS A DATA MINING ENVIRONMENT.
How the Experts Algorithm Can Help Solve LPs Online Marco Molinaro TU Delft Anupam Gupta Carnegie Mellon University.
Sparse Cutting-Planes Marco Molinaro TU Delft Santanu Dey, Andres Iroume Georgia Tech.
7-3 Elimination Using Addition and Subtraction 7-4 Elimination Using Multiplication Objective: Students will be able to: Solve a system of linear equations.
HASE: A Hybrid Approach to Selectivity Estimation for Conjunctive Queries Xiaohui Yu University of Toronto Joint work with Nick Koudas.
Deterministic Algorithms for Submodular Maximization Problems Moran Feldman The Open University of Israel Joint work with Niv Buchbinder.
SystemsOfInequalities. 7-1 Solving Systems by Graphing What is a system of linear equations? “SOLUTION” No solution Infinitely Many Solutions Page 342.
Sparse Cutting-Planes Marco Molinaro Santanu Dey, Andres Iroume, Qianyi Wang Georgia Tech.
Zonk & Wam!!. Topics  Linear Equations (one variable)  Linear Inequalities (one variable)  Compound Inequalities  Absolute Value Equations (one variable)
1 Approximation Algorithms for Generalized Min-Sum Set Cover Ravishankar Krishnaswamy Carnegie Mellon University joint work with Nikhil Bansal and Anupam.
SYSTEMS OF EQUATIONS. SYSTEM OF EQUATIONS -Two or more linear equations involving the same variable.
Scale Selection for Classification of Point-sampled 3-D Surfaces Jean-François Lalonde, Ranjith Unnikrishnan, Nicolas Vandapel and Martial Hebert Carnegie.
Solving systems of equations with three variables January 13, 2010.
1.1 The row picture of a linear system with 3 variables.
Notes 6.5, Date__________ (Substitution). To solve using Substitution: 1.Solve one equation for one variable (choose the variable with a coefficient of.
Solve Linear Systems by Elimination February 3, 2014 Pages
Massive Support Vector Regression (via Row and Column Chunking) David R. Musicant and O.L. Mangasarian NIPS 99 Workshop on Learning With Support Vectors.
Holt McDougal Algebra Solving Systems of Linear Inequalities Solve systems of linear inequalities. Objective.
A Discriminatively Trained, Multiscale, Deformable Part Model Yeong-Jun Cho Computer Vision and Pattern Recognition,2008.
TODAY IN ALGEBRA 2.0…  Review: Solving Linear Systems by Graphing  Learning Goal 1: 3.2 Solving Linear Systems by Substitution with one equation solved.
$200 $400 $600 $800 $1000 $200 $400 $600 $800 $1000 $200 $400 $600 $800 $1000 $200 $400 $600 $800 $1000 $200 $400 $600 $800 $1000 $200.
Beating the Direct Sum Theorem in Communication Complexity with Implications for Sketching Marco Molinaro David Woodruff Grigory Yaroslavtsev Georgia Tech.
3.5 Solving systems of equations in three variables Main Ideas Solve systems of linear equations in three variables. Solve real-world problems using systems.
Objectives: Graph (and write) inequalities on a number line.
Solve Linear Systems By Multiplying First
Students will be able to:
Solve more difficult number problems mentally
Maximum Matching in the Online Batch-Arrival Model
3-2 Graphs of Linear Equations in 2 Variables
5.2 Solving Systems Using Substitution
Solve: 1. 4<
Solving Linear Systems by Linear Combinations
REVIEW: Solving Linear Systems by Elimination
2.7 Two-variable inequalities (linear) 3.3 Systems of Inequalities
Warm Up Graph y = 4x and 16, $10’s and 7 $20’s.
Choose the best answer for each problem.
ICS 353: Design and Analysis of Algorithms
Notes Solving a System by Elimination
Choosing the Best Method
Warm Up Graph y = 4x and 16, $10’s and 7 $20’s.
GRAPHING LINEAR EQUATIONS
Cases. Simple Regression Linear Multiple Regression.
ICS 353: Design and Analysis of Algorithms
Solving a System of Linear Equations
Multiplication and Division of Integers
Learner Objective: Students will write simple two column proofs.
Types of Errors And Error Analysis.
Notes Over 6.1 Graphing a Linear Inequality Graph the inequality.
Practice Geometry Practice
Presentation transcript:

Geometry of Online Packing Linear Programs Marco Molinaro and R. Ravi Carnegie Mellon University

Packing Integer Programs (PIPs) A x ≤ b m n

A Online Packing Integer Programs Adversary chooses values for c, A, b …but columns are presented in random order …when column comes, set variable to 0/1 irrevocably b and n are known upfront x ≤ b c A A A A n A A 1 0

Online Packing Integer Programs

Previous Results do not depend on n depends on n

Main Question and Result Q: Do general PIPs become more difficult for larger n? A: No!

High-level Idea 1.Online PIP as learning 2.Improving learning error using tailored covering bounds 3.Geometry of PIPs that allow good covering bounds 4.Reduce general PIP to above

Online PIP as Learning

2)Solving PIP via learning

Online PIP as Learning 2)Solving PIP via learning

Online PIP as Learning 2)Solving PIP via learning Improve this…

Improved Learning Error Idea 1: Covering bounds via witnesses (handling multiple bad classifiers at a time)

Geometry of PIPs with Small Witness Set

Conclusion

Thank you!