Distributed Optimization Yen-Ling Kuo Der-Yeuan Yu May 27, 2010.

Slides:



Advertisements
Similar presentations
Constraint Satisfaction Problems
Advertisements

Problems and Their Classes
Adopt Algorithm for Distributed Constraint Optimization
1 Constraint Satisfaction Problems A Quick Overview (based on AIMA book slides)
Markov Decision Process
1 An Adaptive GA for Multi Objective Flexible Manufacturing Systems A. Younes, H. Ghenniwa, S. Areibi uoguelph.ca.
GRAPH BALANCING. Scheduling on Unrelated Machines J1 J2 J3 J4 J5 M1 M2 M3.
1 University of Southern California Keep the Adversary Guessing: Agent Security by Policy Randomization Praveen Paruchuri University of Southern California.
Spread of Influence through a Social Network Adapted from :
Interconnect throughput modeling. Important network performance metrics Throughput – Point to point (link bandwidth + end host software overheads) – Aggregate.
Cost-effective Outbreak Detection in Networks Jure Leskovec, Andreas Krause, Carlos Guestrin, Christos Faloutsos, Jeanne VanBriesen, Natalie Glance.
Randomized Sensing in Adversarial Environments Andreas Krause Joint work with Daniel Golovin and Alex Roper International Joint Conference on Artificial.
Preference Elicitation Partial-revelation VCG mechanism for Combinatorial Auctions and Eliciting Non-price Preferences in Combinatorial Auctions.
Seminar In Game Theory Algorithms, TAU, Agenda  Introduction  Computational Complexity  Incentive Compatible Mechanism  LP Relaxation & Walrasian.
Decision Theoretic Planning
Presenter: David Fleeman { D. Juedes, F. Drews, L. Welch and D. Fleeman Center for Intelligent, Distributed & Dependable.
Infinite Horizon Problems
Planning under Uncertainty
1 Stochastic Event Capture Using Mobile Sensors Subject to a Quality Metric Nabhendra Bisnik, Alhussein A. Abouzeid, and Volkan Isler Rensselaer Polytechnic.
Game-Theoretic Approaches to Multi-Agent Systems Bernhard Nebel.
A Decentralised Coordination Algorithm for Maximising Sensor Coverage in Large Sensor Networks Ruben Stranders, Alex Rogers and Nicholas R. Jennings School.
HMM-BASED PATTERN DETECTION. Outline  Markov Process  Hidden Markov Models Elements Basic Problems Evaluation Optimization Training Implementation 2-D.
Jie Gao Joint work with Amitabh Basu*, Joseph Mitchell, Girishkumar Stony Brook Distributed Localization using Noisy Distance and Angle Information.
INFERRING NETWORKS OF DIFFUSION AND INFLUENCE Presented by Alicia Frame Paper by Manuel Gomez-Rodriguez, Jure Leskovec, and Andreas Kraus.
Penn ESE535 Spring DeHon 1 ESE535: Electronic Design Automation Day 5: February 2, 2009 Architecture Synthesis (Provisioning, Allocation)
Achieving Minimum Coverage Breach under Bandwidth Constraints in Wireless Sensor Networks Maggie X. Cheng, Lu Ruan and Weili Wu Dept. of Comput. Sci, Missouri.
U NIVERSITY OF M ASSACHUSETTS, A MHERST Department of Computer Science Optimal Fixed-Size Controllers for Decentralized POMDPs Christopher Amato Daniel.
Multiagent Planning with Factored MDPs Carlos Guestrin Daphne Koller Stanford University Ronald Parr Duke University.
Maximum Network lifetime in Wireless Sensor Networks with Adjustable Sensing Ranges Mihaela Cardei, Jie Wu, Mingming Lu, and Mohammad O. Pervaiz Department.
9/23. Announcements Homework 1 returned today (Avg 27.8; highest 37) –Homework 2 due Thursday Homework 3 socket to open today Project 1 due Tuesday –A.
Penn ESE535 Spring DeHon 1 ESE535: Electronic Design Automation Day 5: February 2, 2009 Architecture Synthesis (Provisioning, Allocation)
Distributed Constraint Optimization * some slides courtesy of P. Modi
Coordinated Sampling sans Origin-Destination Identifiers: Algorithms and Analysis Vyas Sekar, Anupam Gupta, Michael K. Reiter, Hui Zhang Carnegie Mellon.
CS Reinforcement Learning1 Reinforcement Learning Variation on Supervised Learning Exact target outputs are not given Some variation of reward is.
Instructor: Vincent Conitzer
A1A1 A4A4 A2A2 A3A3 Context-Specific Multiagent Coordination and Planning with Factored MDPs Carlos Guestrin Shobha Venkataraman Daphne Koller Stanford.
MAKING COMPLEX DEClSlONS
Distributed Constraint Optimization Michal Jakob Agent Technology Center, Dept. of Computer Science and Engineering, FEE, Czech Technical University A4M33MAS.
QoS-Based Web service Selection and Agreement Marco Comuzzi Dipartimento di Elettronica e Informazione Politecnico di Milano
General Polynomial Time Algorithm for Near-Optimal Reinforcement Learning Duke University Machine Learning Group Discussion Leader: Kai Ni June 17, 2005.
REINFORCEMENT LEARNING LEARNING TO PERFORM BEST ACTIONS BY REWARDS Tayfun Gürel.
An Online Auction Framework for Dynamic Resource Provisioning in Cloud Computing Weijie Shi*, Linquan Zhang +, Chuan Wu*, Zongpeng Li +, Francis C.M. Lau*
Maximum Network Lifetime in Wireless Sensor Networks with Adjustable Sensing Ranges Cardei, M.; Jie Wu; Mingming Lu; Pervaiz, M.O.; Wireless And Mobile.
Practical Dynamic Programming in Ljungqvist – Sargent (2004) Presented by Edson Silveira Sobrinho for Dynamic Macro class University of Houston Economics.
1 Online algorithms Typically when we solve problems and design algorithms we assume that we know all the data a priori. However in many practical situations.
CS483/683 Multi-Agent Systems Lectures 9-10: Distributed Constraint Optimization: Auction-based solutions 16 February 2009 Instructor: Kostas Bekris Computer.
1 Markov Decision Processes Infinite Horizon Problems Alan Fern * * Based in part on slides by Craig Boutilier and Daniel Weld.
1 Markov Decision Processes Infinite Horizon Problems Alan Fern * * Based in part on slides by Craig Boutilier and Daniel Weld.
1 Outline:  Optimization of Timed Systems  TA-Modeling of Scheduling Tasks  Transformation of TA into Mixed-Integer Programs  Tree Search for TA using.
1 Iterative Integer Programming Formulation for Robust Resource Allocation in Dynamic Real-Time Systems Sethavidh Gertphol and Viktor K. Prasanna University.
Algorithmic Game Theory and Internet Computing Vijay V. Vazirani Georgia Tech Primal-Dual Algorithms for Rational Convex Programs II: Dealing with Infeasibility.
CS 3343: Analysis of Algorithms Lecture 19: Introduction to Greedy Algorithms.
Deterministic Algorithms for Submodular Maximization Problems Moran Feldman The Open University of Israel Joint work with Niv Buchbinder.
Cost-effective Outbreak Detection in Networks Presented by Amlan Pradhan, Yining Zhou, Yingfei Xiang, Abhinav Rungta -Group 1.
Coverage Problems in Wireless Ad-hoc Sensor Networks Seapahn Meguerdichian 1 Farinaz Koushanfar 2 Miodrag Potkonjak 1 Mani Srivastava 2 University of California,
DEPARTMENT/SEMESTER ME VII Sem COURSE NAME Operation Research Manav Rachna College of Engg.
Approximation Algorithms based on linear programming.
Multi-Agents System CMSC 691B Gunjan Kalra Peter DSouza.
Lecture 20 Review of ISM 206 Optimization Theory and Applications.
Slide 1 Toward Optimal Sniffer-Channel Assignment for Reliable Monitoring in Multi-Channel Wireless Networks Donghoon Shin, Saurabh Bagchi and Chih-Chun.
Monitoring rivers and lakes [IJCAI ‘07]
Algorithmic Game Theory and Internet Computing
Chapter 6. Large Scale Optimization
MURI Kickoff Meeting Randolph L. Moses November, 2008
Cost-effective Outbreak Detection in Networks
Market-based Dynamic Task Allocation in Mobile Surveillance Systems
CS 416 Artificial Intelligence
Constraint Satisfaction Problems
Chapter 6. Large Scale Optimization
Presentation transcript:

Distributed Optimization Yen-Ling Kuo Der-Yeuan Yu May 27, 2010

Outline [Yu] Optimized Sensing: From Water to the Web Distributed Dynamic Programming Distributed Solutions to Markov Decision Problems

Optimized Sensing Problem Statement Greedy Algorithms and Submodularity Robust Sensing Optimization with Saturate Algorithm Application in Blogs

Problem Statement How do we detect contamination in drinking water distribution networks? Which blogs should we read to learn about the biggest, newest stories on the Web? Fundamental Question: How can we get the most useful information at minimum cost (limited resources)?

Solutions to Optimized Sensing Covers fields of statistics, machine learning, sensor networks, and robotics With partially observable Marko decision processes, we can get optimal solutions But it is difficult to scale POMDP to large problems Introducing a new algorithm based on submodularity

Formulation Sensing quality function F(A) – A: the set of sensor locations Si (i=1~k) – V: the set of all locations We can also have cost constraints – Total cost of sensor deployment no greater than the budget Goal: Find A* – This is NP-hard already

Greedy Algorithm Iteratively find Si This naïve algorithm actually performs pretty well – Why? Submodularity – We get near-optimal solutions Submodularity: diminishing returns

Diminishing Returns

Cost-Effective Lazy Forward-Selection (CELP) Greedy algorithm Lazy evaluations – Delaying computation until the result is required – A computational technique

Robust Sensing Optimization Idea: Protect system against adversaries that know of our deployment of sensors Goal: Maximize the worst-case detection performance Approach Unfortunately, this naïve extension can fail

Failure of Greedy Algorithm on Worst- Case Scenarios I1, I2: two contamination events S1, S2, S3: three possible sensor locations – S1: detect I1 immediately, but never I2 – S2: detect I2 immediately, but never I1 – S3: detect both I1 and I2, but only after a long time We can only place two sensors Greedy would pick S3 first and then either S1 or S2 But we know the optimal solution should be S1 and S2 Solution? Saturate algorithm

Saturate Algorithm Idea: reduce the non-submodular worst-case objective to a submodular optimization problem – Transform non-submodular to submodular Transformation – Guess optimal solution value C using binary search – Try to find A such that F(A) is no less than C

Performance of Saturate

From Water to the Web Blog Reading Problem: Information cascading

Improvements Number-of-posts (NP) model – Reading a big blog can be time-consuming, so they define the cost to be the number of posts CELP tends to choose blogs with many posts NP model tends to choose summarizer blogs – But stories appear in summarizer blogs a little late

Other Thoughts What if we are looking for stories to read instead of blogs to read? – We can reverse our information management goal – Find posts instead of blogs – Ref. 10 End of Paper

Distributed Dynamic Programming for Path Planning Asynchronous Dynamic Programming Learning Real-Time A*

Asynchronous Dynamic Programming Propagate costs from target to start locations

Learning Real-Time A* (LRTA*)

LRTA*(n) LRTA with n agents Faster – Agents break ties differently – They can share the same h-value table

LRTA*(2)

Distributed Solutions to Markov Decision Problems As previously mentioned in the Water to Web paper, MDPs can be difficult to scale to big problems Solution: Exploit independence properties We address the modularity of actions

Action Selection in multiagent MDPs

Implementation

Subtask Distribution A global problem is broken down into subtasks Subtasks are distributed among agents Each agent has different capabilities Problem 25

Contract Net Stages – Recognition – Announce – Bidding – Awarding & Expediting Initial assignment: Not optimal Anytime property – Improve assignment in negotiation process 26

Assignment problem Problem definition – A set N of n agents – A set X of n objects – A set M ⊆ N × X of possible assignment pairs, and – A function v : M → R Find optimal assignment XN M 27

Corresponding Linear Program Linear program (LP) formulation Profit maximization Resource constraint Optimal solution Any LP can be solved in polynomial time O(n 3 ) 28

Competitive Equilibrium Consider a price vector p = (p 1, …, p n ) – The utility from an assignment j to agent i is u(i, j) = v(I, j) - p j A feasible assignment S and a price vector p are in competitive equilibrium when for every pairing (i, j) ∈ S it is the case that ∀ k, u(i, j) ≥ u(i, k) 29 Every agent will not change its selection S is a optimal solution

Naïve Auction Algorithm Round-robin style Bid increment is the difference between the utility to i of the best and second-best object 30 The agent will not overbid

Problem in Naïve Auction When more than one object offers maximal utility for an agent – Bid increment is zero 31

Terminating Auction Algorithm Modify the bid increment – 32 ε-competitive equilibrium: u(i, j) + ε ≥ u(i, k) Agents may overbid some objects

Scheduling Problem Problem definition – N is a set of n agents – X is a set of m discrete and consecutive time slots – q = (q 1,..., q m ) is a reserve price vector – v = (v 1,..., v n ), where v i is the valuation function of agent I Find optimal allocation 33 F

Corresponding Integer Program Integer program (IP) formulation IPs are not solvable polynomial time 34

Competitive Equilibrium – General Form Definition – For all i ∈ N it is the case that F i = argmax T ⊆ X (v i (T) − ∑ j|x j ∈ T p j ) – For all j such that x j ∈ F ∅ it is the case that p j = q j – For all j such that x j ∈ F ∅ it is the case that p j ≥ q j May not exist competitive equilibrium 35 Has a competitive equilibrium solution ↕ The LP relaxation of the associated integer program has a integer solution.

Ascending Auction Algorithm Center advertise an ask price Bid increment is constant 36

Problem in Ascending Auction If the increment is too large May not converge to optimal solution 37

Social Laws and Conventions Social law – A restriction on the given strategies of the agents – Induce a sub-game Social convention – The sub-game consists of a single strategy for all agent Other topics – Social goal negotiation – Social norm negotiation – …. 38