Common Voting Rules as Maximum Likelihood Estimators Vincent Conitzer and Tuomas Sandholm Carnegie Mellon University, Computer Science Department.

Slides:



Advertisements
Similar presentations
Sep 16, 2013 Lirong Xia Computational social choice The easy-to-compute axiom.
Advertisements

Sep 15, 2014 Lirong Xia Computational social choice The easy-to-compute axiom.
On the Robustness of Preference Aggregation in Noisy Environments Ariel D. Procaccia, Jeffrey S. Rosenschein and Gal A. Kaminka.
Common Voting Rules as Maximum Likelihood Estimators Vincent Conitzer (Joint work with Tuomas Sandholm) Early version of this work appeared in UAI-05.
Voting and social choice Vincent Conitzer
Algorithmic Game Theory Uri Feige Robi Krauthgamer Moni Naor Lecture 9: Social Choice Lecturer: Moni Naor.
Voting and social choice Looking at a problem from the designers point of view.
How “impossible” is it to design a Voting rule? Angelina Vidali University of Athens.
IMPOSSIBILITY AND MANIPULABILITY Section 9.3 and Chapter 10.
CS 886: Electronic Market Design Social Choice (Preference Aggregation) September 20.
Complexity of manipulating elections with few candidates Vincent Conitzer and Tuomas Sandholm Carnegie Mellon University Computer Science Department.
Computing Kemeny and Slater Rankings Vincent Conitzer (Joint work with Andrew Davenport and Jayant Kalagnanam at IBM Research.)
Making Group Decisions Mechanism design: study solution concepts
Excursions in Modern Mathematics Sixth Edition
Lirong Xia Friday, May 2, 2014 Introduction to Social Choice.
But, how?? Explaining all possible positional, pairwise voting paradoxes & prop. Don Saari Institute for Math Behavioral Sciences University of California,
Using a Modified Borda Count to Predict the Outcome of a Condorcet Tally on a Graphical Model 11/19/05 Galen Pickard, MIT Advisor: Dr. Whitman Richards,
Manipulation Toby Walsh NICTA and UNSW. Manipulation Constructive  Can we change result so a given candidate wins Destructive  Can we change result.
Socially desirable approximations for Dodgson’s voting rule Ioannis Caragiannis (University of Patras) Christos Kaklamanis (University of Patras) Nikos.
+ Random Tie Breaking Toby Walsh NICTA and UNSW. + Random Tie Breaking Haris Aziz, Serge Gaspers, Nick Mattei, Nina Narodytska, Toby Walsh NICTA and UNSW.
Convergence of Iterative Voting AAMAS 2012 Valencia, Spain Omer Lev & Jeffrey S. Rosenschein.
Using computational hardness as a barrier against manipulation Vincent Conitzer
Using computational hardness as a barrier against manipulation Vincent Conitzer
The Distortion of Cardinal Preferences in Voting Ariel D. Procaccia and Jeffrey S. Rosenschein.
Ties Matter: Complexity of Voting Manipulation Revisited based on joint work with Svetlana Obraztsova (NTU/PDMI) and Noam Hazon (CMU) Edith Elkind (Nanyang.
Preference elicitation Vincent Conitzer
CPS Voting and social choice
Social choice theory = preference aggregation = voting assuming agents tell the truth about their preferences Tuomas Sandholm Professor Computer Science.
1 Algorithms for Large Data Sets Ziv Bar-Yossef Lecture 7 April 20, 2005
Complexity of unweighted coalitional manipulation under some common voting rules Lirong XiaVincent Conitzer COMSOC08, Sep. 3-5, 2008 TexPoint fonts used.
Preference Analysis Joachim Giesen and Eva Schuberth May 24, 2006.
Preference Functions That Score Rankings and Maximum Likelihood Estimation Vincent Conitzer Matthew Rognlie Lirong Xia Duke University Thanks (at least)
Social choice theory = preference aggregation = truthful voting Tuomas Sandholm Professor Computer Science Department Carnegie Mellon University.
Strategic Behavior in Multi-Winner Elections A follow-up on previous work by Ariel Procaccia, Aviv Zohar and Jeffrey S. Rosenschein Reshef Meir The School.
Introduction complexity has been suggested as a means of precluding strategic behavior. Previous studies have shown that some voting protocols are hard.
Social choice (voting) Vincent Conitzer > > > >
Social Choice Lecture 19 John Hey.
CPS Voting and social choice Vincent Conitzer
An Algorithm for the Coalitional Manipulation Problem under Maximin Michael Zuckerman, Omer Lev and Jeffrey S. Rosenschein AAMAS’11.
An Algorithm for the Coalitional Manipulation Problem under Maximin Michael Zuckerman, Omer Lev and Jeffrey S. Rosenschein (Simulations by Amitai Levy)
Automated Design of Multistage Mechanisms Tuomas Sandholm (Carnegie Mellon) Vincent Conitzer (Carnegie Mellon) Craig Boutilier (Toronto)
Elections and Strategic Voting: Condorcet and Borda E. Maskin Harvard University.
1 Elections and Manipulations: Ehud Friedgut, Gil Kalai, and Noam Nisan Hebrew University of Jerusalem and EF: U. of Toronto, GK: Yale University, NN:
Avoiding manipulation in elections through computational complexity Vincent Conitzer Computer Science Department Carnegie Mellon University Guest lecture.
Social choice theory = preference aggregation = voting assuming agents tell the truth about their preferences Tuomas Sandholm Professor Computer Science.
Warm-Up Rank the following soft drinks according to your preference (1 being the soft drink you like best and 4 being the one you like least)  Dr. Pepper.
The Mathematics of Voting Chapter 1. Preference Ballot A Ballot in which the voters are asked to rank the candidates in order of preference. 1. Brownies.
Chapter 9: Social Choice: The Impossible Dream Lesson Plan Voting and Social Choice Majority Rule and Condorcet’s Method Other Voting Systems for Three.
11/24/2008CS Common Voting Rules as Maximum Likelihood Estimators - Matthew Kay 1 Common Voting Rules as Maximum Likelihood Estimators Vincent Conitzer,
Arrow’s Impossibility Theorem
When Are Elections with Few Candidates Hard to Manipulate V. Conitzer, T. Sandholm, and J. Lang Subhash Arja CS 286r October 29, 2008.
0 Fall, 2016 Lirong Xia Computational social choice The easy-to-compute axiom.
Algorithms for Large Data Sets
Introduction to Social Choice
Chapter 10: The Manipulability of Voting Systems Lesson Plan
Introduction to Social Choice
Social choice theory = preference aggregation = voting assuming agents tell the truth about their preferences Tuomas Sandholm Professor Computer Science.
Applied Mechanism Design For Social Good
A Brief Introductory Tutorial on Computational Social Choice
Algorithmic Analysis of Elections: Voting Rules and Manipulability (minicourse) Piotr Faliszewski AGH University Kraków, Poland.
Introduction If we assume
A Crash Course on Computational Social Choice and Fair Division
Maximal Independent Set
Computational Social Choice and Moral Artificial Intelligence
Voting systems Chi-Kwong Li.
Voting and social choice
Preference elicitation/ iterative mechanisms
Introduction to Social Choice
CPS 173 Voting and social choice
CPS Voting and social choice
Presentation transcript:

Common Voting Rules as Maximum Likelihood Estimators Vincent Conitzer and Tuomas Sandholm Carnegie Mellon University, Computer Science Department

Voting (rank aggregation) rules Set of m candidates (alternatives) C n voters; each voter ranks the candidates (the voter’s vote) –E.g. b > a > c > d Voting rule f maps every vector of votes to either: –a winner in C, or –a complete ranking of C E.g. plurality: –every voter votes for a single candidate (equiv. we only consider the candidate’s top-ranked candidate) –candidate with most votes wins/candidates are ranked by score

Two views of voting 1.Voters’ preferences are idiosyncratic; only purpose is to find a compromise winner/ranking 2.There is some absolute sense in which some candidates are better than others, independent of voters’ preferences; votes are merely noisy perceptions of candidates’ true quality a “correct” outcome a agents’ votes a “correct” outcome a vote 1 a vote 2 a vote n … conditional independence assumption Goal: given votes, find maximum likelihood estimate of correct outcome Different noise model  different maximum likelihood estimator/voting rule (outcome=winner or ranking)

History [Condorcet 1785] assumed noise model where voter ranks any two candidates correctly with fixed probability p > 1/2, independently –Gives cyclical rankings with some probability, but does not affect MLE approach –Solved cases of 2 and 3 candidates Two centuries pass… [Young 1995] solved case of arbitrary number of candidates under the same model –Showed that it coincided with rule proposed by Kemeny [Kemeny 1959] [Drissi & Truchon 2002] extend to the case where p is allowed to vary with the distance between two candidates in correct ranking

What is next? Does this suggest using Kemeny rule? –Many other noise models possible –Some of these may correspond to other, better-known rules Goal of this paper: Classify which common rules are a maximum likelihood estimator for some noise model –Positive and negative results –Positive results are constructive Motivation: –Rules corresponding to a noise model are more natural –Knowing a noise model can give us insight into the rule and its underlying assumptions –If we disagree with the noise model, we can modify it and obtain new version of the rule

Independence restriction Without any independence restriction, it turns out that any rule has a noise model: P(vote vector|outcome) > 0 if and only if f(vote vector)=outcome a “correct” outcome a agents’ votes a “correct” outcome a vote 1 a vote 2 a vote n … conditional independence assumption So, will focus on conditionally independent votes If a rule has a noise model in this setup we call it an –MLEWIV rule if producing winner –MLERIV rule if producing ranking –(IV = Independent Votes)

Any scoring rule is MLEWIV and MLERIV Scoring rule gives a candidate a 1 points if it is ranked first, a 2 points if it is ranked second, etc. –plurality rule: a 1 = 1, a i = 0 otherwise –Borda rule: a i = m-i –veto rule: a m = 0, a i = 1 otherwise MLEWIV noise model: P(v|w) = 2 a l(v,w) where l(v,w) is the rank of w in v –want to choose w to maximize Π v 2 a l(v,w) = 2 Σ v a l(v,w) MLERIV noise model: P(v|r) = Π 1≤i≤m (m+1-i) a l(v,r i ) where r i is the candidate ranked ith in r

Single Transferable Vote (STV) is MLERIV STV rule: Candidate ranked first by fewest voters drops out and is removed from rankings, final ranking is inverse of order in which they dropped out MLERIV noise model: –Let r i be the candidate ranked ith in r –Let δ v (r i ) = 1 if all the candidates ranked higher than r i in v are ranked lower in r (i.e. they are all contained in {r i+1, r i+2, …, r m }), otherwise 0 –P(v|r) = Π 1≤i≤m k i δ v (r i ) where k i+1 << k i < 1

Lemma to prove negative results For any noise model, if there is a single outcome that maximizes the likelihood of both vote vector 1 and vote vector 2, then it must also maximize the likelihood of vote vector 3 Hence, a voting rule that produces the same outcome on both vector 1 and vector 2 but a different one on vector 3 cannot be a maximum likelihood estimator correct outcome vote 1 vote k vote k+1 vote n vote vector 1 vote vector 2 vote vector 3 … …

STV rule is not MLEWIV STV rule: Candidate ranked first by fewest voters drops out and is removed from rankings, final ranking is inverse of order in which they dropped out First vote vector: –3 times c > a > b –4 times a > b > c –6 times b > a > c –c drops out first, then a wins Second vote vector: –3 times b > a > c –4 times a > c > b –6 times c > a > b –b drops out first, then a wins But: taking all votes together, a drops out first! –(8 votes vs. 9 for the others)

Bucklin rule is not MLEWIV/MLERIV Bucklin rule: –For every candidate, consider the minimum k such that more than half of the voters rank that candidate among the top k –Candidates are ranked (inversely) by their minimum k –Ties are broken by the number of voters by which the “half” mark is passed First vote vector: –2 times a > b > c > d > e –1 time b > a > c > d > e –gives final ranking a > b > c > d > e Second vote vector: –2 times b > d > a > c > e –1 time c > e > a > b > d –1 time c > a > b > d > e –gives final ranking a > b > c > d > e But: taking all votes together gives final ranking b > a > c > d > e –(b goes over half at k=2, a does not)

Pairwise election graphs Pairwise election: take two candidates and see which one is ranked above the other in more votes Pairwise election graph has edge of weight k from a to b if a defeats b by k votes in the pairwise election E.g. votes a > b > c and b > a > c together produce pairwise election graph:

(Roughly) all pairwise election graphs can be realized Lemma: any graph with even weights is the pairwise election graph for some votes Proof: can increase the weight of edge from a to b by two by adding the following two votes: –a > b > c 1 > c 2 > … > c m-2 –c m-2 > c m-1 > … c 1 > a > b Hence, from here on, we will simply show the pairwise election graph rather than the votes that realize it

Copeland is not MLEWIV/MLERIV Copeland rule: candidate’s score = number of pairwise victories – number of pairwise defeats –i.e. outdegree – indegree of vertex in pairwise election graph a: 3-1 = 2 b: 2-1 = 1 c: 2-2 = 0 d: 1-2 = -1 e: 1-3 = -2 a: 3-1 = 2 b: 2-1 = 1 c: 2-2 = 0 d: 1-2 = -1 e: 1-3 = -2 b: 2-0 = 2 a: 2-1 = 1 c: 2-2 = 0 d: 1-2 = -1 e: 0-2 = -2 + =

Maximin is not MLEWIV/MLERIV maximin rule: candidate’s score = score in worst pairwise election –i.e. candidates are ordered inversely by weight of largest incoming edge a: 6 b: 8 c: 10 d: 12 + = a: 6 b: 8 c: 10 d: 12 c: 2 a: 4 d: 6 b: 8

Ranked pairs is not MLEWIV/MLERIV ranked pairs rule: pairwise elections are locked in according by margin of victory –i.e. larger edges are “fixed” first, an edge is discarded if it introduces a cycle b > d fixed a > b fixed d > a discarded b > c fixed c > d fixed result: a > b > c > d + = a > c fixed c > d fixed d > a discarded b > c fixed a > b fixed result: a > b > c > d d > a fixed c > d fixed a > c discarded b > d fixed a > b discarded b > c fixed result: b > c > d > a

Conclusions MLERIVnot MLERIV MLEWIVscoring rules (incl. plurality, Borda, veto) hybrids of MLEWIV and (not MLERIV) rules not MLEWIVSTVBucklin, Copeland, maximin, ranked pairs Thank you for your attention! We asked the question: which common voting rules are maximum likelihood estimators (for some noise model)? If votes are not independent given outcome (winner/ranking), any rule is MLE If votes are independent given outcome, some rules are MLEWIV (MLE for winner), some are MLERIV (MLE for ranking), some are both: