Isoelastic Agents and Wealth Updates in Machine Learning Markets

Recently, prediction markets have shown considerable promise for developing flexible mechanisms for machine learning. In this paper, agents with isoelastic utilities are considered. It is shown that the costs associated with homogeneous markets of agents with isoelastic utilities produce equilibrium prices corresponding to alpha-mixtures, with a particular form of mixing component relating to each agent's wealth. We also demonstrate that wealth accumulation for logarithmic and other isoelastic agents (through payoffs on prediction of training targets) can implement both Bayesian model updates and mixture weight updates by imposing different market payoff structures. An iterative algorithm is given for market equilibrium computation. We demonstrate that inhomogeneous markets of agents with isoelastic utilities outperform state of the art aggregate classifiers such as random forests, as well as single classifiers (neural networks, decision trees) on a number of machine learning benchmarks, and show that isoelastic combination methods are generally better than their logarithmic counterparts.

[1]  C. Manski.  Interpreting the Predictions of Prediction Markets , 2004 .

[2]  S. Amari.  Integration of Stochastic Models by Minimizing -Divergence , 2007, Neural Computation.

[3]  Paul C. Tetlock,et al.  The Promise of Prediction Markets , 2008, Science.

[4]  Prithviraj Dasgupta,et al.  A multi‐agent system for analyzing the effect of information on prediction markets , 2011, Int. J. Intell. Syst..

[5]  Nathan Lay,et al.  Supervised Aggregation of Classifiers using Artificial Prediction Markets , 2010, ICML.

[6]  Nathan Lay,et al.  An introduction to artificial prediction markets for classification , 2011, J. Mach. Learn. Res..

[7]  P. Samuelson,et al.  Foundations of Economic Analysis. , 1948 .

[8]  K. Arrow,et al.  EXISTENCE OF AN EQUILIBRIUM FOR A COMPETITIVE ECONOMY , 1954 .

[9]  John Langford,et al.  Learning performance of prediction markets with Kelly bettors , 2012, AAMAS.

[10]  R. Cole,et al.  Fast-Converging Tatonnement Algorithms for the Market Problem , 2007 .

[11]  Pedro M. Domingos.  Bayesian Averaging of Classifiers and the Overfitting Problem , 2000, ICML.

[12]  Nimrod Megiddo,et al.  Equilibrium in prediction markets with buyers and sellers , 2010 .

[13]  R. Hanson,et al.  An experimental test of combinatorial information markets , 2009 .

[14]  Jacob D. Abernethy,et al.  A Collaborative Mechanism for Crowdsourcing Prediction Problems , 2011, NIPS.

[15]  Amos J. Storkey,et al.  Machine Learning Markets , 2011, AISTATS.

[16]  David S. Lee,et al.  Bayesian Learning and the Pricing of New Information: Evidence from Prediction Markets , 2009 .

[17]  Thomas P. Minka,et al.  Bayesian model averaging is not model combination , 2002 .

[18]  Rich Caruana,et al.  An empirical comparison of supervised learning algorithms , 2006, ICML.

[19]  Amin Saberi,et al.  A Fast and Simple Algorithm for Computing Market Equilibria , 2008, WINE.

[20]  Jennifer Wortman Vaughan,et al.  An optimization-based framework for automated market-making , 2010, EC '11.

[21]  Dalei Wu.  Parameter Estimation for -GMM Based on Maximum Likelihood Criterion , 2009, Neural Computation.

[22]  Nikhil R. Devanur,et al.  Market equilibrium via a primal--dual algorithm for a convex program , 2008, JACM.

[23]  Xiaotie Deng,et al.  On the complexity of equilibria , 2002, STOC '02.

[24]  V. Vazirani.  Algorithmic Game Theory: Combinatorial Algorithms for Market Equilibria , 2007 .

[25]  Yehuda Koren,et al.  Lessons from the Netflix prize challenge , 2007, SKDD.

[26]  Michael P. Wellman,et al.  Representing Aggregate Belief through the Competitive Equilibrium of a Securities Market , 1997, UAI.

[27]  L. V. Williams,et al.  Prediction Markets , 2003 .

[28]  Pedro M. Domingos.  Why Does Bagging Work? A Bayesian Account and its Implications , 1997, KDD.

[29]  Jennifer Wortman Vaughan,et al.  A new understanding of prediction markets via no-regret learning , 2010, EC '10.

[30]  Yinyu Ye,et al.  A path to the Arrow–Debreu competitive market equilibrium , 2007, Math. Program..

[31]  Jie-Jun Tseng,et al.  Statistical properties of agent-based models in markets with continuous double auction mechanism , 2010, 1002.0917.

[32]  Sanmay Das,et al.  Comparing Prediction Market Structures, With an Application to Market Making , 2010, ArXiv.

[33]  Marco Ottaviani,et al.  Aggregation of Information and Beliefs in Prediction Markets , 2007 .

[34]  David M. Pennock,et al.  An Empirical Comparison of Algorithms for Aggregating Expert Predictions , 2006, UAI.