Skip to main content

Machine Learning for Optimal Economic Design

  • 774 Accesses

Part of the Studies in Economic Design book series (DESI)

Abstract

This position paper anticipates ways in which the disruptive developments in machine learning over the past few years could be leveraged for a new generation of computational methods that automate the process of designing optimal economic mechanisms.

We would like to thank Yang Cai, Vincent Conitzer, Constantinos Daskalakis, Scott Kominers, Alexander Rush, and participants at seminars at the Simons Institute for the Theory of Computing, Dagstuhl, London School of Economics, IJCAI’18, AAMAS’18, The Technion, the EC’18 WADE workshop, the German Economic Association annual meeting, Google, MIT/Harvard Theory Seminar, WWW’19 workshop, and HBS for their helpful feedback.

This is a preview of subscription content, access via your institution.

Buying options

Chapter
USD   29.95
Price excludes VAT (USA)
  • DOI: 10.1007/978-3-030-18050-8_70
  • Chapter length: 21 pages
  • Instant PDF download
  • Readable on all devices
  • Own it forever
  • Exclusive offer for individuals only
  • Tax calculation will be finalised during checkout
eBook
USD   139.00
Price excludes VAT (USA)
  • ISBN: 978-3-030-18050-8
  • Instant PDF download
  • Readable on all devices
  • Own it forever
  • Exclusive offer for individuals only
  • Tax calculation will be finalised during checkout
Softcover Book
USD   179.99
Price excludes VAT (USA)
Hardcover Book
USD   179.99
Price excludes VAT (USA)
Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5

Notes

  1. 1.

    Results are known for additive i.i.d. U(0, 1) values on items (Manelli and Vincent 2006), additive, independent and asymmetric distributions on item values (Daskalakis et al. 2017; Giannakopoulos and Koutsoupias 2015; Thirumulanathan et al. 2016), additive, i.i.d. exponentially distributed item values (Daskalakis et al. 2017) and extended to multiple items (Giannakopoulos 2015), additive, i.i.d. Pareto distributions on item values (Hart and Nisan 2012), unit-demand valuations with item values i.i.d. \(U(c,c+1)\), \(c>0\) (Pavlov 2011), and unit-demand, independent, uniform and asymmetric distributions on item values (Thirumulanathan et al. 2017).

  2. 2.

    Working in increasingly general settings, relevant results on DSIC auction design include Chawla et al. (2007, 2010), Alaei (2014), Kleinberg and Weinberg (2012), Hart and Nisan (2012), Li and Yao (2013), Babaioff et al. (2014), Yao (2015), Rubinstein and Weinberg (2015), Cai et al. (2016), Cai and Zhao (2017), Dütting et al. (2017). These mechanisms are simple, and reveal the structural ingredients that are important for the design of mechanisms with good revenue properties.

  3. 3.

    IC means that no agent can benefit, in equilibrium, by misreporting its type, and can hold in a dominant-strategy equilibrium (DSIC) or a Bayes-Nash equilibrium (BIC). We will generally be interested in DSIC.

  4. 4.

    See Parkes and Wellman (2015) for a discussion on the role of AI in the mediation of economic transactions.

  5. 5.

    See Carroll (2013), Mennle and Seuken (2014), Lubin and Parkes (2012), Mennle and Seuken (2017) for some discussion of approximate notions of incentive compatibility.

  6. 6.

    For example, we could also penalize failure of weak-monotonicity (Bikhchandani et al. 2006), or insist that the implied pricing-function is agent independent (with prices to an agent that are do not depend on its report, conditioned on an allocation).

  7. 7.

    Daskalakis et al. (2014) give a complexity result for optimal mechanism design. There is also a recent literature on the sample complexity of auctions and mechanisms, including revenue-optimal auctions (Elkind 2007; Cole and Roughgarden 2014; Dughmi et al. 2014; Morgenstern and Roughgarden 2015, 2016; Huang et al. 2015; Devanur et al. 2016; Narasimhan and Parkes 2016; Gonczarowski and Nisan 2017; Cai and Daskalakis 2017).

  8. 8.

    The use of machine learning for mechanism design was earlier pioneered by Dütting et al. (2015), who use support vector machines to design payment rules for a given allocation rule (which can be designed to be scalable). But their framework can fail to even closely approximate incentive compatibility then the rule is not implementable, and does not support design objectives that are stated on payments. Earlier, Procaccia et al. (2009) studied the learnability of specific classes of voting rules, but without considering incentives; see also Xia (2013), who suggests a learning framework that incorporates specific axiomatic properties.

  9. 9.

    Deep learning, which refers typically to the use of multi-layer neural networks, has gained a great deal of attention in recent years. This is because of the existence of large data sets, the development of tool chains that make experimentation easy, optimized hardware to speed-up training (GPUs), as well as massive investment from the private sector. Whether a network is considered ‘deep’ or not is a matter of taste.

  10. 10.

    We have also explored network architectures that leverage characterization results; Myerson (1981) and Rochet (1987) for optimal auction design, and Moulin (1980) for facility location problems.

  11. 11.

    The sigmoid activation function is \(\sigma (z)\,{=}\,1/(1+e^{-z})\). The softmax activation function for item j is \( softmax _i(s_{1j}, \ldots , s_{nj}, s_{n+1,j})\,{=}\, e^{s_{ij}}/\sum _{k=1}^{n+1} e^{s_{kj}}\), where \(s_{n+1,j}\) is a dummy input that corresponds to the item not being allocated to any bidder. In another variation, we handle unit-demand valuations of bidders by using an additional set of softmax activation functions, one per agent, and taking the minimum of these item-wise and agent-wise softmax components in defining the output layer.

  12. 12.

    The output \(p_i\) that corresponds to bidder i is the amount the bidder should pay in expectation, for a particular bid profile. This can be converted into an equivalent lottery on payments, such that a bidder’s payment is no greater than her value for any realized allocation (the property of ex post IR).

  13. 13.

    In more recent work (Dütting et al. 2019) we take an adversarial-style approach, using a gradient-based approach for estimating regret for a given profile. The gradient-based approach requires that the valuation space is continuous and the utility function is differentiable, but is more scalable and stable for larger settings.

  14. 14.

    With a suitably large penalty parameter \(\rho \), the method of augmented Lagrangian is guaranteed to converge to a (locally) optimal solution to the original problem (Wright and Nocedal 1999). In practice we find that even for small values of \(\rho \) and enough iterations, the solver converges to auction designs that yield near-optimal revenue while closely satisfying the regret constraints.

  15. 15.

    A training iteration is one mini-batch gradient updated in the solver that we use for stochastic gradient descent.

  16. 16.

    Pai and Vohra (2014) design the optimal, single-item BIC auction, while Malakhov and Vohra (2008) provide the state-of-the-art result for the optimal, single-item DSIC auction (for two bidders, and with a weaker “conditional” form of DSIC). These results build on earlier results for more stylized settings (Che and Gale 1998, 2000; Maskin 2000; Laffont and Robert 1996). There are also a few approximation results for DSIC and BIC designs (Borgs et al. 2005; Bhattacharya et al. 2012; Chawla et al. 2011).

  17. 17.

    Ehlers and Gordon (2011) and Heo (2013) provide characterizations for the special case of \(K=2\) under additional assumptions. Heo (2013) assumes anonymity and an additional property, users only, which means that agents cannot influence the locations of facilities they will not use. Ehlers and Gordon (2011) assume that agents have lexmax preferences over facilities, and thus do not only care about the peak closest to them.

  18. 18.

    For the single facility location problem, it is w.l.o.g. to consider mechanisms that operate on agent peaks (Border and Jordan 1983). This extends also to a more general “voting under constraints” setting (Barberà et al. 1997). For multiple facilities there are DSIC mechanisms that do not depend only on agent peaks. For example, one can consider the example of \(K=2\) facilities and \(n=2\), where one facility is placed at the peak of agent 1 and the other at some location that depends on the shape of agent 1’s report. Still, we retain this simple representation in our current work.

  19. 19.

    In a multi-facility, dictatorial rule, each facility is determined by a separate dictatorial rule. This is equivalent to having a serial-dictatorial rule for all K facilities. In a multi-facility, percentile rule, each facility is determined by a separate percentile rule (Sui et al. 2013). A constant rule places each facility in the same location all the time.

  20. 20.

    If \(p_1, \ldots , p_n\) are the agent peaks in sorted order, then a facility at location x has percentile 0 if \(x \le p_1\), has percentile 1 if \(x \ge p_n\), and has percentile \(\frac{i-1}{n-1} + \frac{x - p_i}{(n-1)(p_{i+1} - p_i)}\) if \(p_i \le x < p_{i+1}\).

References

  • Abadi, M., Chu, A., Goodfellow, I. J., McMahan, H. B., Mironov, I., Talwar, K., et al. (2016). Deep learning with differential privacy. Proceedings of the ACM Conference on Computer and Communications Security (pp. 308–318).

    Google Scholar 

  • Alaei, S., Fu, H., Haghpanah, N., & Hartline, J. D. (2013). The simple economics of approximately optimal auctions. Proceedings of the 54th IEEE Symposium on Foundations of Computer Science (pp. 628–637).

    Google Scholar 

  • Alaei, S., Fu, H., Haghpanah, N., Hartline, J. D., & Malekian, A. (2012). Bayesian optimal auctions via multi- to single-agent reduction. Proceedings of the 13th ACM Conference on Electronic Commerce (p. 17).

    Google Scholar 

  • Alaei, S. (2014). Bayesian combinatorial auctions: Expanding single buyer mechanisms to many buyers. SIAM Journal on Computing, 43, 930–972.

    CrossRef  Google Scholar 

  • Albert, M., Conitzer, V., & Stone, P. (2017). Automated design of robust mechanisms. Proceedings of the 31st AAAI Conference on Artificial Intelligence (pp. 298–304).

    Google Scholar 

  • Aziz, H., Chan, H., Lee, B. E., & Parkes, D. C. (2018). Mechanism design without money for common goods. arXiv:1806.00960.

  • Babaioff, M., Immorlica, N., Lucier, B., & Weinberg, S. M. (2014). A simple and approximately optimal mechanism for an additive buyer. Proceedings of the 55th IEEE Symposium on Foundations of Computer Science (pp. 21–30).

    Google Scholar 

  • Barberà, S., Massó, J., & Neme, A. (1997). Voting under constraints. Journal of Economic Theory, 76, 298–321.

    CrossRef  Google Scholar 

  • Bhalgat, A., Gollapudi, S., & Munagala, K. (2013). Optimal auctions via the multiplicative weight method. Proceedings of the 14th ACM Conference on Economics and Computation (pp. 73–90).

    Google Scholar 

  • Bhattacharya, S., Goel, G., Gollapudi, S., & Munagala, K. (2012). Budget-constrained auctions with heterogeneous items. Theory of Computing, 8, 429–460.

    CrossRef  Google Scholar 

  • Bikhchandani, S., Chatterji, S., Lavi, R., Mu’alem, A., Nisan, N., & Sen, A. (2006). Weak monotonicity characterizes deterministic dominant-strategy implementation. Econometrica, 74, 1109–1132.

    CrossRef  Google Scholar 

  • Border, K. C., & Jordan, J. S. (1983). Straightforward elections, unanimity and phantom voters. Review of Economic Studies, 50, 153.

    CrossRef  Google Scholar 

  • Borgs, C., Chayes, J., Immorlica, N., Mahdian, M., & Saberi, A. (2005). Multi-unit auctions with budget-constrained bidders. Proceedings of the 6th ACM Conference on Electronic Commerce (pp. 44–51).

    Google Scholar 

  • Cai, Y., & Daskalakis, C. (2017). Learning Multi-Item Auctions with (or without) Samples. Proceedings of the 58th IEEE Symposium on Foundations of Computer Science (pp. 516–527).

    Google Scholar 

  • Cai, Y., & Huang, Z. (2013). Simple and nearly optimal multi-item auctions. Proceedings of the 24th ACM-SIAM Symposium on Discrete Algorithms (pp. 564–577).

    Google Scholar 

  • Cai, Y., & Zhao, M. (2017). Simple mechanisms for subadditive buyers via duality. Proceedings of the 49th ACM Symposium on Theory of Computing (pp. 170–183).

    Google Scholar 

  • Cai, Y., Daskalakis, C., & Weinberg, M. S. (2012a). Optimal multi-dimensional mechanism design: Reducing revenue to welfare maximization. Proceedings of the 53rd IEEE Symposium on Foundations of Computer Science (pp. 130–139).

    Google Scholar 

  • Cai, Y., Daskalakis, C., & Weinberg, S. M. (2012b). An algorithmic characterization of multi-dimensional mechanisms. Proceedings of the 44th ACM Symposium on Theory of Computing (pp. 459–478).

    Google Scholar 

  • Cai, Y., Daskalakis, C., & Weinberg, S. M. (2013a). Reducing revenue to welfare maximization: approximation algorithms and other generalizations. Proceedings of the 24th ACM-SIAM Symposium on Discrete Algorithms (pp. 578–595).

    Google Scholar 

  • Cai, Y., Daskalakis, C., & Weinberg, S. M. (2013b). Understanding incentives: Mechanism design becomes algorithm design. Proceedings of the 54th IEEE Symposium on Foundations of Computer Science (pp. 618–627).

    Google Scholar 

  • Cai, Y., Devanur, N. R., & Weinberg, S. M. (2016). A duality based unified approach to bayesian mechanism design. Proceedings of the 48th ACM Symposium on Theory of Computing (pp. 926–939).

    Google Scholar 

  • Cai, Y., & Daskalakis, C. (2015). Extreme value theorems for optimal multidimensional pricing. Games and Economic Behavior, 92, 266–305.

    CrossRef  Google Scholar 

  • Carroll, G. (2013). A Quantitative Approach to Incentives: Application to Voting Rules. Technical report, Microsoft Research and Stanford University.

    Google Scholar 

  • Caruana, R., Lou, Y., Gehrke, J., Koch, P., Sturm, M., & Elhadad, N. (2015). Intelligible models for healthcare: Predicting pneumonia risk and hospital 30-day readmission. Proceedings of the 21sth ACM Conference on Knowledge Discovery and Data Mining (pp. 1721–1730).

    Google Scholar 

  • Chawla, S., Hartline, J.D., & Kleinberg, R. D. (2007). Algorithmic pricing via virtual valuations. Proceedings of the 8th ACM Conference on Electronic Commerce (pp. 243–251).

    Google Scholar 

  • Chawla, S., Hartline, J. D. Malec, D. L., & Sivan, B. (2010). Multi-parameter mechanism design and sequential posted pricing. Proceedings of the 42th ACM Symposium on Theory of Computing (pp. 311–320).

    Google Scholar 

  • Chawla, S., Malec, D. L., & Malekian, A. (2011). Bayesian mechanism design for budget-constrained agents. Proceedings of the 12th ACM Conference on Electronic Commerce (pp. 253–262).

    Google Scholar 

  • Che, Y.-K., & Gale, I. (1998). Standard auctions with financially constrained bidders. The Review of Economic Studies, 65, 1–21.

    CrossRef  Google Scholar 

  • Che, Y.-K., & Gale, I. (2000). The optimal mechanism for selling to a budget-constrained buyer. Journal of Economic Theory, 92, 198–233.

    CrossRef  Google Scholar 

  • Chen, R., Lucier, B., Singer, Y., & Syrgkanis, V. (2017). Robust optimization for non-convex objectives. Proceedings of the 21st Conference on Neural Information Processing Systems (pp. 4708–4717).

    Google Scholar 

  • Cole, R., & Roughgarden, T. (2014). The sample complexity of revenue maximization. Proceedings of the 46th ACM Symposium on Theory of Computing (pp. 243–252).

    Google Scholar 

  • Conitzer, V., & Sandholm, T. (2002). Complexity of mechanism design. Proceedings of the 18th Conference in Uncertainty in Artificial Intelligence (pp. 103–110).

    Google Scholar 

  • Conitzer, V., & Sandholm, T. (2003). Applications of automated mechanism design. Proceedings of the 4th Bayesian Modelling Applications Workshop.

    Google Scholar 

  • Daskalakis, C., Deckelbaum, A., & Tzamos, C. (2014). The complexity of optimal mechanism design. Proceedings of the 25th Annual ACM-SIAM Symposium on Discrete Algorithms (pp. 1302–1318).

    Google Scholar 

  • Daskalakis, C., Deckelbaum, A., & Tzamos, C. (2017). Strong duality for a multiple-good monopolist. Econometrica (pp. 735–767).

    Google Scholar 

  • Daskalakis, C. (2015). Multi-item auctions defying intuition? SIGecom Exchanges, 14, 41–75.

    CrossRef  Google Scholar 

  • Devanur, N. R., Huang, Z., & Psomas, C. -A. (2016). The sample complexity of auctions with side information. Proceedings of the 48th ACM Symposium on Theory of Computing (pp. 426–439).

    Google Scholar 

  • Doshi-Velez, F., Wallace, B. C. & Adams, R. (2015). Graph-sparse LDA: A topic model with structured sparsity. Proceedings of the 29th AAAI Conference on Artificial Intelligence (pp. 2575–2581).

    Google Scholar 

  • Dughmi, S., Han, L., & Nisan, N. (2014). Sampling and representation complexity of revenue maximization. Proceedings of the 10th Conference on Web and Internet Economics (pp. 277–291).

    Google Scholar 

  • Dütting, P., Feldman, M., Kesselheim, T., & Lucier, B. (2017). Prophet inequalities made easy: stochastic optimization by pricing non-stochastic inputs. Proceedings of the 58th IEEE Symposium on Foundations of Computer Science (pp. 540–551).

    Google Scholar 

  • Dütting, P., Feng, Z., Narasimhan, H., Parkes, D. C., & Ravindranath, S. S. (2019). Optimal auctions through deep learning. Proceedings of the 36th International Conference on Machine Learning (first version, 2017)

    Google Scholar 

  • Dütting, P., Fischer, F. A., Jirapinyo, P., Lai, J. K., Lubin, B., & Parkes, D. C. (2015). Payment rules through discriminant-based classifiers. ACM Transactions on Economics and Computation, 3, 5:1–5:41.

    CrossRef  Google Scholar 

  • Ehlers, L., & Gordon, S. (2011). Strategy-Proof Provision of Two Public Goods: The LexMax Extension. Technical report, Mimeo Université de Montréal.

    Google Scholar 

  • Elkind, E. (2007). Designing and learning optimal finite support auctions. Proceedings of the 18th ACM-SIAM Conference on Discrete Algorithms (pp. 736–745).

    Google Scholar 

  • Fawzi, A., Fawzi, O., & Frossard, P. (2018). Analysis of classifiers’ robustness to adversarial perturbations. Machine Learning, 107, 481–508.

    CrossRef  Google Scholar 

  • Feng, Z., Narasimhan, H., & Parkes, D. C. (2018). Deep learning for revenue-optimal auctions with budgets. Proceedings of the 17th Conference on Autonomous Agents and Multi-Agent Systems (pp. 354–362).

    Google Scholar 

  • Fotakis, D., & Tzamos, C. (2014). On the power of deterministic mechanisms for facility location games. ACM Transactions on Economics and Computation, 2, 15:1–15:37.

    Google Scholar 

  • Fotakis, D., & Tzamos, C. (2016). Strategyproof facility location for concave cost functions. Algorithmica, 76, 143–167.

    CrossRef  Google Scholar 

  • Giannakopoulos, Y. & Koutsoupias, E. (2014). Duality and optimality of auctions for uniform distributions. Proceedings of the 15th ACM Conference on Economics and Computation (pp. 259–276).

    Google Scholar 

  • Giannakopoulos, Y. & Koutsoupias, E. (2015). Selling two goods optimally. Proceedings of the 42nd International Colloquium on Automata, Languages and Programming (pp. 650–662).

    Google Scholar 

  • Giannakopoulos, Y. (2015). Bounding the optimal revenue of selling multiple goods. Theoretical Computer Science, 581, 83–96.

    CrossRef  Google Scholar 

  • Golowich, N., Narasimhan, H., & Parkes, D.C. (2018). Deep learning for multi-facility location mechanism design. Proceedings of the 27th International Joint Conference on Artificial Intelligence (pp. 261–267).

    Google Scholar 

  • Gonczarowski, Y. A., & Nisan, N. (2017). Efficient empirical revenue maximization in single-parameter auction environments. Proceedings of the 49th ACM Symposium on Theory of Computing (pp. 856–868).

    Google Scholar 

  • Goodfellow, I. J., Bengio, Y., & Courville, A. C. (2016). Deep Learning, Adaptive Computation and Machine Learning. MIT Press: Cambridge.

    Google Scholar 

  • Goodfellow, I. J., Shlens, J., & Szegedy, C. (2015). Explaining and harnessing adversarial examples. Proceedings of the 3rd International Conference on Learning Representations.

    Google Scholar 

  • Guo, M., & Conitzer, V. (2010). Computationally feasible automated mechanism design: general approach and case studies. Proceedings of the 24th AAAI Conference on Artificial Intelligence.

    Google Scholar 

  • Guo, M., & Conitzer, V. (2009). Worst-case optimal redistribution of VCG payments in multi-unit auctions. Games and Economic Behavior, 67, 69–98.

    CrossRef  Google Scholar 

  • Hart, S., & Nisan, N. (2012). Approximate revenue maximization with multiple items. Proceedings of the 13th ACM Conference on Economics and Computation (p. 656).

    Google Scholar 

  • Heo, E. J. (2013). Strategy-proof rules for two public goods: double median rules. Social Choice and Welfare, 41, 895–922.

    CrossRef  Google Scholar 

  • Huang, Z., Mansour, Y., & Roughgarden, T. (2015). Making the most of your samples. Proceedings of the 16th ACM Conference on Economics and Computation (pp. 45–60).

    Google Scholar 

  • Kingma, D. P., & Ba, J. (2015). Adam: A method for stochastic optimization. Proceedings of the 3rd International Conference on Learning Representations.

    Google Scholar 

  • Kleinberg, R., & Weinberg, S. M. (2012). Matroid prophet inequalities. Proceedings of the 44th ACM Symposium on Theory of Computing (pp. 123–136).

    Google Scholar 

  • Laffont, J.-J., & Robert, J. (1996). Optimal auction with financially constrained buyers. Economics Letters, 52, 181–186.

    CrossRef  Google Scholar 

  • Li, X., & Yao, A. C.-C. (2013). On revenue maximization for selling multiple independently distributed items. Proceedings of the National Academy of Sciences, 110, 11232–11237.

    CrossRef  Google Scholar 

  • Lubin, B., & Parkes, D. C. (2012). Approximate strategyproofness. Current Science, 103, 1021–1032.

    Google Scholar 

  • Malakhov, A., & Vohra, R. V. (2008). Optimal auctions for asymmetrically budget constrained bidders. Review of Economic Design, 12, 245.

    CrossRef  Google Scholar 

  • Manelli, A., & Vincent, D. (2006). Bundling as an optimal selling mechanism for a multiple-good monopolist. Journal of Economic Theory, 127, 1–35.

    CrossRef  Google Scholar 

  • Maskin, E. S. (2000). Auctions, development, and privatization: efficient auctions with liquidity-constrained buyers. European Economic Review, 44, 667–681.

    CrossRef  Google Scholar 

  • Mennle, T., & Seuken, S. (2014). An axiomatic approach to characterizing and relaxing strategyproofness of one-sided matching mechanisms. Proceedings of the 15th ACM Conference on Economics and Computation (pp. 37–38).

    Google Scholar 

  • Mennle, T., & Seuken, S. (2017). Partial strategyproofness: Relaxing strategyproofness for the random assignment problem. arXiv:1401.3675v4.

  • Moosavi-Dezfooli, S., Fawzi, A., & Frossard, P. (2016). DeepFool: A simple and accurate method to fool deep neural networks. Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (pp. 2574–2582).

    Google Scholar 

  • Morgenstern, J., & Roughgarden, T. (2015). On the pseudo-dimension of nearly optimal auctions. Proceedings of the 29th Conference on Neural Information Processing Systems.

    Google Scholar 

  • Morgenstern, J., & Roughgarden, T. (2016). Learning simple auctions. Proceedings of the 29th Conference on Learning Theory (pp. 1298–1318).

    Google Scholar 

  • Moulin, H. (1980). On strategy-proofness and single-peakedness. Public Choice, 35, 55–74.

    CrossRef  Google Scholar 

  • Myerson, R. (1981). Optimal auction design. Mathematics of Operations Research, 6, 58–73.

    CrossRef  Google Scholar 

  • Narasimhan, H., & Parkes, D. C. (2016). A general statistical framework for designing strategy-proof assignment mechanisms. Proceedings of the 32nd Conference on Uncertainty in Artificial Intelligence (pp. 527–536).

    Google Scholar 

  • Narasimhan, H., Agarwal, S., & Parkes, D. C. (2016). Automated mechanism design without money via machine learning. Proceedings of the 25th International Joint Conference on Artificial Intelligence (pp. 433–439).

    Google Scholar 

  • Niepert, M., Ahmed, M., & Kutzkov, K. (2016). Learning convolutional neural networks for graphs. Proceedings of The 33rd International Conference on Machine Learning (Vol. 48, pp. 2014–2023).

    Google Scholar 

  • Orhan, A. E., & Ma, W. J. (2017). Efficient probabilistic inference in generic neural networks trained with non-probabilistic feedback. Nature Communications, 8, 138.

    CrossRef  Google Scholar 

  • Pai, M. M., & Vohra, R. (2014). Optimal auctions with financially constrained buyers. Journal of Economic Theory, 150, 383–425.

    CrossRef  Google Scholar 

  • Parkes, D. C., & Wellman, M. P. (2015). Economic reasoning and artificial intelligence. Science, 349, 267–272.

    CrossRef  Google Scholar 

  • Pavlov, G. (2011). Optimal mechanism for selling two goods. B.E Journal of Theoretical Economics, 11, 1–35.

    Google Scholar 

  • Procaccia, A. D., & Tennenholtz, M. (2013). Approximate mechanism design without money. ACM Transactions on Economics and Computation, 1, 1–26.

    CrossRef  Google Scholar 

  • Procaccia, A., Zohar, A., Peleg, Y., & Rosenschein, J. (2009). The learnability of voting rules. Artificial Intelligence, 173, 1133–1149.

    CrossRef  Google Scholar 

  • Raghu, M., Poole, B., Kleinberg, J. M., Ganguli, S., & Sohl-Dickstein, J. (2016). Survey of expressivity in deep neural networks. arXiv:1611.08083.

  • Ribeiro, M. T., Singh, S., & Guestrin, C. (2016). “Why should i trust you?”: explaining the predictions of any classifier. Proceedings of the 22nd ACM Conferecne on Knowlegde Discovery and Data Mining (pp. 1135–1144).

    Google Scholar 

  • Rochet, J.-C. (1987). A necessary and sufficient condition for rationalizability in a quasilinear context. Journal of Mathematical Economics, 16, 191–200.

    CrossRef  Google Scholar 

  • Ross, A. S., Hughes, M. C., & Doshi-Velez, F. (2017). Right for the Right Reasons: Training Differentiable Models by Constraining their Explanations. Proceedings of the 26th International Joint Conference on Artificial Intelligence (pp. 2662–2670).

    Google Scholar 

  • Rothkopf, M. H., Pekec, A., & Harstad, R. M. (1998). Computationally manageable combinational auctions. Management Science, 44, 1131–1147.

    CrossRef  Google Scholar 

  • Rubinstein, A., & Weinberg, S. M. (2015). Simple mechanisms for a subadditive buyer and applications to revenue monotonicity. Proceedings of the 16th ACM Conference on Economics and Computation (pp. 377–394).

    Google Scholar 

  • Sandholm, T., & Likhodedov, A. (2015). Automated design of revenue-maximizing combinatorial auctions. Operations Research, 63, 1000–1025.

    CrossRef  Google Scholar 

  • Shalev-Shwartz, S., & Wexler, Y. (2016). Minimizing the maximal loss: how and why. Proceedings of the 33rd International Conference on Machine learning (pp. 793–801).

    Google Scholar 

  • Smilkov, D., Thorat, N., Nicholson, C., Reif, E., Viégas, F. B., & Wattenberg, M. (2016). Embedding projector: interactive visualization and interpretation of embeddings. arXiv:1611.05469.

  • Sui, X., Boutilier, C., & Sandholm, T. (2013). Analysis and optimization of multi-dimensional percentile mechanisms. Proceedings of the 23rd International Joint Conference on Artificial Intelligence (pp. 367–374).

    Google Scholar 

  • Szegedy, C., Zaremba, W., Sutskever, I., Bruna, J., Erhan, D., Goodfellow, I. J., & Fergus, R. (2014). Intriguing properties of neural networks. Proceedings of the 2nd International Conference on Learning Representations.

    Google Scholar 

  • Thirumulanathan, D., Sundaresan, R., & Narahari, Y. (2016). Optimal mechanism for selling two items to a single buyer having uniformly distributed valuations. Proceedings of the 12th International Conference on Web and Internet Economics (pp. 174–187).

    Google Scholar 

  • Thirumulanathan, D., Sundaresan, R., & Narahari, Y. (2017): On optimal mechanisms in the two-item single-buyer unit-demand setting. arXiv:1705.01821.

  • Vinyals, O., Fortunato, M., & Jaitly, N. (2015). Pointer networks. Proceedings of the 29th Conference on Neural Information Processing Systems (pp. 2692–2700).

    Google Scholar 

  • Wang, F., & Rudin, C. (2015). Falling rule lists. Proceedings of the 18th Conference on Artificial Intelligence and Statistics.

    Google Scholar 

  • Wright, S., & Nocedal, J. (1999). Numerical optimization. Springer Science, 35, 67–68.

    Google Scholar 

  • Xia, L. (2013). Designing social choice mechanisms using machine learning. Proceedings of the 13th Conference on Autonomous Agents and Multi-Agent Systems (pp. 471–474).

    Google Scholar 

  • Yao, A. C.-C. (2015). An n-to-1 bidder reduction for multi-item auctions and its applications. Proceedings of the 26th ACM-SIAM Symposium on Discrete Algorithms (pp. 92–109).

    Google Scholar 

  • Yao, A. C.-C. (2017). Dominant-strategy versus bayesian multi-item auctions: maximum revenue determination and comparison. Proceedings of the 18th ACM Conference on Economics and Computation (pp. 3–20).

    Google Scholar 

Download references

Acknowledgements

This research is supported in part by NSF EAGER #124110.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to David C. Parkes .

Editor information

Editors and Affiliations

Rights and permissions

Reprints and Permissions

Copyright information

© 2019 Springer Nature Switzerland AG

About this chapter

Verify currency and authenticity via CrossMark

Cite this chapter

Dütting, P., Feng, Z., Golowich, N., Narasimhan, H., Parkes, D.C., Ravindranath, S.S. (2019). Machine Learning for Optimal Economic Design. In: Laslier, JF., Moulin, H., Sanver, M., Zwicker, W. (eds) The Future of Economic Design. Studies in Economic Design. Springer, Cham. https://doi.org/10.1007/978-3-030-18050-8_70

Download citation