PINGZHONG TANG
Associate Professor
IIIS, Tsinghua University
Computational Economics Lab
Office: FIT 4-608-5
100084, Beijing, China
CV
Research statement
I am a tenured associate professor at IIIS (aka. Yao class), Tsinghua University
In Tsinghua, I teach game theory and economics and computation
My current research focuses on the interdisciplinary topics between AI, multi-agent systems and economics. I enjoy both theoretical and applied problems. Examples of my past work include simple and optimal auctions, dynamic ad auctions, water right market design (fielded in Gansu, China) as well as reinforcement mechanism design (fielded in Baidu advertising and Taobao search). During PhD, I wrote computer programs to automatically discover theorems in game theory. Two theorems discovered are reported in a top game theoretical journal and the methodology invented has now become a standard technique in social choice theory. I started doing AI and multi-agent systems research back in 2003 when I was a member of the world champion robot soccer team WrightEagle.
Education
Before Tsinghua, I was a postdoc with Tuomas Sandholm at CS, CMU, 2010-2012
I completed PhD in CS, HKUST, 2010 with Fangzhen Lin and bachelor in CS, USTC, 2005
Academic visits
In Fall 2015, I spent a semester as visiting scientist at the Simons Institute, UC. Berkeley, as a participant of the Economics and Computation Program.
During PhD, I was exchange student at the EconCS group of CS at Harvard 2010 and the Multiagent lab of CS at Stanford 2008-2009.
Industry collaborations
I am a consultant for Bytedance (Tiktok, TopBuzz). In the past, I have been research consultant for Alibaba, Baidu, DiDi and Microsoft research.
News:
? I was among the MIT TR35 China in the year of 2018.
? Our paper Non-clairvoyant dynamic mechanism design is R&R at Econometrica!
? I gave the Early Career Spotlight Talk at IJCAI-2017 in Melbourne, August 2017
? Our paper on water market Stability of generalized two-sided markets with transaction thresholds has been nominated for best paper and best student paper of AAMAS-2017.
RESEARCH INTERESTS
Artificial Intelligence
Multi-agent System, Electronic Commerce, Machine Learning, Optimization, Recommendation
Economics
Mechanism Design, Auction, Market Design, Game Theory
Application
Water right market design, Kidney exchange market design, Ranking algorithm optimization for e-commerce, Ad auction revenue optimization
Publications by topic
Reinforcement mechanism design
a) Reinforcement mechanism design. IJCAI-2017, PDF.
b) Reinforcement mechanism design, with applications to dynamic reserve pricing in sponsored search auctions. (With Weiran Shen, et. al.) Working paper. PDF.
c) Reinforcement mechanism design for fraudulent behavior in e-commerce. (With Qingpeng Cai, Aris-Filos Ratzikas, Yiwei Zhang). AAAI-2018. PDF.
d) Reinforcement mechanism design for e-commerce. (With Qingpeng Cai, Aris-Filos Ratzikas, Yiwei Zhang). WWW-2018. PDF
Water right market design (Fielded in Gansu Province, China)
a) A two-phase model for trade matching and price setting in double auction water market (With Tingting Xu, JianShi Zhao, Hang Zheng). Water Resource Research. 2018 (The most prestigious journal in water resource research). PDF.
b) Stability of generalized two-sided markets with transaction thresholds. (With Zhiyuan Li, Yicheng Liu, Tingting Xu, Wei Zhan). AAMAS-2017. PDF. Nominee for both best paper and best student paper awards. Top 4 among all papers.
c) Optimizing trading assignments in water right markets. (With Yicheng Liu, Tingting Xu, Hang Zheng) AAAI-2016. Phoenix, USA. PDF.
Auctions, Revenue maximization, Internet advertising
a) Optimal Bayesian commitments in asymmetric auctions with incomplete information. (With Zihe Wang and Michael Zhang). Working paper under submission. (full version of the EC-16 paper below). PDF.
b) Optimal dynamic mechanisms with ex-post IR via bank accounts. (With Vahab Mirrokni, Renato Paes Leme, Song Zuo). Working paper. PDF
1. Ad auction workshop, ACM EC-2016
c) Automated mechanism design via neural networks. (with Weiran Shen and Song Zuo). AAMAS-2019. PDF
d) Buyer signaling games in auctions. (With Weiran Shen and Yulong Zeng). AAMAS-2019. PDF.
e) Making money from what you know ? how to sell information. (With Shani Alkoby, David Sarne, Zihe Wang). AAAI-2019. PDF.
f) Optimal dynamic auctions are virtual welfare maximizers. (With Vahab Mirrokni, Renato Paes Leme, Song Zuo). AAAI-2019. PDF
g) The price of prior dependence in auctions. (With Yulong Zeng). ACM EC 2018.
1. Former version at AGT and Data science workshop ACM EC-2016 under the title How to manipulate truthful prior-dependent mechanisms?. PDF.
h) Non-clairvoyant dynamic mechanism design. (With Vahab Mirrokni, Renato Paes Leme, Song Zuo). ACM EC 2018. Full version R&R at Econometrica. SSRN
i) A Closed-Form Characterization of Buyer Signaling Schemes in Monopoly Pricing. Weiran Shen, Pingzhong Tang, Yulong Zeng, AAMAS 2018. PDF.
j) Buyer optimal distributions. Weiran Shen, Pingzhong Tang, Yulong Zeng, AAMAS 2018. PDF.
k) Theory and Practice of revenue optimal mechanism design. (with Zihe Wang). IJCAI-2017, Tutorial.
l) Optimal mechanisms with simple menus. (with Zihe Wang), Journal of Mathematical economics, 2017. PDF link.
1. Subsume the EC-14 paper below.
m) Practical versus optimal mechanisms. (With Weiran Shen), AAMAS-2017. PDF.
n) Fans economy and all-pay auctions with proportional allocations. (With Yulong Zeng, Song Zuo). AAAI-2017. PDF.
o) Dynamic auctions with bank accounts. (With Vahab Mirrokni, Renato Paes Leme, Song Zuo). IJCAI -2016. New York. USA. PDF.
p) Optimal auctions for negatively correlated items. (With Zihe Wang) ACM EC-2016. PDF.
q) Optimal commitments in auctions with incomplete information. (with Zihe Wang, Michael Zhang) ACM EC-2016. PDF.
r) Discrete action spaces cause little loss in single-item auctions. (With Yicheng Liu) Extended abstract AAMAS-2016. PDF
s) Optimal auctions for partially rational bidders. (with Zihe Wang) IJCAI-2015, Buenos Aires, Argentina. PDF.
t) Optimal mechanisms with simple menus. (with Zihe Wang), ACM EC-2014. Palo Alto, USA.
u) Optimal Auctions for Spiteful Bidders. (With Tuomas Sandholm). AAAI-2012, July, Toronto, Canada. PDF.
v) Mixed Bundling Auctions with Reserve Prices. (With Tuomas Sandholm). AAMAS-2012, June, Valencia, Spain. PDF.
1. Also presented at Informs-11, Charlotte, USA.
w) Approximating optimal combinatorial auctions for complements using restricted welfare maximization. (with Tuomas Sandholm). In IJCAI-2011, Barcelona, Spain. PDF.
1. Also presented at Informs-2011, Charlotte, USA.
2. Also presented ACM EC-2011 Workshop on Bayesian Mechanism Design (WBMD), June, 2011, San Jose, CA.
Applied mechanism design and optimization: matchings, water right market, kidney exchange, lung exchange, digital good exchange,
d) Optimal vehicle dispatching schemes via dynamic pricing. (With Mengjing Chen, Weiran Shen, Song Zuo) Working paper. PDF.
1. 2-page poster version at WWW-2018
e) Learning the optimal strategy to commit to. (With Binghui Peng, Weiran Shen, Song Zuo). AAAI-2019.
f) Coalition manipulations of the Gale-Shapley algorithm. (With Yuan Deng, Weiran Shen). AAAI-2018. PDF
g) Reinforcement mechanism design for fraudulent behavior in e-commerce. (With Qingpeng Cai, Aris-Filos Ratzikas, Yiwei Zhang). AAAI-2018. PDF.
h) Coalitional Permutation Manipulations in the Gale-Shapley Algorithm. Yuan Deng, Weiran Shen, Pingzhong Tang, AAMAS 2018. PDF.
i) Ranking mechanism design for price-setting agents in e-commerce. Qingpeng Cai, Pingzhong Tang, Yulong Zeng. AAMAS, 2018. PDF.
j) Balanced outcomes in wage bargaining. Pingzhong Tang, Dingli Yu. AAMAS, 2018. PDF.
k) Reinforcement mechanism design. IJCAI-2017, PDF.
l) Efficient near-optimal algorithms for barter exchange. (With Zhipeng Jia, Ruosong Wang, Hanrui Zhang). AAMAS-2017. PDF.
m) Mechanism design for personalized recommender systems. (With Qingpeng Cai, Aris-Filos Ratsikas, Chang Liu) ACM Recsys-2016. PDF. Tsinghua News.
n) Digital good exchanges. (With Wenyi Fang, Song Zuo). IJCAI -2016. New York. USA. PDF.
1. Abstract presented in AAMAS-16
o) Facility location with minimax envy. (With Qingpeng Cai, Aris-Filos Ratsikas). IJCAI -2016. New York. USA. PDF.
p) Mechanism design and implementation for lung exchange. (with Suiqian Luo) IJCAI-2015, Buenos Aires, Argentina. PDF. Media press paper. Top 4 among all papers
q) Mechanism design for resource allocation with applications to centralized multi-commodity routing. (with Qipeng Liu, Yicheng Liu) Extended abstract, AAMAS-2015, Istanbul, Turkey. Full version.
r) Randomized assignments for barter exchanges: fairness vs. efficiency. (With Wenyi Fang, Aris Filos-Ratskas, Soren Stiil-Fredriksen, Song Zuo). ADT-2015, Kentucky, USA. PDF.
s) Internally stable matchings and exchanges. (with Yicheng Liu and Wenyi Fang). AAAI-2014. Quebec City, Canada. PDF.
1. Presented at Informs-14
t) Egalitarian Pair-wise Kidney Exchange: Fast Algorithms via Linear Programming and Parametric Flow. (With Jian Li, Yicheng Liu). AAMAS-2014. Paris, France. PDF.
u) Mechanism design for route allocation in multiple-commodity network. (With Qipeng Liu and Yicheng Liu). AAMAS-2014. Paris, France. PDF.
v) Approximation of barter exchanges with cycle length constraints. Working paper. PDF.
Beyond Nash equilibrium: computation in games, bounded rationality in repeated games
a) Unit-sphere games. (With Hanrui Zhang). International Journal of Game Theory, 2017 PDF.
b) K-memory strategies in repeated games. (With Lijie Chen, Fangzhen Lin, Kangning Wang, Shiheng Wang, Ruosong Wang). AAMAS-2017 extended abstract.
c) Computational issues in time-inconsistent planning. (With Yifeng Teng, Zihe Wang, Shengke Xiao, Yichong Xu). AAAI-2017. PDF.
d) Bounded rationality of restricted Turing machines. (with Lijie Chen, Ruosong Wang), AAAI-2017. PDF. Full version.
1. Subsume the AAMAS extended abstract below
e) Complexity and algorithms of K-implementation. (With Yuan Deng, Shuran Zheng). AAMAS-2016. Singapore, PDF.
f) Bounded rationality of restricted Turing machines. (with Lijie Chen), Extended abstract AAMAS-2015, Istanbul, Turkey.
g) Optimal machine strategy to commit to in two-person repeated games. (with Song Zuo). AAAI-2015, Austin, USA. PDF
Worst case analysis of mechanism design, online mechanism design
1. Efficient mechanism design for online scheduling (Extended abstract). (With Bo Zheng et. al.) IJCAI-2017.
2. Efficient mechanism design for online scheduling. (With Bo Zheng et. al.). Journal of AI Research, 2016. Link.
3. Online non-preemptive story scheduling in web advertising. (With Tie-Yan Liu, Weidong Ma, Tao Qin, Guang Yang, Bo Zheng). AAMAS-2016, Singapore, PDF.
4. The multi-shop ski rental problem. (With Lingqing Ai, Xian Wu, Lingxiao Huang, Longbo Huang, Jian Li). ACM Sigmetrics-2014. Austin, USA. PDF.
Reinforcement learning
1. Warm-up cold start advertisement: improve CTR predictions via learning to learn ID embeddings. (With Feiyang Pan) SIGIR-2019. PDF.
2. Policy gradient for contextual recommendation. (With Qingpeng Cai, Qing He, Feiyang Pan, Fuzhen Zhuang.) WWW-2019. PDF.
3. A deep reinforcement learning framework for rebalancing dockless bike sharing system. (With Qingpeng Cai, Longbo Huang, Zhixuan Huang, Ling Pan). AAAI-2019. PDF.
4. Policy optimization with model based exploration. (With Qingpeng Cai, Qing He, Feiyang Pan, Pingzhong Tang) AAAI-2019. PDF
Computer-aided theorem discovery in economic theory (PhD dissertation work)
a) Discovering Theorems in Game Theory: Two-Person Games with Unique Nash Equilibria Payoff. (With Fangzhen Lin). Artificial Intelligence, 2011.PDF.Link.
1. Using AI techniques to automatically discover and prove theorems in game theory
b) Computer-aided Theorem Discovery - A New Adventure and its Application to Economic Theory. PhD dissertation, HKUST, 2010. PDF.
c) Two Equivalence Results for Two-person Strict Games. (With Fangzhen Lin) Games and Economic Behavior, 2011. PDF. Link..
1. Potential games and Super-modular games are equivalent
d) Discovering Theorems in Game Theory: Two-Person Games with Unique Nash Equilibria Payoff. (With Fangzhen Lin). In IJCAI-2009, July, Pasadena, USA. PDF.
e) Computer Aided Proofs of Arrows and Other Impossibility Theorems. (With Fangzhen Lin) Artificial Intelligence, 2009. PDF. Link.
1. This paper initiates automated theorem discovery in economics
f) Coalitional Structure of the Muller-Satterthwaite Theorem. (With Tuomas Sandholm). CoopMas-2012, June, Valencia, Spain. PDF.
g) Computer Aided Proofs of Arrows and Other Impossibility Theorems. (With Fangzhen Lin) In AAAI-2008, July, Chicago, USA. PDF
h) Computer-Aided Proofs of Theorems in Implementation Theory. (With Fangzhen Lin) Draft. PDF
Game theory in sports: Team competition
1. On the power of dominated players in team competitions. (With Kai Jin, Shiteng Chen). AAMAS-2016. Singapore, PDF.
2. Designing Competitions between Teams of Individuals. (With Yoav Shoham and Fangzhen Lin). Artificial Intelligence, 2010. PDF. Link.
3. Team Competition. (With Yoav Shoham and Fangzhen Lin) In AAMAS-2009, May, Budapest, Hungary. PDF
4. Two-person Bridge. (With Yiling Chen) Working paper.
Social choice theory, Voting
1. Bayesian vote manipulation: optimal strategies and impact on welfare. (with Craig Boutilier, Tyler Lu, Ariel Procaccia), UAI-2012, Catalina Island, US. PDF
2. A Framework for Quantitative Evaluation of Voting Rules. (With Mike Munie, Yoav Shoham). In Logic, Game Theory and Social Choice (LGS6), 2009. PDF
Survey articles in Chinese
1. Economics and Computation. (in Chinese) Communication of CCF, 2017. PDF.
2. Review of the AAMAS-16 conference. (in Chinese, with Bo An) Communication of CCF, 2016. PDF.
3. Computational economics and optimal mechanism design. (in Chinese) Communication of CCF, 2013. PDF.
Students
PhD students:
Yulong Zeng (5th year) Song Zuo (5th year)
Qingpeng Cai (4th year) Weiran Shen (4th year)
Shenke Xiao (3rd year) Mengjing Chen (2nd year)
Xun Wang (1st year) Yadong Xu (1st year)
Master students:
Yuanqi Li (1st year)
Alumni:
1. Zihe Wang PhD graduated in 2016. PhD dissertation: Geometric approaches to auction design. First job: Shanghai Univ. of Fina. and Econ.
2. Bo Zheng PhD graduated in 2016. PhD dissertation: Incentive compatible online scheduling for cloud computing. First job: Government of Ningbo
3. Wenyi Fang Master graduated in 2016. First job: China construction bank
4. Yicheng Liu Master graduated in 2016. Pony.ai
5. Suiqian Luo Master graduated in 2016. Guazi
Undergraduate students:
Yao Class 2019 Dingli Yu
Yao Class 2017: Hanrui Zhang (PhD at Duke), Shuran Zheng (PhD at Harvard)
Yao Class 2016: Yu Xia (PhD student at MIT), An Yi (Master student at UCSD)
Yao Class 2015: Yuan Deng (PhD student at Duke) Qipeng Liu (PhD student at Princeton) Yifeng Teng (PhD student at U-Wisconsin) Yichong Xu (PhD student at CMU)
Yao Class-2014: Weiyi Chen (master student, NYU) Ning Jiang(Master student, UMich) Junxing Wang (PhD student at CMU) Qianru Zhu (Master at CMU)
Projects
1. NSFC-ISF joint project (China - Israel), Information brokers in multi-agent systems and mechanism design. 2015 - 2018. Principal investigator.
2. NSFC project. Optimal mechanism design: two computational approaches. 2014 - 2016. Principal investigator.
3. Tsinghua initiative research program. Optimal mechanism design. 2014 ? 2016. Principal investigator.
4. National 1000-youth program. 2014. Principal investigator.
