UM-SJTU JI
News:
  • Openings for master, PhDs, postdocs, research associates or interns: If you're interested in my research, please send me your CV.
  • Prospective students: here are for some previous projects I advised.
  • Feb. 2019: I will serve as a PC member for UAI 2019
  • Feb. 2019: I will serve as a PC member for PRICAI 2019
  • Feb. 2019: I will serve as a PC member for JFPDA 2019
  • Feb. 2019: I will serve as a reviewer for NeurIPS 2019
  • Jan. 2019: one paper accepted at WWW 2019
  • Jan 2019: I will serve as a senior PC member for IJCAI 2019
  • Dec. 2018: I will serve as a reviewer for ICML 2019
  • June 2018: I will serve as a senior PC member for AAAI 2019
  • May 2018: I will serve as a PC member for ACML 2018
  • Apr. 2018: I will serve as a reviewer for DA2PL 2018
  • Apr. 2018: I will serve as a reviewer for NIPS 2018
  • Mar. 2018: I will serve as a PC member for PRICAI 2018, special track on RL
  • Jan. 2018: I will serve as a chair of the demonstration track for IJCAI-ECAI 2018

The high-level goals of our group are to help build decision-support systems (recommender systems, decision aid, search...) and automatic decision systems (autonomous agents, robots, traders...). Our research work lies at the intersection of artificial intelligence, decision theory, machine learning and operations research. In particular, it deals with (sequential) decision making under uncertainty, multiobjective optimization, preference elicitation/learning.

Here's a classification of some recent work (downloadable here or send me an email):

Applications
  • Energy related domains
    • Decomposition Methods for Distributed Optimal Power Flow: Panorama and Case Studies of the DC Model - Amini, M. Hadi; Bahrami, Shahab; Kamyab, Farhad; Mishra, Sakshi; Jaddivada, Rupamathi; Boroojeni, Kianoosh; Weng, Paul; Xu, Yinliang; Classical and Recent Aspects of Power System Optimization 2018
    • Hierarchical Electric Vehicle Charging Aggregator Strategy Using Dantzig-Wolfe Decomposition - Amini, M. Hadi; McNamara, Paul; Weng, Paul; Karabasoglu, Orkun; Xu, Yinliang; IEEE Design & Test 2017
    • Optimal Threshold Policies for Robust Data Center Control - Weng, Paul; Qiu, Zeqi; Costanzo, John; Yin, Xiaoqi; Sinopoli, Bruno; International Conference on Advanced Engineering – Theory and Applications 2017
  • Social Network and Recommender systems
    • Dual Graph Attention Networks for Deep Latent Representation of Multifaceted Social Effects in Recommender Systems - Dual Graph Attention Networks for Deep Latent Representation of Multifaceted Social Effects in Recommender Systems; WWW 2019
    • Adversarial Training Model Unifying Feature Driven and Point Process Perspectives for Event Popularity Prediction - Wu, Qitian; Yang, Chaoqi; Zhang, Hengrui; Gao, Xiaofeng; Weng, Paul; Chen, Guihai; ACM International Conference on Information and Knowledge Management 2018
  • Shortest Path
    • Finding Risk-averse Shortest Path with Time-dependent Stochastic Costs - Li, Dajian; Weng, Paul; Karabasoglu, Orkun; Multi-Disciplinary International Workshop on Artificial Intelligence, Lecture Notes in Artificial Intelligence 2016 Best paper award
  • Computer vision
    • Representing Relative Visual Attributes with a Reference-Point-Based Decision Model - Law, Marc; Weng, Paul; International Conference on Pattern Recognition 2018
  • Argumentation as sequential decision-making
    • Mediation of Debates with Dynamic Argumentative Behaviors - Hadoux, Emmanuel; Beynier, Aurélie; Maudet, Nicolas; Weng, Paul; International Conference on Computational Models of Argument 2018
    • Optimization of probabilistic argumentation with Markov Decision Models - Hadoux, Emmanuel; Beynier, Aurélie; Maudet, Nicolas; Weng, Paul; Hunter, Anthony; International Joint Conference on Artificial Intelligence 2015
Multiobjective/multicriteria decision-making
  • Multi-objective Bandits: Optimizing the Generalized Gini Index - Busa-Fekete, Róbert; Szörenyi, Balázs; Weng, Paul; Mannor, Shie; International Conference on Machine Learning 2017
  • An Efficient Primal-Dual Algorithm for Fair Combinatorial Optimization Problems - Nguyen, Viet Hung; Weng, Paul; International Conference on Combinatorial Optimization and Applications 2017
  • A Compromise Programming Approach to Multiobjective Markov Decision Processes - Ogryczak, Wlodzimierz; Perny, Patrice; Weng, Paul; International Journal of Information Technology and Decision Making 2013 12 pp. 1021-1053
  • Approximation of Lorenz-Optimal Solutions in Multiobjective Markov Decision Processes - Perny, Patrice; Weng, Paul; Goldsmith, Judy; Hanna, Josiah; Conference on Uncertainty in Artificial Intelligence 2013
  • On WOWA Rank Reversal - Ogryczak, Wlodzimierz; Perny, Patrice; Weng, Paul; International Conference on Modelling Decisions for Artificial Intelligence 2012 7647 pp. 66-77 LNAI
  • A Compromise Programming Approach to Multiobjective Markov Decision Processes - Ogryczak, Wlodzimierz; Perny, Patrice; Weng, Paul; International Conference on Multicriteria Decision Making 2011 pp. 167-167
  • On Minimizing Ordered Weighted Regrets in Multiobjective Markov Decision Processes - Ogryczak, Wlodzimierz; Perny, Patrice; Weng, Paul; International Conference on Algorithmic Decision Theory 2011 6992 pp. 190-204 LNAI
  • On finding compromise solutions in multiobjective Markov decision processes - Perny, Patrice; Weng, Paul; European Conference on Artificial Intelligence Multidisciplinary Workshop on Advances in Preference Handling 2010 pp. 55-60
Qualitative/ordinal decision-making
  • Optimizing Quantiles in Preference-based Markov Decision Processes - Gilbert, Hugo; Weng, Paul; Xu, Yan; AAAI Conference on Artificial Intelligence 2017
  • Model-Free Reinforcement Learning with Skew-Symmetric Bilinear Utilities - Gilbert, Hugo; Zanuttini, Bruno; Viappiani, Paolo; Weng, Paul; Nicart, Esther; International Conference on Uncertainty in Artificial Intelligence 2016
  • From Preference-based to Multiobjective Sequential Decision-making - Weng, Paul; Multi-Disciplinary International Workshop on Artificial Intelligence, Lecture Notes in Artificial Intelligence 2016
  • Qualitative Multi-Armed Bandits: A Quantile-Based Approach - Szorenyi, Balazs; Busa-Fekete, Robert; Weng, Paul; Hüllermeier, Eyke; International Conference on Machine Learning 2015
  • Preference-based Reinforcement Learning: Evolutionary Direct Policy Search using a Preference-based Racing Algorithm - Busa-Fekete, Robert; Szorenyi, Balazs; Weng, Paul; Cheng, Weiwei; Hüllermeier, Eyke; Machine Learning 2014 97 pp. 327-351
  • Preference-based Evolutionary Direct Policy Search - Busa-Fekete, Robert; Szorenyi, Balazs; Weng, Paul; Cheng, Weiwei; Hüllermeier, Eyke; ICRA Autonomous Learning Workshop 2013
  • Top-k Selection based on Adaptive Sampling of Noisy Preferences - Busa-Fekete, Robert; Szorenyi, Balazs; Weng, Paul; Cheng, Weiwei; Hüllermeier, Eyke; International Conference on Machine Learning 2013
  • Ordinal Decision Models for Markov Decision Processes - Weng, Paul; European Conference on Artificial Intelligence 2012
  • Committee Selection with a Weight Constraint Based on a Pairwise Dominance Relation - Delort, Charles; Spanjaard, Olivier; Weng, Paul; International Conference on Algorithmic Decision Theory 2011 6992 pp. 28-41 LNAI
  • Markov Decision Processes with Ordinal Rewards: Reference Point-Based Preferences - Weng, Paul; International Conference on Automated Planning and Scheduling 2011 pp. 282-289
Preference elicitation/learning in sequential decision-making
  • Reducing the Number of Queries in Interactive Value Iteration - Gilbert, Hugo; Spanjaard, Olivier; Viappiani, Paolo; Weng, Paul; International Conference on Algorithmic Decision Theory 2015 LNCS
  • Teacher-Student Framework: a Reinforcement Learning Approach - Zimmer, Matthieu; Viappiani, Paolo; Weng, Paul; AAMAS Workshop Autonomous Robots and Multirobot Systems 2014
  • Interactive Q-Learning with Ordinal Rewards and Unreliable Tutor - Weng, Paul; Busa-Fekete, Robert; Hüllermeier, Eyke; ECML/PKDD Workshop Reinforcement Learning with Generalized Feedback 2013
  • Interactive Value Iteration for Markov Decision Processes with Unknown Rewards - Weng, Paul; Zanuttini, Bruno; International Joint Conference on Artificial Intelligence 2013 pp. 2415-2421
Sequential decision-making in non-stationary environments
  • Sequential Decision-Making under Non-stationary Environments via Sequential Change-point Detection - Hadoux, Emmanuel; Beynier, Aurélie; Weng, Paul; First International Workshop on Learning over Multiple Contexts @ ECML 2014
  • Solving Hidden-Semi-Markov-Mode Markov Decision Problems - Hadoux, Emmanuel; Beynier, Aurélie; Weng, Paul; International Conference on Scalable Uncertainty Management 2014
General framework and decision models for sequential decision-making
  • Solving MDPs with Skew Symmetric Bilinear Utility Functions - Gilbert, Hugo; Spanjaard, Olivier; Viappiani, Paolo; Weng, Paul; 24th International Joint Conference on Artificial Intelligence 2015
  • Markov Decision Processes with Functional Rewards - Spanjaard, Olivier; Weng, Paul; 7th Multi-Disciplinary International Workshop on Artificial Intelligence 2013 8271 pp. 269-280 Lecture Notes in Computer Science Best paper award
  • Algebraic Markov Decision Processes - Perny, Patrice; Spanjaard, Olivier; Weng, Paul; 19th International Joint Conference on Artificial Intelligence 2005 pp. 1372-1377
Study of qualitative/ordinal decision models
  • Axiomatic Foundations of Generalized Qualitative Utility - Weng, Paul; Multi-Disciplinary International Workshop on Artificial Intelligence 2013 LNAI
  • An Axiomatic Approach to Qualitative Decision Theory with Binary Possibilistic Utility - Weng, Paul; European Conference on Artificial Intelligence 2006 pp. 467-471
  • Axiomatic Foundations for a Class of Generalized Expected Utility: Algebraic Expected Utility - Weng, Paul; International Conference on Uncertainty in Artificial Intelligence 2006 pp. 520-527
  • Qualitative Decision-Making Under Possibilistic Uncertainty: Toward More Discriminating Criteria - Weng, Paul; International Conference on Uncertainty in Artificial Intelligence 2005 pp. 615-622