AOS4 - Décision multicritère et sous incertitudes : introduction

Topic outline

• 

  General

  • News forum

    General news about the course
  • Exchange forum

    Forum whose main goal is to exchange and ask questions about the courses, the papers to be read for the evaluation, etc.

• 

  Course content and evaluation

  Content

  This course will be in different modules, each of them introducing a particular aspect of decision theory. This year will be divided into two modules:

  • Decision under uncertainty (S. Destercke, 12h)
  • Multi-criteria decision-making, preferences and connected aspects (K. Belahcene, 12h)

  Evaluation

  Evaluation will be as follows: by groups of at most 3 students, each group will

  • pick a scientific paper, either from the given list or proposed by the students (and validated by the teachers);
  • read it and understand it;
  • deliver a report explaining the paper in at most 6 pages (this can be through examples, implemented simulation, but also a consistent summary or any form deemed fit by the students) - at least the goal and main theme of the paper should be understandable by anyone with a high-school degree;
  • make a poster about the paper, that the group will explain in the last 2 hours of the course.

  Lecturers

  • Sébastien Destercke, Heudiasyc laboratory
  • Khaled Belahcene, Heudiasyc laboratory

  • Paper report guidelines File
• 

  Introductory notes

  Lecture slides will be provided here, once they have been polished and corrected as much as we can. If you let us know if there is some error or typos remaining, we will be thankful.

  This year course will be online, and each teacher will handle the situation to the best of its human capabilities. We ask the students to be understanding, and apologize for any (almost inevitable) decrease in pedagogical quality.

  • Introduction File
• 

  Part 1: Decision under uncertainty

  Materials concerning the part on decision under uncertainty

  Lectures in this part are done by small (and not so small) modules, with a non-linear, tree-structured way to go through them. At the end of a module, the idea is that students can choose what topic will be covered next. This means some of them will be covered in class, others will not be, but most will be provided here. In addition, the provided course trees (given in the navigation .pdf files) will lead you, when those have been done, to videos allowing students to catch up with the materials, or to visit materials not covered in class.

  • Module navigation: part 1 File
  • Module navigation: part 2 File
  • Introduction through examples (module A) File
  • Probability and expected utility: formal presentation (module B) File
  • Expected utility: Savage justification (module C1) File
  • Expected utility: De Finetti justification (module C2) File
  • Expected utility: Von Neumann justification (module C3) File
  • Expected Utility: is it always working? (module D1) File
  • Comparing probabilities without utilities (module D2) File
  • Decision under ignorance (module E1) File
  • Decision with RDU (module E2) File
  • Decision with probability sets (Module F) File
  • IP and Learning (Module G) File
  • Naive Credal Classifier (Module H1) File
• 

  Part 2: Multi-criteria decision making

  Materials concerning the part on Multi-criteria decision

  • Lesson 1: Basic notions and procedures File
  • Lesson 2: more advanced notions File
  • Tutorial on preference elicitation (author V. Mousseau) File
  • Tutorial on bi-objective optimization (author: V. Mousseau) File
• 

  Part 3: Multi-objective optimisation

  Materials concerning the part on multi-objective optimisation

  • Lecture 8 : MOO - Theory and algorithms File
  • Tutorial works MOO File
  • Lecture 9 : Multi-Objective Evolutionary Algorithms (MOEA) File
  • Paper related to Lecture 9 (for further reading) File
• 

  Non-exhaustive list of possible papers

  Selected papers

  • Diecidue, E., & Wakker, P. P. (2001). On the intuition of rank-dependent utility. Journal of Risk and Uncertainty, 23(3), 281-298.
    Topic: introduction to rank-dependent utility, able to include risk-averse behaviours
    Group: Patrick Deutschmann and Pierre-Louis Pecheux
  • Chevaleyre, Y., Koriche, F., Lang, J., Mengin, J., & Zanuttini, B. (2010). Learning ordinal preferences on multiattribute domains: The case of CP-nets. In Preference learning (pp. 273-296). Springer, Berlin, Heidelberg.
    Topic: CP-net learning
    Group: Estia Maliqari, Jerome Gaigne, Denis Demko
  • Yaman, F., Walsh, T. J., & Littman, M. L. (2010). Learning lexicographic preference models. In Preference learning (pp. 251-272). Springer, Berlin, Heidelberg.
    Group: Mathilde Roblot, Gautier Daures, Pierre Romon
  • Bandyopadhyay, Sanghamitra, et al. "A simulated annealing-based multiobjective optimization algorithm: AMOSA." IEEE transactions on evolutionary computation 12.3 (2008): 269-283.
    Group: Shengzhe Zhang, Yuhui Wang and Ziyu Lu
  • Deb, K., Pratap, A., Agarwal, S., & Meyarivan, T. A. M. T. (2002). A fast and elitist multiobjective genetic algorithm: NSGA-II. IEEE transactions on evolutionary computation, 6(2), 182-197.
    Group: Yufei Zhao, Dan Yan and Sizhe He
  • Troffaes, M. C. (2007). Decision making under uncertainty using imprecise probabilities. International journal of approximate reasoning., 45(1), 17-29.
    Topic: decision with imprecise probabilities, comparison of rules
    Group: Yiwen WANG and Yinong Qiu
  • Sébastien Destercke, Gen Yang:
    Cautious Ordinal Classification by Binary Decomposition. ECML/PKDD (1) 2014: 323-337
    Topic: cautious prediction by binary decomposition
    Group: Doris Fejza, Thomas Deroo and Elona Karaj
    
  • Q. Liu, W. Cai, J. Shen, Z. Fu X. Liu and N. Linge, A speculative approach to spatial-temporal efficiency with multi-objective optimization in a heterogeneous cloud environment, Security and Communication Networks, 2016, 9, 4002--4012.
    Topic: MapReduce, Extreme Learning Machine (ELM), Kernel ELM, Genetic Algorithms, Application to cloud computing and load balacing.
    Group: Yumeng MA, Zhishan TAO
    
  • Destercke, S. (2018). A generic framework to include belief functions in preference handling and multi-criteria decision. International Journal of Approximate Reasoning, 98, 62-77.
    Topic: including uncertainty in preference assessments, through belief function (also a Part 2 compatible paper)
    Group: Bolun CAI, Mengshen ZHU and me, Yiru SHEN

  Part 1: decision under uncertainty
  
  

  • Any survey paper by Fishburn, really (works also for Part 2 to some extent).
    Topic: order structures and decision under uncertainty (for most of them).
  • Ellsberg, D. (1961). Risk, ambiguity, and the Savage axioms. The quarterly journal of economics, 643-669.
    Topic: seminal paper indicating the paradox
    
  • Dubois, D., Fargier, H., Prade, H., & Sabbadin, R. (2009). A survey of qualitative decision rules under uncertainty. Decision‐making Process: Concepts and Methods, 435-473.
    Topic: decision theory with a qualitative perspective (no number at all), with a Savage-like justification
  • Chateauneuf, A., & Cohen, M. (2009). Cardinal Extensions of the EU Model Based on the Choquet Integral. Decision‐making Process: Concepts and Methods, 401-433
    Topic: extensions allowing to avoid the Ellsberg/Allais Paradox
    
    

  Part 2: multi-criteria decision making and preferences

  • Grabisch, M., & Labreuche, C. (2010). A decade of application of the Choquet and Sugeno integrals in multi-criteria decision aid. Annals of Operations Research, 175(1), 247-286.
    Topic: review paper about application on non-additive models of preferences
  • Grabisch, M., & Roubens, M. (2000). Application of the Choquet integral in multicriteria decision making. Fuzzy Measures and Integrals-Theory and Applications, 348-374.
    Topic: review paper on application of the Choquet integral
  • Boutilier, C., Brafman, R. I., Domshlak, C., Hoos, H. H., & Poole, D. (2004). CP-nets: A tool for representing and reasoning with conditional ceteris paribus preference statements. Journal of artificial intelligence research, 21, 135-191.
    Topic: seminal (big) paper introducing the idea of CP-net

  Part 3: Multi-Objective Optimisation (MOO)

  • T. Prasad anf N.S. Park, Multiobjective genetic algorithms for design of water distribution networks, Journal of Water Resources Planning and Management, Vol. 130, 1 (2004)
    Topic: Modeling, Reliability, Resilience, Multiobjective genetic algorithms, Pareto solutions, application.
    
  • A. Ng, S. Bandaru and M. Frantzen, Innovative design and analysis of production systems by multi-objective optimization and data mining, Procedia CIRP 50 (2016), 665--671.
    Topic: Multicriteria Decision making, MOO, Knowledge discovery, interactive dama mining, real-world production line design

  Past Year Papers (cannot be taken this year, but can provide ideas of papers to ask for).

  • Zaffalon, M., Corani, G., & Mauá, D. (2012). Evaluating credal classifiers by utility-discounted predictive accuracy. International Journal of Approximate Reasoning, 53(8), 1282.
    Topic: how to compare cautious predictions and precise predictions in classification problems
    
  • Zaffalon, M. (2002). The naive credal classifier. Journal of statistical planning and inference, 105(1), 5-21.
    Topic: cautious prediction for naive Bayesian network (imprecise probabilistic extension)
    
  • Zhengwu Wang, Yang Cai, Yuping Zeng and Jie Yu, Multi-Objective Optimization for Plug-In 4WD Hybrid Electric Vehicle Powertrain, Appl. Sci., 2019, 9, 4068; doi:10.3390/app9194068
    
  • Yaochu Jin and Bernhard Sendhoff, Pareto-Based Multiobjective Machine Learning: An Overview and Case Studies, in IEEE Transactions on Systems Man and Cybernetics Part C (Applications and Reviews), June 2008. DOI: 10.1109/TSMCC.2008.919172
    
  • Erik Dovgan, Matjaž Gams, Bogdan Filipič (2019), A Real-Time Multiobjective Optimization Algorithm for Discovering Driving Strategies. Transportation Science 53(3):695-707. https://doi.org/10.1287/trsc.2018.0872
  • Hiroki Omagari and Shin–Ichiro Higashino, Provisional-Ideal-Point-Based Multi-objective Optimization Method for Drone Delivery Problem, International Journal of Aeronautical & Space Sciences (2018) 19:262–277, https://doi.org/10.1007/s42405-018-0021-7
    
  • Bräuning, M., Hüllermeier, E., Keller, T., & Glaum, M. (2017). Lexicographic preferences for predictive modeling of human decision making: A new machine learning method with an application in accounting. European Journal of Operational Research, 258(1), 295-306.
    
  • Benabbou, N., Perny, P., & Viappiani, P. (2017). Incremental elicitation of choquet capacities for multicriteria choice, ranking and sorting problems. Artificial Intelligence, 246, 152-180.
    Topic: online optimal learning of preferences, with a minimax/robust approach
    

  • De Schuymer, B., De Meyer, H., De Baets, B., & Jenei, S. (2003). On the cycle-transitivity of the dice model. Theory and Decision, 54(3), 261-285.
    Topic: study of the transitivity properties of statistical preferences