Postgraduate Course: Decision Making in Robots and Autonomous Agents (INFR11090)
|School||School of Informatics
||College||College of Science and Engineering
|Credit level (Normal year taken)||SCQF Level 11 (Postgraduate)
||Availability||Available to all students
|Summary||This course is intended as a specialized course on models and techniques for decision making in autonomous agents, such as intelligent robots, that must function in rich interactive settings involving environments with other agents and people.
This course will cover decision theoretic algorithms, interactive decision making including game theoretic techniques, learning in games and social settings, as well as selected topics involving decentralized systems. We will also look at aspects of human decision making, both to ask what people actually do and to consider what agents must do in light of this.
Issues of intelligent and fluid interaction by autonomous robots/agents, operating in environments including other strategic agents (either other autonomous agents or people), are becoming increasingly more important - with the advent of systems that routinely embody rich and sophisticated multi-modal interfaces, making it possible for us to now consider issues of interactive behaviour. At the same time but from a seemingly opposite perspective, 'market design' approaches are becoming increasingly more
suitable to the needs of collections of individually simple robots and agents (and people) that must work together on sophisticated large scale tasks.
The content of this course has connections to other courses within our existing curriculum, such as Reinforcement Learning and Algorithmic Game Theory. A noteworthy difference is that this course will focus more heavily on issues of modelling - how tasks associated with robotics and autonomous agents could/should be expressed and analysed using the formal language of these models, and also have more coverage of learning and potential connections to mechanisms of (boundedly rational) human decision making. This course will be self contained, discussing salient algorithmic techniques associated with some of the major models being considered. However, we expect this knowledge to be complemented by the more detailed discussion of techniques in the Reinforcement Learning and Algorithmic Game Theory and its Applications. Similarly, students will benefit from prior exposure to robotics at the level of the Robotics:Science and Systems (or some equivalent exposure to autonomous agent design), which provides the perspective necessary to fully appreciate the concerns of this course.
The DMR course will cover the following major themes:
- Problems involving interaction: Strategically rich human-robot interaction; Teams of autonomous agents; Market design
- Survey of existing models of interaction: from psychology, cognitive science and machine learning
- The utility maximization framework of decision theory
- Bandit problems, online learning and related models (e.g., matching problems)
- Markov Decision Processes and variants
Interactive Decision Making:
- Tools and techniques of game theoretic models
- Game theoretic models with incomplete information; models such as Interactive POMDP
- Repeated interaction
- Models of bargaining and negotiation (including the incomplete information case)
- Strategic learning in games
Mechanism Design and Related Topics in Decentralized Systems:
- Introduction to mechanism design and social choice
- Learning and mechanism design
- Graphical games, coordination games and social learning models
- Special topics: models of asymmetric information and privacy
Human Decision Making and Behavioural Issues:
- Behavioural aspects of human decision making - how real people think about risk, games, etc.
- Reconciling behavioural findings with formal models
Entry Requirements (not applicable to Visiting Students)
||Other requirements|| This course is open to all Informatics students including those on joint degrees. For external students where this course is not listed in your DPT, please seek special permission from the course organiser.
Prior exposure to mathematical models; Multivariate Calculus, Probability (expectation, conditional probability) & Stochastic Processes, principles of optimization (linear programming, gradient decent)
Ability to program in a high level environment such as Matlab, or a programming language such as Java/C++.
Information for Visiting Students
Course Delivery Information
|Academic year 2014/15, Available to all students (SV1)
|Learning and Teaching activities (Further Info)
Lecture Hours 18,
Summative Assessment Hours 2,
Programme Level Learning and Teaching Hours 2,
Directed Learning and Independent Learning Hours
|Assessment (Further Info)
|Additional Information (Assessment)
||You should expect to spend approximately 25 hours on the coursework for this course.
If delivered in semester 1, this course will have an option for semester 1 only visiting undergraduate students, providing assessment prior to the end of the calendar year.
||Hours & Minutes
|Main Exam Diet S2 (April/May)||2:00|
| - formulate practical problems involving interaction (e.g., human-robot interaction) in the language of decision and game theory
- analyze and evaluate conceptual problems with decision models involving multiple agents
- analyze and implement selected learning algorithms that consider incomplete information and partial observability
- demonstrate understanding of key issues related to decision making in humans; identify when, why and how standard models fail to capture real behaviour
|I. Gilboa, Theory of Decision Under Uncertainty, Cambridge University Press, 2009.|
H.P. Young, Strategic Learning and its Limits, Oxford University Press, 2004.
N. Nisan, T. Roughgarden, E. Tardos, V.V. Vazirani, Algorithmic Game Theory, Cambridge University press, 2007.
P.W. Glimcher, Foundations of Neuroeconomic Analysis, Oxford University Press, 2011.
|Course organiser||Dr Subramanian Ramamoorthy
Tel: (0131 6)50 9969
|Course secretary||Ms Katey Lee
Tel: (0131 6)50 2701
© Copyright 2014 The University of Edinburgh - 12 January 2015 4:12 am