This is the evolving homepage of the Machine Learning course for
Ph.D. students at AASS, based on the contributions of the course members themselves. Please send any additions, comments or corrections to Tom Duckett: `tom.duckett@tech.oru.se`

Course Members

Seminar Series 1: Theory of Machine Learning

Seminar Series 2: Applications of Machine Learning

Additional Topics of Interest

Book

The course consists of two seminar series, to be presented by each of the course members in turn. The first series (Theory), based on Tom Mitchell's book, is compulsory and worth 3 points to Ph.D. students for attending and presenting your chosen book chapter. The second series (Applications) consists of an optional assignment, where you present the state-of-the-art in applications of ML in a subject area of your choice, worth 2 extra points (or it can any sensible related topic, to be agreed with the course leader). Alternatively, students can do a programming assignment for 2 points. All course members are also asked to contribute to this web page, as a resource for current and future students.

- Tom Duckett (course leader)

- Abdelbaki Bouguerra
- Alexander Skoglund
- Christoffer Wahlgren
- Grzegorz Cielniak
- Henrik Andreasson
- Johan Larsson
- Kjell Mårdensjö
- Linn Robertsson
- Malin Lindquist
- Marco Gritti
- Martin Magnusson

- Presenter: Tom Duckett.
- Where and when: T1210, 2005-03-10, 14:00.
- Mitchell slides: ML_chapter01.pdf
- Extra readings:
- Ulrich Nehmzow and Tom Mitchell. The Prospective Student's Introduction to the Robot Learning Problem. University of Manchester Technical Report Series. 1996.
- George Bekey. Autonomy and Learning in Mobile Robots. International Conference on Rehabilitation Robotics. 1999.
*(Gives a very nice introduction to some recent trends in robotics.)* - Claude E. Shannon. A Mathematical Theory of Communication. The Bell System Technical Journal. 1948.
*(For those interested, this is the seminal paper on information theory.)* - Group discussion: read the paper by Nehmzow and Mitchell above. Looking back nearly 10 years later, has the field of robot learning progressed? (are some of the subproblems now solved?). With hindsight, is this a complete description of the robot learning problem?

- Presenter: Christoffer Wahlgren.
- Where and when: T1210, 2005-03-17, 14:00.
- Own presentation: chapter2.pdf
- Mitchell slides: ML_chapter02.pdf Interactive Statistics
- Extra readings:
- Java applet of the candidate-elimination algorithm (Very good, but only if you know Dutch...)

- Presenter: Abdelbaki Bouguerra.
- Where and when: T1210, 2005-03-24, 14:00.
- Mitchell slides: ML_chapter03.pdf
- Extra readings:
- Floriana Esposito, Donato Malerba and Giovanni Semararo. A Comparative Analysis of Methods for Pruning Decision Trees. IEEE Trans. Pattern Analysis and Machine Intelligence. 1997.
- Karen Zita Haigh and Manuela M. Veloso. Learning Situation-Dependent Costs: Improving Planning from Probabilistic Robot Execution. Autonomous Agents. 1998.
*(This paper reports on the use of real robot navigational execution data to learn rules that help the navigation planner generate paths that are appropriate for a given situation.)*

- Presenter: Kjell Mårdensjö.
- Where and when: T1210, 2005-03-31, 14:00.
- Own presentation: kmo_ann.pdf
*"A picture (pattern) says more than a thousand words.''* - Mitchell slides: ML_chapter04.pdf
- Extra readings:
- Ted Hesselroth, Kakali Sarkar and P. Patrick van der Smagt. Neural Network Control of a Pneumatic Robot Arm, 2005.
- Java Neural Network Simulator (Tuebingen University)
- Sumeet's Neural Networks Page

- Presenter: Henrik Andreasson.
- Where and when: T1210, 2005-04-07, 14:00.
- Own presentation: chapter5.pdf
- Mitchell slides: ML_chapter05.pdf
- Extra readings:

- Presenter: Martin Magnusson.
- Where and when: T1210, 2005-04-28, 11:00.
- Own presentation: chapter6.pdf
- Mitchell slides: ML_chapter06.pdf
- Extra readings:
- Anti-Spam Web Page.
- Frank Dellaert. The Expectation Maximization Algorithm (a tutorial)
- Expectation-Maximization as lower bound maximization (alternative derivation of the EM algorithm)
- Sean Borman. The Expectation Maximization Algorithm: A short tutorial (self-contained, very mathematical)

- Presenter: Alexander Skoglund.
- Where and when: T1210, 2005-04-28, 14:00.
- Own presentation: mlcourse_chapter8.pdf
- Java applet for testing a RBF network
- Mitchell slides: ML_chapter08.pdf
- Extra readings:
- Christopher G. Atkeson, Andrew W. Moore and Stefan Schaal. Locally Weighted Learning
- Christopher G. Atkeson, Andrew W. Moore and Stefan Schaal. Locally Weighted Learning for Control
- Stefan Schaal, Christopher G. Atkeson and Sethu Vijayakumar. Scalable Techniques from Nonparametric Statistics for Real Time Robot Learning

- Presenter: Linn Robertsson.
- Where and when: T1210, 2005-05-19, 14:00.
- Mitchell slides: ML_chapter09.pdf
- Extra readings:
- J.J. Fernandez and I.D. Walker. Biologically inspired robot grasping using genetic programming. 1998.
- Introduction to GAs with Java applets

- Presenter: Johan Larsson.
- Where and when: T1210, 2005-05-19, 14:00.
- Mitchell slides: ML_chapter10.pdf
- Extra readings:
- Koren Ward, Alexander Zelinsky and Phillip McKerrow. Learning Robot Behaviours by Extracting Fuzzy Rules from Demonstrated Actions, 2001.

- Presenter: Malin Lindquist.
- Where and when: T1210, 2005-05-19, 14:00.
- Mitchell slides: ML_chapter11.pdf
- Own presentation: chapter11.pdf
- Extra readings:
- DeJong, G. and Mooney, R. Explanation-based learning: An alternative view. Machine learning, 1986, 1, 145-176.
*(Gives a good overview of explanation based learning and identifies problems in the explanation-based generalization presented in Mitchell, T.M., Keller, R., & Kedar-Cabelli, S. Explanation-based generalization: A unifying view. Machine learning,1986, 1, 47-80. An alternative approach to explanation-based learning problem is presented.)* - Thrun, S. An approach to learning mobile robot navigation. Robotics and Autonomous Systems, 1995, 15:4, 301-319.
*(Combines inductive learning and analytical learning, using an explanation-based neural network. When there is no domain theory a network is used to find it. The obtained domain theory is not perfect.)*

- Presenter: Marco Gritti.
- Where and when: T1210, 2005-05-20, 14:00.
- Mitchell slides: ML_chapter13.pdf
- Extra readings (TD-backgammon):
- Temporal Difference Learning and TD-Gammon from IBM's website.
- TD-Gammon algorithm from the book of Sutton and Barto.
- A review of the "Family FunPak" in which the backgammon computer game was contained.
- G. Tesauro. Practical Issues in Temporal Difference Learning.
*A paper from Tesauro, in which he describes in detail implementation issues and optimizations of the TD-Learning algorithm for the development of his game player.*

- Presenter: Kjell Mårdensjö.
- Where and when: T1210, 2005-05-26, 14:00.
- Own presentation: kmo_able.pdf

- Presenter: Christoffer Wahlgren.
- Where and when: T1210, 2005-06-02, 14:00.
- Own presentation: Robotic_Mapping.pdf
- Robotic mapping and the probabilistic approach:
- Bayes filter:
- Bayesian Filtering for Location Estimation (a nice magazine article on Bayes filter implementations)

- Kalman filters:
- The Kalman Filter (this is definitely where one should start looking for information)
- Peter Maybeck's book (a short introduction, excerpt from the book)

- EM algorithm: (see also references above in Chapter 6)
- S. Thrun, W. Burgard and D. Fox. A Probabilistic Approach to SLAM for Mobile Robots. Machine Learning and Autonomous Robots (joint issue). 1998.
*Describes the EM algorithm and Bayesian learning pretty clearly, in a mapping context.*

- S. Thrun, W. Burgard and D. Fox. A Probabilistic Approach to SLAM for Mobile Robots. Machine Learning and Autonomous Robots (joint issue). 1998.
- Other:

- Presenter: Alexander Skoglund.
- Where and when: T1210, 2005-06-16, 14:00.
- Own presentation: LWLforcontrol.pdf

- Presenter: Linn Robertsson.
- Where and when: T1210, 2005-06-23, 14:00.
- Own presentation: GAsforDextrous.pdf
- Extra readings:
- J.J. Fernandez and I.D. Walker. Biologically Inspired Robot Grasping Using Genetic Programming. ICRA 1998.

- Presenter: Johan Larsson.
- Where and when: T1210, 2005-06-23, 15:00.
- Own presentation: ML_for_Mines.pdf

- Presenter: Martin Magnusson.
- Where and when: T1210, 2005-06-23, 16:00.
- Own presentation: 3Dmapping.pdf

- Presenter: Henrik Andreasson
- Where and when: T1210, 2005-06-30, 14:00.
- Own presentation: ML_obj_rec.pdf

- Presenter: Malin Lindquist.
- Where and when: T1210, 2005-06-30, 15:00.
- Own presentation: SOM.pdf
- Extra readings:
- M. Zuppa, C. Distante, P. Siciliano and K.C. Persaud. Drift Counteraction with multiple self-organising maps for an electronic nose. Sensors and Actuators B, 2004.

- Presenter: Tom Duckett.
- Where and when: T1210, 2005-06-30, 16:00.
- Own presentation: Robot_Learning.pdf
- Extra readings:
- T. Dietterich. Machine Learning Research: Four Current Directions. AI Magazine, 1997.
- S. Schaal. Is imitation learning the route to humanoid robots?. Trends in Cognitive Sciences, 1999
- A.P. Shon, D.B. Grimes, C.L. Baker and R.P.N. Rao. A Probabilistic Framework for Model-Based Imitation Learning. CogSci. 2004.
- J. Weng, J. McClelland, A. Pentland, O. Sporns, I. Stockman, M. Sur and E. Thelen, Autonomous Mental Development by Robots and Animals. Science, 2000.
- M. Lopes, A. Bernardino and J. Santos-Victor. A Developmental Roadmap for Task Learning by Imitation in Humanoid Robots. AISB 2005 Symposium on Imitation in Animals and Artifacts, 2005.

Tom Mitchell. Machine Learning. McGraw-Hill. 1997. ISBN 0-07-042807-7

Book home page

Simon Haykin. Neural Networks: A Comprehensive Foundation. Prentice-Hall. 1998.

Michael A. Arbib (editor). The Handbook of Brain Theory and Neural Networks. MIT Press. 1995.

Richard S. Sutton and Andrew G. Barto, Reinforcement Learning: An Introduction. MIT Press. 1998.

David E. Goldberg. Genetic Algorithms in Search, Optimization and Machine Learning. Addisson-Wesley. 1989.

Rajesh P.N. Rao, Bruno A. Olshausen and Michael S. Lewicki (editors). Probabilistic Models of the Brain. MIT Press. 2002.