Swedish web site
Theoretical Aspects of Automated Planning2026VT, 6.0 credits
|
|
Course plan
Introduction
This course's aim is to cover a range of theoretical results related to the
field of automated planning. We will start the course by an introduction to
complexity theory for non-specialists, and follow with a series of
presentations given by the students enrolled in the course.
Each student will have to read a paper covering a theoretical result in
automated planning, and present the main results to the rest of the
participants.
The goal is to sharpen the students' presentation skills and ease, as well as
their abilities to talk about theoretical aspects of their sub-field.
Recommended for
PhD students in computer science specializing in automated planning, or sequential decision making in general
No of lectures
The course will have two introductory lectures, followed by 4 to 6 seminars, depending on the number of participants.
Goals
The goal of this course is to offer a different perspective on various topics
related to automated planning. By learning about the theoretical foundations of
the models and algorithms used in the field, the students will acquire a better
understanding of the guarantees and challenges of planning.
In addition, the students will learn to present complex theoretical results in
limited time, thus improving their science communication skills.
Prerequisites
A solid background in computer science and mathematics is required. In particular, the students would benefit from knowledge about linear algebra and discrete maths (logic, combinatorics and graph theory), but also algorithms, data structure, and complexity theory.
Organization
After the initial lectures, the course will be run as a weekly seminar in which
up to two participants present their respective papers, and lead a following
discussion on the topic.
We will meet on the first week of the course to find the time that accommodates
the most students.
Content
This course covers different aspects related to theoretical computer science
and automated planning. Such aspects include:
- Applications of automata theory to search in planning (temporally-extended
goals, factored state representation...)
- Complexity analysis of various problems in planning, with respect to the
models used
- Formal guarantees of search algorithms
- Methods for proving and certifying the unsolvability of planning tasks
- Logic applied to planning
Literature
A selection of recent papers will be provided on the course webpage. Participants are more than welcome to suggest papers beyond the provided options.
Examination
Each participant will have to choose a research article of their choice, select
a theorem shown in the paper, and present the proof. In addition, the student
should be able to discuss the broader implications of the result.
In addition, each participant is expected to take part in the discussion about
the papers that are presented at each session.
Examiner
Arnaud Lequen
Credits
6
Page responsible: Anne Moe
