KEPS 2011 - The 3rd workshop on learning and planning

Planning and Learning

The 3rd workshop on learning and planning aims to provide a forum for the discussion of issues surrounding the use of learning techniques in planning, continuing the lineage of the events of ICAPS 2007 and 2009. This year, the workshop will be held in parallel to the learning track of the International Planning Competition, and will be a suitable forum to also present the ideas behind the planners running the competition.

Call For Papers

Planning has been defined as the process of thinking before acting, while machine learning has been defined as the process of improving with experience. Although these two areas seem to be quite different, machine learning is actually very useful in all stages of planning, from learning models for planning problems, to learning domain-specific search control, and even online learning during problem solving. This workshop aims to provide a forum for discussing current advances in using learning techniques for all areas of planning.

Automated planners traditionally reason about correct and complete descriptions of planning tasks. These descriptions include models of the actions that can be carried out in the environment together with a specification of the state of the environment and the goals to achieve. In the real-world, actions may result in numerous outcomes, the perception of the state of the environment may be partial and the goals may not be completely defined. Specifying planning tasks from scratch under these conditions becomes complex, even for experts.

Furthermore, despite great progress that had been made in the field of domain independent planning - powerful domain-independent heuristics, useful landmarks analysis or novel propagators for use in a planning-as-CSP framework, to name but a few - hand-crafted domain-specific planners tend to outperform general domain independent planners. The drawback of such guidance is the amount of human effort needed to produce suitable guidance for each domain - the key motivation behind domain-independent approaches.

Machine learning can be used to help with both of these problems. The aim is to eliminate the human bottleneck by automating the process of acquiring domain-specific knowledge (either in the form of a domain model, or as search guidance). In doing so, the system as a whole becomes domain independent once again - the learning system can be used on each domain of interest.

This workshop aims to provide a forum for discussing issues surrounding the use of learning techniques in planning, continuing the lineage of the events of ICAPS 2007 and 2009. The topics that will be covered include, but are not limited to:

Approaches to learning search guidance
Approaches to learning of planning models - action modelling, model-lite planning, ...
Representation of learned knowledge - control rules, heuristics, macro-actions....
Applying learning to portfolio-based planners
Hybrid learned-guidance--generic-heuristic search
Applications of planning and learning
Learning during planning
Future Challenges for the IPC Learning Part
Using machine learning in activity/plan/goal recognition
The impact of problems sets on what can be learned
We invite contributions from researchers who have considered the application of learning to planning. We also welcome theoretical contributions considering the expressive power and/or limitations of various forms of learned knowledge representation. Additionally, we also welcome descriptions of the systems participating in the learning part of IPC-2011.

Important Dates

Submission Deadline: February 11, 2011
Notifications and Technical Program: March 11, 2011
Workshop Date: June 12th or 13th, 2011
Submission Procedure

We ask authors to submit technical papers in PDF format. Papers should be formatted in accordance with the AAAI style template and may be at most 8 pages long, including figures and bibliography. Visit the url http://www.aaai.org/Publications/Author/author.php for formatting instructions.

Please note that all submitted papers will be carefully peer-reviewed by multiple reviewers, and that low-quality or off-topic papers will not be accepted.

Detailed submission instructions will be published at a later date.

Organizing Committee

Sergio Jiménez Celorrio
Planning and Learning Group
Computer Science Department
Universidad Carlos III de Madrid
Avda de la Universidad, 30
28911-Leganés, Madrid
Spain
sjimenez-AT-uc3m.es
Erez Karpas
Faculty of Industrial Engineering and Management
Technion - Israel Institute of Technology
Technion City, Haifa 32000
Israel
karpase-AT-technion.ac.il
Subbarao Kambhampati
Dept. of Computer Science & Engineering
Fulton School of Engineering
Arizona State University, Tempe Arizona 85287-5406
USA
rao-AT-asu.edu
Program Committee

Daniel Borrajo, Universidad Carlos III de Madrid
Alan Fern, Oregon State University
Alfonso Gerevini, Università degli Studi di Brescia
Bob Givan, Purdue University
Hakim Newton, NICTA
Adele Howe, Colorado State University
Roni Khardon, Tufts University
Shaul Markovitch, Technion
Lee McCluskey, University of Huddersfield
Ioannis Refanidis, University of Macedonia
Scott Sanner, NICTA and ANU
Prasad Tadepalli, Oregon State University
Jia-Hong Wu
Sungwook Yoon, PARC
Shlomo Zilberstein, University of Massachusetts, Amherst