Overview
The last decade has shown significant advances in machine learning (ML)-based robotics, fostered in particular by the DARPA Challenges aimed at autonomous vehicles since 2005. Similar advances in the domain of multi-agent systems (MAS) and swarm robotics (SR) are likely and eagerly expected.
One main barrier for ML and data mining (DM)-based advances in robotics and SR was the fragmentation of the research community due to the rapid development of the research fields involved. Therefore quite a few venues have been organized to address this fragmentation, gradually making the ML and DM communities more aware of the critical issues faced by the Robotics and Multi-Agent communities.
This workshop aims at fostering research in Machine Learning and Data Mining applied to Robots. Below is a non-exhaustive list of expected contributions, as papers on relevant topics are also welcome.
Program | |
10:25-10:30 | Opening (Einoshin Suzuki) |
4 presentations (Chair: Shin Ando) | |
10:30-11:00 | Characterizing Anomalous Behaviors and Revising Robotic Controllers (David Meunier, Michele Sebag, and Shin Ando) |
11:00-11:30 | Adaptive Windowing for Online Learning from Multiple Inter-Related Data Streams (Elena Ikonomovska, Kurt Driessens, Saso Dzeroski, and Joao Gama ) |
11:30-12:00 | Lifted-Rollout for Approximate Policy Iteration of Markov Decision Process (Wang-Zhou Dai, Yang Yu, and Zhi-Hua Zhou) |
12:00-12:30 | Implementing Camshift on a Mobile Robot for Person Tracking and Pursuit (Somar Boubou, Asuki Kouno, and Einoshin Suzuki) |
Open log data of physical robots (in preparation)
We provide log data of a physical robot to encourage submissions
from a wide range of researchers in ML/DM who have no access to physical
robots. The log data consist of thousands of images (3-6 minutes +
6-9 minutes) with sensor values (0-9 minutes, explanation)
taken by a mobile robot, accompanied with a movie (6-9 minutes).
You may tackle a data mining/machine learning problem on the data, e.g.,
clustering of sensor values from the log data using images.
You may also propose a data mining/machine learning algorithm which is
executed on-board of the robot, e.g., detection of unusual situations
from sensor readings and images by the robot.
Please acknowledge our JST-ANR project on Integrating Symbolic
Discovery with Numerical Machine Learning for Autonomous Swarm Control
in your publication using the data.
To uncover the hidden pictures and the movie, please submit an article to this workshop.
Key Dates | |
Deadline for papers | August 5, 2011 |
Notification of the results | September 20, 2011 |
Camera readies due | October 14, 2011 |
Workshop | December 11, 2011 (ICDM organizers changed to 11) |
Program Committee | ||
Chair | (Kyushu University, Japan) | |
Chair | (CNRS & Université Paris-Sud, France) | |
(Gunma University, Japan) | ||
(Universidad de Cantabria, Spain) | ||
(Ecole Polytechnique Fédérale de Lausanne, Switzerland) | ||
(University of Ljubljana, Slovenia) | ||
(Université Paris-Sud, France) | ||
(University of Porto, Portugal) | ||
(CWI, Netherlands) | ||
(University of Tokyo, Japan) | ||
(Fraunhofer IAIS & University of Bonn, Germany) | ||
(Max Planck Institute for Biological Cybernetics, Germany) | ||
(Osaka University, Japan) |
Paper format
According to the Workshop Chairs, we can have either full (8 pages)
or short (6 pages) papers and up to 2 pages per paper can be purchased
at 125 USD per page.
Please use IEEE 2-column
format.
We don't adopt the triple blind review system of ICDM 2011 so please do not
hide your identity in your submission.
Related links
LEMIR 2009,
JST-ANR project,
SYMBRION project,
PASCAL Network of Excellence