Call for Participation: Morpho Challenge 2010 – Semi-supervised and Unsupervised Analysis

http://www.cis.hut.fi/morphochallenge2010/

Morpho Challenge 2010 – Semi-supervised and Unsupervised Analysis

Part of the EU Network of Excellence PASCAL2 Challenge Program.
Participation is open to all.

The objective of the Challenge is to design a statistical machine
learning algorithm that discovers which morphemes (smallest individually
meaningful units of language) words consist of. Ideally, these are basic
vocabulary units suitable for different tasks, such as text
understanding, machine translation, information retrieval, and
statistical language modeling.

The scientific goals are:
* To learn of the phenomena underlying word construction in natural
languages
* To discover approaches suitable for a wide range of languages
* To advance machine learning methodology

Morpho Challenge 2010 is a follow-up to our previous Morpho Challenge
2005, 2007, 2008 and 2009. The task in 2010 is similar to 2009, where
the aim was to find the morpheme analysis of the word forms in the data.
As a new task we will provide a possibility for semi-supervised learning
using the available linguistic gold standard morpheme analysis.

Participation in the previous challenges is by no means a prerequisite
for participation in Morpho Challenge 2010. Everyone is welcome and we
hope to attract many participating teams. The results will be presented
in a workshop organized at our university in 2-3 September 2010. Please
read the rules and see the schedule at the home page.

If you now decided to participate in Morpho Challenge, please contact
the organizers and ask to be added in our mailing list. We will use this
mailing list to provide news about the tasks, data and evaluations.

We are looking forward to an interesting challenge!

Mikko Kurimo, Krista Lagus, Sami Virpioja and Ville Turunen
Adaptive Informatics Research Centre, Aalto University (previously
known as Helsinki University of Technology)
The organizers

http://www.cis.hut.fi/morphochallenge2010/

call for Ph.D.: ‘Identification of Medical Endocrine Systems’

The research group Syscon in the department of Information Technology of Uppsala University (Sweden) (http://www.it.uu.se/research/syscon) invites applications for a Ph.D. studentship (4-5 years). The succesful applicant will work in the context of the ERC project on engineering for endocrine systems, entitled “Systems and Signals Tools for Estimation and Analysis of Mathematical Models in Endocrinology and Neurology”. Especially motivated students with a strong background in engineering, computer science, applied mathematics or alike are encouraged to apply. This research will be conducted under supervision of Kristiaan Pelckmans, Alexander Medvedev and Petre Stoica.

In this Ph.D. project we will explore different approaches to unravell complex interaction networks as observed in hormonal, dynamical systems of the human body. The study of such endocinous systems is important in order to understand cause and consequence of different diseases, notably Parkinsons’ and type II Diabetes. The key point is that such systems use complex interaction strategies, typically taking the form of pulsatile signals of concentration of hormones in the blood stream. As classical modeling tools are often not satifactoy for such signals, ample room for improvement and further exploration of either theoretical or applied research remains in this area.

The applicant will work in the framework of a larger cooperation with different medical teams, and work towards a practical solution based on a sound theoretical foundation. In order to obtain this, we will combine the best of such exciting research areas as system identification, machine learning, signal processing, theoretical computer science, flavoured with a healthy engineering perspective.

The closing date for applications is at the end of februari. The official anoucement will follow in due time. If you are interested or have any question, contact Kristiaan Pelckmans (kp(at)it.uu.se).

PhD student / postdoc positions in PLT+AD

Seeking PhD students and postdocs interested in an elegant combination
of functional programming and big-iron style numeric computing.

Functional Programming and Automatic Differentiation
PhD Studentships
Postdoctoral Positions

We are adding exact first-class derivative calculation operators
(Automatic Differentiation or AD) to the lambda calculus, and
embodying the combination into a production-quality optimising
compiler. Our research prototype compiler generates object code
competitive with the fastest current systems, which are based on
FORTRAN. We are seeking PhD students and postdocs with interest and
experience in relevant areas: programming language theory, numeric
computing/numeric linear algebra, or differential geometry; and a
burning drive to help lift big iron numeric computing out of the 1960s
and into a newer higher order. Specific sub-projects include:
compiler and numeric programming environment construction; writing,
simplifying, and generalising numeric algorithms through the use of AD
operators; and associated type/lambda calculus/PLT/real computation
issues.

The project headquarters will be in the Hamilton Institute, NUI
Maynooth, Ireland, http://www.hamilton.ie/.

Applications to:
“Barak A. Pearlmutter”

French Spring School in Machine Learning

We invite applications to attend the 1st French Spring School in Machine
Learning (EPAT 2010) which will be held at Cap Hornu, in the beautiful Baie
de Somme, from may 2 to may 7 2010. Detailed information is available at

http://www.hds.utc.fr/epat2010/

Targeted Audience:

The school is meant for PhD students, young researchers in machine
learning and researchers from other research areas, such as signal
processing or bioinformatics.
Limited grants are available for PASCAL researchers.
Note that the lectures will be given in French.

Application deadline: April 2, 2010

Aims:

Our objective is to gradually present a few important themes, from basics to
recent advances. The selected themes are:

– Machine Learning: a Historical Perspective
– Optimization Techniques for Non-differentiable Problems in ML
– Kernel Methods
– Data Analysis
– Bioinformatics

Confirmed Lecturers:

Stéphane Chrétien, Université de Franche Comté
Antoine Cornuéjols, AgroParisTech
Marco Cuturi, Princeton University
Mohamed Nadif , Université Paris Descartes
Stéphane Robin, Institut National de la Recherche Agronomique (INRA)
Jean-Philippe Vert, Mines ParisTech & Institut Curie, INSERM

Program Committee:

Stéphane Canu, LITIS, INSA Rouen
Antoine Cornuéjols, LRI, AgroParisTech
François Denis, LIF, Université de Provence
Rémi Gilleron, INRIA, Lille
Yves Grandvalet, Heudiasyc, Université de Technologie de Compiègne
Rémi Munos, Mostrare INRIA, Université de Lille
Liva Ralaivola, LIF, Université de Provence
Marc Sebban, Laboratoire Hubert Curien, Université Jean Monnet
Jean-Philippe Vert, Mines ParisTech & Institut Curie, INSERM

Organisers:

Heudiasyc, CNRS and University of Compiègne joint research unit, France
CNRS, Nord-Pas de Calais & Picardie delegation.

Supported by PASCAL, Heudiasyc, CNRS, UT Compiègne, Picardie, INRIA, GdR BiM, SFdS

Special Issue Call for Papers: Model and Optimization for Machine Learning -*DEADLINE EXTENDED*

Details of the call can be found at the following link:

http://www.pascal-network.org/cfp/specialissuenov09.pdf

The paper submission deadline has been extended to 28 February 2010

Second call for papers: MLSP 2010, the Twentieth IEEE International Workshop on Machine Learning for Signal Processing

Second Call for Papers

for the

Twentieth IEEE International Workshop on
Machine Learning for Signal Processing
(MLSP 2010)

August 29 – September 1, 2010, Kittila, Finland

Website: http://mlsp2010.conwiz.dk

IMPORTANT DATES:

Submission of full papers: April 1, 2010
Notification of acceptance: May 28, 2010
Camera-ready paper
and author registration: June 18, 2010
Advance registration before: June 23, 2010

The 2010 IEEE International Workshop on MACHINE LEARNING FOR SIGNAL
PROCESSING (MLSP 2010) will be held in Kittila, Finland, in
August-September 2010. MLSP 2010 is the twentieth workshop in the
series of workshops sponsored by IEEE Signal Processing Society. It
will present the most recent and exciting contributions in machine
learning for signal processing through keynote talks as well as
special and regular single-track sessions.

INVITED SPEAKERS:

– Prof. Zoubin Ghahramani, University of Cambridge
– Prof. Tom Mitchell, Carnegie Mellon University
– Dr. Henry Tirri, Head of Nokia Research Center

ORGANIZATION:

General chair: Erkki Oja
Program chairs: Samuel Kaski, David Miller
Special session chairs: Samy Bengio, Mikko Kurimo
Publicity chairs: Marc Van Hulle, Jaakko Peltonen
Web and publication chairs: Antti Honkela, Jan Larsen
Data competition chairs: Vince Calhoun, Kenneth Hild, Mikko Kurimo
Local arrangements: Tapani Raiko (chair), Francesco Corona,
Ali Faisal, Mari-Sanna Paukkeri

VENUE:

MLSP 2010 will be held in the Levi Summit conference and exhibition
centre in Kittila, Finland. Levi is one of the largest resorts in
Finnish Lapland, north of the Arctic Circle. In the summer, Levi
offers many sports activities as well as lots of wild northern nature.
The conference centre is located high on the hillside of the Levi
fell, accessible by gondola from the main village.

CONFERENCE TOPICS:

Machine learning in signal processing is concerned with tasks such as
detection, estimation, prediction, classification, and optimization,
with a wide range of applications. The following is a non-exhaustive
list of topics for MLSP 2010:

– Bayesian learning and signal processing
– Cognitive information processing
– Graphical and kernel methods
– Information-theoretic learning
– Learning theory and algorithms, including bounds on performance
– Supervised learning, including signal detection, pattern
recognition and classification
– Unsupervised learning, reinforcement learning
– Source separation and component analysis
– Data fusion and integration
– Feature extraction, information visualization
– Sparse and structured representations
– Neural network learning
– Time-series analysis
– Adaptive filtering
– Data mining, information retrieval
– Sequential learning and sequential decision methods
– Hardware implementation of machine learning in signal processing
– Applications of machine learning: Bioinformatics, Biomedical and
neural signal processing, Neuroinformatics, Speech and audio
processing, Image and video processing, Computer vision,
Sensor networks, Robot control, Communications, Cognitive radio,
Multimodal interfaces and context modeling, Intelligent multimedia
and web processing

SPECIAL SESSION:

A special session “Towards multimodal proactive interfaces using
large-scale machine learning” is being organized. For more
information see http://mlsp2010.conwiz.dk .

DATA COMPETITION:

In conjunction with the workshop, a data and signal analysis
competition “Mind Reading” is being organized. Winners will present
their works and receive their award during the Workshop. For more
information see http://mlsp2010.conwiz.dk .

PAPER SUBMISSION PROCEDURE:

Authors are invited to submit a double column paper of up to six pages
using the electronic submission procedure described at
http://mlsp2010.conwiz.dk .

Accepted papers will be published by IEEE Press; electronic
proceedings will be distributed at the workshop and included in
IEEE Xplore.

JOURNAL SPECIAL ISSUE:

Authors of selected papers will be invited to submit extended versions
to a special issue of an international journal.

SPONSORS: MLSP 2010 is supported by IEEE, by the IEEE Signal
Processing Society, by the PASCAL2 Network of Excellence, and by
the Federation of Finnish Learned Societies. The data competition is sponsored by Nokia and the PASCAL2 Challenge Program.

========= See http://mlsp2010.conwiz.dk for more details! =========

Workshop on Active Learning and Experimental Design (collocated with AISTATS)

Workshop on Active Learning and Experimental Design
—————————————————————————-
Collocated with AISTATS 2010
May 16, 2010
Chia Laguna resort
http://clopinet.com/isabelle/Projects/AISTATS2010/

Abstract submission deadline: March 8, 2010

Guest speakers:
– Donald Rubin, Department of Statistics, Harward University (co-inventor of the EM algorithm, expert in experimental design and causal inference)
– Burr Settles, Machine Learning Department, Carnegie Mellon University (author of an authoritative tutorial on active learning)
– David Jensen, Computer Science Department, Universtity of Massachusetts (expert in quasi-experimental design)

The results of the Active Learning challenge http://clopinet.com/al will be discussed at the workshop and plans will be made for an upcoming challenge on experimental design.

ICDL (deadline EXTENDED): 9th International Conference on Development and Learning

Submission deadline EXTENDED to March 6

9th International Conference on Development and Learning (ICDL)
University of Michigan, Ann Arbor, MI, USA
http://www.icdl-2010.org
August 18-21, 2010

ICDL is the premiere venue for interdisciplinary research that blends the boundaries between robotics, artificial intelligence, machine learning, developmental psychology, neuroscience, and philosophy. The scope of development and learning covered by this conference includes perceptual, cognitive, motor, behavioral, emotional and other related capabilities that are exhibited by humans, higher animals, artificial systems and robots.

While most other conferences focus on either mechanisms or organisms, ICDL focuses on both! The papers presented at the conference are split approximately 50-50 between the “natural intelligence side,” such as neuroscience and psychology, and the “artificial intelligence side,” such as machine intelligence and robotics. This diversity is mirrored in the composition of the organizing committee and the ICDL governing board. Please join us in 2010 when we celebrate our 10-th anniversary.

Topics of interest include, but are not limited to:

* General principles of development
* Cognitive and perceptual development
* Developmental learning: schedules and architectures
* New methodologies to study natural and artificial intelligence.
* Statistical learning in humans and machines
* Embodied cognition
* Play and exploration in animals, infants and robots
* Interactive learning
* Cultural learning
* Social and emotional development
* Theory of mind
* Language acquisition
* Skill acquisition
* Intrinsic motivation
* Dynamic systems
* Attention mechanisms and their role in development
* Philosophical issues of development and learning
* Differences between learning and development
* Interactions of learning and development with evolution
* Grounding of knowledge and representations
* Studies and models of developmental disorders, e.g., autism
* Using robots to study development and learning
* Human-Robot interaction
* Visual, auditory, and tactile systems and their development
* Motor systems and their development
* Biological and biologically inspired developmental architectures
* Neural plasticity during development.

ICDL 2010 will accept two types of submissions:

1) Full six-page paper submissions. Accepted papers will be included
in the conference proceedings and will be selected for either an oral
presentation or a featured poster presentation. Featured posters
will have a 1 minute “teaser” presentation as part of the main
conference session and will be showcased in the poster sessions.

2) Two-page poster abstract submissions. To encourage late-breaking
results or for work that is not sufficiently mature for a full paper,
ICDL will accept 2-page abstracts. These submissions will NOT be
included in the conference proceedings. Accepted abstracts will be
presented during the evening poster sessions.

Important dates:
** Mar 06, 2010 Full 6-page paper submissions due ** Changed from Feb 20
May 20, 2010 Notification of accept/reject for papers
May 27, 2010 2-page poster abstracts due
June 10, 2010 Notification of accept/reject for abstracts
June 20, 2010 Camera-Ready Copy due
July 20, 2010 Early Registration Deadline
Aug. 18-21, 2010 Conference

General Chairs:
* Benjamin Kuipers, University of Michigan
* Thomas Shultz, McGill University

Program Chairs:
* Alexander Stoytchev, Iowa State University
* Chen Yu, Indiana University, Bloomington

Publicity chairs:
* Ian Fasel, University of Arizona, USA (for North America)
* Jochen Triesch, Frankfurt Institute for Advanced Studies, Germany (for Europe)
* Jun Tani, RIKEN, Japan (for Asia).

Sponsored by:
* IEEE Computational Intelligence Society
* Cognitive Science Society

For more information please check the conference web site:
http://www.icdl-2010.org/

Sony and Oxford Brookes Research Fellow, computer vision for computer games

The Sony EyeToy was launched onto the video games market five years ago. Since then, it has found its way into over 8 million homes worldwide. Using a small video camera plugged into a PlayStation, it tracks the player’s motion, allowing them to play games by moving their bodies rather than using a controller.

Sony Computer Entertainment Europe and Oxford Brookes University have established a Knowledge Transfer Partnership (KTP), aimed at transferring recent results in the academic community into gaming experience for camera based games. We are seeking a Research Associate to lead this 27 month project.

The main responsibility of the post holder will be to research and develop machine learning and computer vision methods and algorithms including:

· Hand tracking

· Interactive computer games

· Image Segmentation

with the aim of developing new camera based games based on this research for the EyeToy.

You should have:

· a good first degree in maths, engineering, physics or computer science

· a PhD or equivalent research experience in computer vision and machine learning

· strong academic and commercial knowledge of computer vision

· ability to research and develop new computer vision algorithms

· good C++

· strong mathematics

· a passion for computer games

You will be self motivated, organised and good at communicating. You should be able to create novel computer vision algorithms and turn these into production quality systems. You should be able to take direction and also contribute to the overall strategy and goals of the group.

Informal inquiries to:

philiptorr(at)brookes.ac.uk,

Professor Phil Torr,
Department of Computing,
Oxford Brookes University,
Wheatley,
Oxford
OX33 1HX.
http://cms.brookes.ac.uk/staff/PhilipTorr/

The Synthetic Visual Reasoning Test Challenge

The first dead-line has been moved to August 31, 2010.

This competition is part of the PASCAL2 challenge program
http://pascallin2.ecs.soton.ac.uk/Challenges/

INTRODUCTION

We are pleased to announce a new challenge for machine learning and
computer vision: The Synthetic Visual Reasoning Test (SVRT). One
motivation is to expose some limitations of current methods for
pattern recognition, and thereby to argue for making a larger
investment in other paradigms and strategies, emphasizing the pivotal
role of relationships among parts, complex hidden states and a rich
dependency structure.

This test consists of a series of 23 hand-designed, image-based,
binary classification problems. The images are binary and with
resolution 128×128. For each problem we have implemented a generator
in C++, which allows one to produce as many i.i.d samples as desired.
A pdf document containing examples of images is available at

http://www.idiap.ch/~fleuret/svrt/svrt.pdf

The Bayes error rate of each problem is virtually zero, and nearly all
of them can be perfectly solved by humans after seeing fewer than ten
examples from each class. Nonetheless, some of them are probably as
difficult as various “real” problems featured in previous challenges
and widely known data-sets. In particular, solving these synthetic
visual tasks with high accuracy requires “reasoning” about
relationships among shapes and their poses.

Human experiments were conducted in the laboratory of Prof. Steven
Yantis, a cognitive psychologist at Johns Hopkins University; those
results will appear in a future publication. A number of people were
asked to solve the problems and the number of samples required to
master each concept was recorded.

SVRT challenge participants who follow the rules described below and
whose results are noteworthy for either their originality or sheer
performance will be invited to co-author a comprehensive, and
hopefully visible, article summarizing the performance of their
methods, including a discussion of the performance of humans (and
possibly monkeys) on the same tasks.

CHALLENGE

The generators for a randomly-selected subset of 13 problems are made
available to participants. Using these 13 problems as “case studies,”
the challenge is to develop or adapt a learning algorithm which inputs
a training set and outputs a classifier for labeling a binary image.

An important performance metric is the number of training examples
required to obtain any given accuracy. Algorithms should be designed
to be trained on sets of varying sizes.

Participants have until August 31, 2010, for development, and are
required to make public the results achieved on the 13 problems as
well as the source code required to reproduce these results and to
test the algorithm on other problems.

The source code and test error rates must be sent to the challenge
organizers Francois Fleuret (francois.fleuret(at)idiap.ch) and Donald
Geman (geman(at)jhu.edu) before midnight EST, August 31, 2010.

The test error rates must be provided in a single text file, with one
line per problem and number of training examples. At minimum, results
are to be provided for exactly 10, 100 and 1000 training examples per
class per problem. Participants may choose to also send their results
for higher powers of ten. On each line there should be the problem
number, followed by the number of training samples, followed by ten
test error rates estimated on ten different runs, with 10,000 test
samples per class. Numbers should be separated by commas.

On September 1, 2010, we will publish the ten remaining problems
(i.e., make the generators available). Participants will measure the
performance of their algorithms *with no additional change* on this
new set of problems and send the performance by mail to the challenge
organizers before midnight EST, September 31, 2010. At that point, we
may use the participants’ code to verify the reported performance.

* DOWNLOAD

The source code of the generators can be downloaded from

http://www.idiap.ch/~fleuret/svrt/

A pdf document containing ten samples of each class of each problem,
together with the error rate of a baseline classifier trained with
Boosting, is available at

http://www.idiap.ch/~fleuret/svrt/svrt.pdf

* CONTACT

François Fleuret, Idiap Research Institute
francois.fleuret(at)idiap.ch
http://www.idiap.ch/~fleuret/

Donald Geman, Johns Hopkins University
geman(at)jhu.edu
http://www.cis.jhu.edu/people/faculty/geman/