SPECIAL ISSUE PROPOSAL FOR IEEE TRAN. ON NEURAL NETWORKS - LEARNING IN NONSTATIONARY & EVOLVING ENVIRONMENTS

[Robi Polikar]

Dear Colleagues

Cesare Alippi and I are planning on submitting a special issue
proposal titled “Learning in Nonstationary and Evolving Environments”
to IEEE Transactions on Neural Networks (soon to be renamed IEEE
Transactions on Neural Networks and Learning Systems).

The pre-proposal is copied below to this e-mail for your review, where
you will find the justification for the proposed issue, the topics
relevant to this issue, as well as a proposed timeline. Of course, all
papers will go through a standard IEEE TNN review, and I suspect that
each of the contributors will also be requested to review other
contributions. I was wondering whether you would be interested in
contributing to this special issue if it is approved. If you are
interested and are able to contribute, please let me know soon (I will
send this to Editor in Chief next week) and I will add your name to
the list of invited authors.


Many thanks and warmest regards,

Robi.

SPECIAL ISSUE PROPOSAL FOR
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
LEARNING IN NONSTATIONARY & EVOLVING ENVIRONMENTS
Guest Editors: Cesare Alippi, Robi Polikar

One of the fundamental goals in neural networks and learning systems
is to mimic brain-like intelli-gence, a remarkable property requiring,
in turn, the ability to incrementally learn from dynamic,
nonstationary (or time-variant) or evolving environments. Here,
adaptation is a primary task that becomes particularly challenging in
the presence of noisy, missing, incomplete, faulty or unbalanced
information.
Using a computational model to learn under various environments has
been a well-researched field that produced relevant results;
unfortunately, the majority of these efforts rely on three fundamental
assumptions: i) there is a sufficient and representative data set to
configure and assess the model performance; ii) data are drawn from a
fixed – albeit unknown – distribution; and iii) samples are mostly
supposed to be independent. Alas, all these assumptions often do not
hold in many real-world applications, such as in analysis of climate
or financial data, network intrusion, spam and fraud detection,
electricity demand and pricing prediction, and industrial quality
inspection among many others. Recent efforts towards incremental and
online learning have allowed us to relax the “sufficiency” requirement
by continuously updating a model to learn from small batches or online
data. Yet, even in incremental learning algorithms, the data that
become available are still assumed to be drawn from a fixed
distribution. More recently, approaches commonly called concept drift
algorithms – and to some extend domain adaptation algorithms – in
collaboration with change detection tests, have attempted to remove
this assumption, by accommodating a stream or batches of data whose
underlying distribution change over time. However, early efforts have
made other assumptions, such as restricting the type of changes
affecting the distribution, and/or they are primarily heuristic in
nature with several free parameters that need to be fine-tuned.
Furthermore, these efforts are yet to be validated on large scale real-
world applications.
Against this background, the need for a general framework for learning
from – and adapting to – a changing environment can be hardly
overstated. Combined with a growing number of real-world appli-cations
that can immediately benefit from such algorithms, such as learning
from financial, climate, epi-demiological data, it is clear that there
is much work to be done for developing such a general frame-work.
A special issue that discusses state-of-the-art approaches and latest
results on detecting and adapting to changes in underlying data
distributions is very timely.
The topics relevant to this special issue include, but are not limited
to
• Faults, changes or anomaly detection in data streams
• Incremental, lifelong and cumulative learning from nonstationary
data
• Domain adaptation
• Data mining from streams of data
• Learning in non-stationary, drifting or dynamic environments
• Architectures, techniques, algorithms – supervised, semi-supervised
or unsupervised – for learning in such environments
• Applications that require learning in dynamic and nonstationary
environments
• Adaptive learning in a missing, faulty or unbalanced data context
• Development of test-sets / benchmarks for evaluating algorithms
learning in such environ-ments
• Issues relevant to above mentioned or related fields

Several authors, including

• Robi Polikar – Rowan University, New Jersey, USA
• Cesare Alippi – Polytechnico Milano, Milan, Italy
• Gavin Brown – University of Manchester, England, UK
• Yaochu Jin –Univ. of Surrey, England, UK
• Ludmilla Kuncheva – Univ. of Bangor, Wales, UK
• Haibo He – Univ. of Rhose Island, RI, USA
• Shengxiang Yang – Brunel Univ., England, UK
• Will add new names as we receive confirmation


have already expressed interest in participating in this effort by
submitting manuscripts that describe their most recent research to
this special issue. We will also widely publicize this special issue
at various venues including all conferences the proposing guest
editors and their students / associates / colleagues attend; the IEEE
CIS newsletter, the CIS Magazine; and relevant mailing lists.

Tentative Timeline

15 October 2011 – Approval of the Special Issue by the Editor-in-Chief
– start publicizing and disseminating Call-for-Papers
15 April 2012 – Initial deadline for manuscript submission
15 May 2012 – Extended deadline for manuscript submission
15 July 2012 – Notification of authors
1 September 2012– Deadline for submission of revised manuscripts
15 September 2012 – Final decisions
January/February 2013 – Special Issue Publication in IEEE TNNLS








About the Guest Editors:

Cesare Alippi received the Dr.Ing. degree in electronic engineering
and the Ph.D. degree in computer engineering from Politecnico di
Milano, Milan, Italy. He has been a visiting researcher at the
University College London, London, U.K., the Massachusetts Institute
of Technology, Cambridge, USA and the École Supérieure de Physique et
de Chimie Industrielles, France. Alippi is a Fellow of the IEEE,
Associate Editor of the IEEE-Transactions on Neural Networks, Past
Associate Editor of the IEEE-Transactions on Instrumentation and
Measurements, Past Chair of the IEEE Neural Networks technical
Committee of the IEEE Computational Intelligence Society. In 2004 he
received the IEEE Instrumentation and Measurement Society Young
Engineer Award. Current research activity addresses adaptation and
learning in non-stationary environments, Neural Networks and Active
and Passive Intelligent Wireless Sensor Networks. Research is also
carried out on the industrial front with several well known companies.
Alippi is conference chair of IEEE IJCNN12 and has been program co-
chair of IEEE IJCNN11.

Robi Polikar received his Ph.D. from Iowa State University in Ames, IA
in 2000 in Electrical Engineering and Biomedical Engineering. He is a
Professor of Electrical and Computer Engineering at Rowan University,
Glassboro,NJ, where he chairs the department and also directs the
Signal Processing and Pattern Recognition Laboratory. His recent and
current works are funded primarily through National Science
Foundation’s CAREER and Energy, Power and Adaptive Systems Programs.
His primary research interests encompass various related areas of
computational intelligence, neural networks and learning systems,
including ensemble based learning, incremental and nonstationary
learning, data and decision fusion, and their real-world applications,
in which he has over 120 peer-reviewed publications. He is also
interested in developing educational paradigms that allow
undergraduate and entry level graduate students to participate in
rigorous computational intelligence research. Dr. Polikar is an
Associate Editor of IEEE Transactions on Neural Networks and Learning
Systems.






============================================
Robi Polikar, Ph.D.
Professor & Chair, Electrical and Computer Engineering
Rowan University, Glassboro, NJ 08028
Tel: (856) 256 5372; E-Mail: pol...@rowan.edu
On the web: http://users.rowan.edu/~polikar
=============================================