CALL FOR PAPERS
IEEE SIGNAL PROCESSING MAGAZINE
Special Issue on Advances in Kernel-based Learning for Signal Processing
The importance of learning and adaptation in statistical Signal Processing creates a
symbiotic relation with Machine Learning. However, the two disciplines possess
different momentum and emphasis, which makes it attractive to periodically review
trends and new developments in their overlapping spheres of influence. Looking at
the recent trends in Machine Learning, we see increasing interest in kernel methods,
Bayesian reasoning, causality, information theoretic learning, reinforcement learning,
non numeric data processing, just to name a few. While some of the machine learning
community trends are clearly visible in Signal Processing, such as the increase
popularity of the Bayesian methods and graphical models, others such as the kernel
approaches are still less prominent. Kernel methods have a number of very attractive
merits for Signal Processing. More specifically:
• Linear operators in RKHS naturally yield nonlinear filters in the input space,
so this opens up many possibilities for optimum nonlinear system design.
• Kernels simplify the computation and bear the promise of on-line nonlinear
optimal filter implementations.
• Recent advances on embedding probability distributions into RKHS bring the
promise of nonparametric statistical inference with functional methods.
• Complementing the previous point, kernel methods may yield a practical
alternative to perform functional data analysis.
• A link between Information Theoretic Learning and RKHS theory was
established using Renyi’s entropy, which suggests other connections and
potential impact both on Information Theory and Signal Processing.
• Since positive definite functions can be defined in abstract spaces, RKHS
yields new opportunities to expand signal processing algorithms beyond
numerical data.
Scope of Topics of the Special Issue include:
• Nonlinear Adaptive Filtering using learning methods (e.g., kernels,
GPs, neural networks, etc.)
• On-line Learning with kernels
• Hypothesis testing with kernels
• Bayesian filtering in kernel spaces
• Information theory in RKHS
• Sampling theory using RKHS
• Analysis of non-numerical data in kernel spaces
• Issues in kernel design
• Optimization in kernel spaces
• Information fusion with kernels for example in multi-modal data
• Applications (e.g, Biology, Social Media, Engineering)

Tentative Schedule:
• White paper due: July 15, 2012
• Invitation notification: August 7, 2012
• Manuscript due: October 15, 2012
• Acceptance notification: November 31, 2012
• Revised manuscript due: December 30, 2012
• Final Acceptance notification: January 31, 2013
• Final manuscript due: February 20, 2013
• Publication data: July 2013
Submission procedure:
White papers, limited to 2 single-space double-column pages, should summarize the
motivation, the significance of the topic, a brief summary, an outline of the content
and key references. Prospective authors should use the web submission system at:
http://mc.manuscriptcentral.com/spmag-ieee.
Guest Editors:
Klaus-Robert Müller, TU Berlin (klaus-robert.mueller@tu-berlin.de)
Tulay Adali, UMBC (adali@umbc.edu)
Kenji Fukumizu, ISM, (fukumizu@ism.ac.jp)
Jose C. Principe, UFL (principe@cnel.ufl.edu)
Sergios Theodoridis U Athens (stheodor@di.uoa.gr)