[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[PVS] CfP: IEEE TNNLS Special Issue on Learning in Non-(geo)metric Spaces




CALL FOR PAPERS
 
IEEE TRANSACTIONS ON NEURAL NETWORKS AND LEARNING SYSTEMS
 
Special Issue on
 Learning in Non-(geo)metric Spaces

 
Traditional machine learning and pattern recognition techniques are
 intimately linked to the notion of “feature space.” Adopting this
 view, each object is described in terms of a vector of numerical
 attributes and is therefore mapped to a point in a Euclidean
 (geometric) vector space so that the distances between the points
 reflect the observed (dis)similarities between the respective objects.
 This kind of representation is attractive because geometric spaces
 offer powerful analytical as well as computational tools that are
 simply not available in other representations. However, the geometric
 approach suffers from a major intrinsic limitation which concerns the
 representational power of vectorial, feature-based descriptions. In
 fact, there are numerous application domains where either it is not
 possible to find satisfactory features or they are inefficient for
 learning purposes. By departing from vector-space representations one
 is confronted with the challenging problem of dealing with
 (dis)similarities that do not necessarily possess the Euclidean
 behavior or not even obey the requirements of a metric. The lack of
 the Euclidean and/or metric properties undermines the very foundations
 of traditional machine learning theories and algorithms, and poses
 totally new theoretical/computational questions and challenges that
 the research community is currently trying to address. The goal of the
 special issue is to consolidate research efforts in this area by
 soliciting and publishing high-quality papers which, together, will
 present a clear picture of the state of the art.
 

SCOPE OF THE SPECIAL ISSUE
 We will encourage submissions of papers addressing theoretical,
 algorithmic, and practical issues related to the two fundamental
 questions that arise when abandoning the realm of vectorial,
 feature-based representations, namely:
 
- how can one obtain suitable similarity information from data
 representations that are more powerful than, or simply different from,
 the vectorial?
 - how can one use similarity information in order to perform learning
 and classification tasks?
 
Accordingly, topics of interest include (but are not limited to):
 
- Embedding and embeddability
 - Graph spectra and spectral geometry
 - Indefinite and structural kernels
 - Game-theoretic models of pattern recognition and learning
 - Characterization of non-(geo)metric behavior
 - Foundational issues
 - Measures of (geo)metric violations
 - Learning and combining similarities
 - Multiple-instance learning
 - Applications
 
We aim at covering a wide range of problems and perspectives, from
 supervised to unsupervised learning, from generative to discriminative
 models, and from theoretical issues to real-world applications.
 

IMPORTANT DATES
 October 1, 2013 – Deadline for manuscript submission
 April 1, 2014 – Notification to authors
 July 1, 2014 – Deadline for submission of revised manuscripts
 October 1, 2014 – Final decision
 

GUEST EDITORS

 Marcello Pelillo, Ca Foscari University,Venice, Italy (pelillo@xxxxxxxxxxxx)

 Edwin Hancock, University of York, UK (edwin.hancock@xxxxxxxxxx)

 Xuelong Li, Chinese Academy of Sciences, China (xuelong_li@xxxxxxxx)

 Vittorio Murino, Italian Institute of Technology, Italy
 (vittorio.murino@xxxxxxxx)
 

SUBMISSION INSTRUCTIONS
 1. Read the information for authors at: http://cis.ieee.org/publications.html
 2. Submit the manuscript by October 1, 2013 at the IEEE-TNNLS webpage
 (http://mc.manuscriptcentral.com/tnnls) and follow the submission
 procedure. Please, clearly indicate on the first page of the
 manuscript and in the author's cover letter that the manuscript has
 been submitted to the Special Issue on Learning in non-(geo)metric
 spaces. Send also an e-mail to the guest editors to notify them of
 your submission.

Attachment: Special_Issue_LNGS_TNNLS.pdf
Description: Adobe PDF document