C.S. Leung, M. Lee, and J.H. Chan (Eds.): ICONIP 2009, Part II, LNCS 5864, pp. 611–619, 2009.
© Springer-Verlag Berlin Heidelberg 2009
String Pattern Recognition Using Evolving Spiking
Neural Networks and Quantum Inspired Particle Swarm
Optimization
Haza Nuzly Abdull Hamed
1
, Nikola Kasabov
1
, Zbynek Michlovský
2
,
and Siti Mariyam Shamsuddin
3
1
Knowledge Engineering and Discovery Research Institute (KEDRI),
Auckland University of Technology, New Zealand
{hnuzly,nkasabov}@aut.ac.nz
2
Faculty of Information Technology, Brno University of Technology, Brno, Czech Republic
imichlov@fit.vutbr.cz
3
Soft Computing Research Group, Universiti Teknologi Malaysia, Malaysia
mariyam@utm.my
Abstract. This paper proposes a novel method for string pattern recognition us-
ing an Evolving Spiking Neural Network (ESNN) with Quantum-inspired Parti-
cle Swarm Optimization (QiPSO). This study reveals an interesting concept of
QiPSO by representing information as binary structures. The mechanism opti-
mizes the ESNN parameters and relevant features using the wrapper approach
simultaneously. The N-gram kernel is used to map Reuters string datasets into
high dimensional feature matrix which acts as an input to the proposed method.
The results show promising string classification results as well as satisfactory
QiPSO performance in obtaining the best combination of ESNN parameters and
in identifying the most relevant features.
Keywords: String Kernels, Text Classification, Evolving Spiking Neural Net-
work, Particle Swarm, Quantum Computing.
1 Introduction
String pattern recognition is an approach for determining which group a string belongs
to, according to its contents. This task, despite being quite challenging, is very impor-
tant to certain areas such as internet security and virus detection. Strings can be texts,
musical symbols or others which are not necessarily in numerical formats. Since most
classifier algorithms can only accept numerical values, transformation from strings to
numerical values is required. String kernels are a well-known method to transform
string input values into high dimensional input vectors. There are several well-known
string kernels such as Bag of Words (BOW) and N-gram Kernels. Output from the
kernel process which is the kernel matrix is used as an input to the algorithm for classi-
fication, clustering or ranking tasks. This technique is quite simple yet quite effective
to transform input from string to the numerical value.
KEDRI-NICT Project Report - APPENDIX:E