vii

Preface

e, the majority of biological data pattern discovery tasks have

om traditional univariate approaches, by which the activity of a

olecule is analysed using a statistical test, to more sophisticated

es. Along with the fast development and modernisation of

ology such as the microarray technology, the next-generation

ng and mass-spectroscopy technology, biological pattern

y has gradually become multivariate, nonlinear, multi-species and

ics as well as network-based. It is the so-called big-data era,

ot only has the promising prospect but also brings with it

es such as the computing facility problem, the model complexity

the generalisation capability problem, and importantly the

ation and extrapolation capability problem.

rchers and scientists have therefore made huge efforts to

e such challenges. Among many successful developments,

learning has always played and still plays a very important as

nique role in biological pattern discovery. Almost all machine

areas have greatly engaged in biological pattern discovery from

ised to supervised machine learning, from deterministic

g to stochastic modelling, from classical algorithms to more and

ing-edge algorithms. The variety of biological data, such as from

oarray gene expression data to the next-generation sequencing

m image data to spectroscopy data, have all seen the involvement

s advanced machine learning algorithms and approaches. The

cess of biological pattern discovery could not have been realised

he support of machine learning.