395

Chapter 9

Advanced Subjects

ological/medical systems are highly complex. It is still a

allenge to improve the accuracy, robustness and

erpretation capability of approaches or models for

ological pattern discovery. The previous chapters in this

ok have only covered a very small proportion of biological

ttern discovery subjects. Even for these subjects, there have

en ongoing advanced studies. This last chapter is dedicated

a brief introduction of several promising new trends in this

a.

al networks and deep learning

on to the accuracy of a constructed machine learning model, a

mon question in most biological/medical experiments is how the

non change can be well explained by the genetic components. For

the explanation must be given when a disease has been diagnosed

nswer must be sought when one set of genes is very active while

e found silent in an experiment with an applied stress.

e using machine learning algorithms to model biological/medical

pattern discovery, the conventional statistical models, which are

linear, are used, and they normally have an excellent

ation capability. For instance, a linear regression model can show

xplanatory variables have the most significant impact on the

non change of a predictor variable. Moreover, a linear

ant analysis model can explain how an object is labelled when

model has been constructed. In addition to the interpretation