Database Reference
In-Depth Information
adaptability and so on.A NIS comprises a complex
system of cells, molecules and organs that jointly
represent an identification mechanism capable of
perceiving and combating exogenous infectious
microorganisms, which contain many antigens that
are substances that can trigger immune responses,
resulting in production of antibodies as part of the
body's defence against infection and disease to
neutralize related antigens.
Generally speaking, an artificial immune sys-
tem (AIS) is a specific computational algorithm
which takes its inspiration from the way how a NIS
learns to respond to those exogenous invaders. It
simulates the key features, such as adaptability,
pattern recognition, learning, and memory acquisi-
tion of the NIS in order to deal with the problems
(Dasgupta 1998) in computer security, anomaly
detection, fault diagnosis, pattern recognition
and a variety of other applications (Timmis et
al. 2003) in science and engineering. As one of
the main areas of the financial market, the stock
market has an important concept -noise, which
is defined as the fluctuations of price and volume
that can confuse interpretation of market direction.
Accordingly, those investors undertaking trades
which generate such confusions are termed as
noise traders. Some noise traders are described
as essential players of the stock market (Black
1985) whereas some insiders illegitimately take
advantage of exclusive information which is still
unavailable to the public to trade securities and
disclose some information through a public signal
consisting of a noisy transformation of his or her
own private information (Gregoire 2001). Some
market manipulators try to influence the price of
a security in order to create false or misleading
patterns of active trading to bring in more trad-
ers. Often, the newly brought-in traders further
cause significant or even disastrous deviation of
security prices from the underlying values of the
related assets resulting to the failure of correct
interpretation of the market direction.
Lee & Yang (2005) proposed an expansile
and adaptive abnormal-trading detection system
with the characteristics of good self-learning
and memory capacities. Their artificial abnormal
detection system has the following advantages.
Firstly, its adaptivity means that the system is able
to learn the trading patterns of different stocks;
and also that it is able to learn the different trad-
ing patterns of the same stock according to each
economic period, trading period, and the locality
of financial markets. Secondly, it is anticipated that
better proxies will be continuously found; the ex-
pansibility allows those newly discovered proxies
be added to the system. Thirdly, the system can be
used as a tool to compare proxies so as to search
for a more proper proxy for a specified stock in
a certain period at a certain place. Finally, those
proxies used in the system will be eliminated or
superseded before they are out of date. Thus with
the continuous introduction of the new proxies
into the system, the limited memory capacity of
the system can be effectively utilized.
AIAS was tested and evaluated on high-
frequent artificial and real stock market data. The
test on real stock market is to see whether AIAS
is able to detect one recorded insider trader case.
It is recorded that insider trader trades occurred
on 24 April, 2001, just before Qantas announced
at the end of April 2001 that it would take over
the operations of Impulse Airlines. As a result,
AIAS detected three suspicious transactions on
20th and two on 24th respectively. It matches
the insider trading case which happened on 24
April 2001.
AIAS needs to be researched on its time usage,
and its ability of detecting anomalies needs to be
improved and tested on more datasets.
other techniques
Some other techniques used for exception mining
in stock market surveillance are decision tree,
logistic regression, neural network, etc.
Decision tree is a predictive model that maps
from observations about an item to conclusions
about its target value (Utgoff 1989). In the tree
Search WWH ::




Custom Search