Information Technology Reference
In-Depth Information
strings. When the number of “blank” bits is zero, the corresponding candidate detec-
tor in h-NSA has the same probability to survive as that in t-NSA. The survivable
L
S
N
, where
S
N
is the number of the
left self strings to be matched when the number of “blank” bits is zero. Obvi-
ously,
probability can be represented by
P
=
(
−
P
)
S
m
L
S
, so the survivable probability
P
of a random detector in h-NSA is
higher than in t-NSA. Especially, as for h-NSA, an initial candidate detector has
N
<
N
S
r
−
1
2
−
r
initial random candidate detectors in t-NSA.
And in h-NSA the survivable probability
1
“blanks”. Therefore, it represents
2
−
r
1
P
of an initial random detector is
times higher than it in t-NSA.
Therefore, according to equation (5), the time cost of h-NSA is less than that of t-
NSA.
In addition, the space complexity of h-NSA is equal to that of t-NSA, as shown in
section 2. It is noted that to store “blank” bits, every bit of a detector in h-NSA needs
two bits. For example, “00” means” “0”, “01” means “1”, and “10” (or “11”) can be
used to denote “*”.
6 Conclusions and Future Works
A heuristic detector generation algorithm for negative selection algorithm with Ham-
ming distance partial matching rule is proposed in this paper. This is a good supple-
ment for negative selection algorithm since previous efficient detector generation
algorithms are most for the r-continuous-bits partial matching rule. There are also
some future works to be done, such as a heuristic detector generation algorithm on
higher alphabets while not a binary space. Another, this heuristic detector generation
algorithm will also be applied to practical applications. Generally, the practical data
set has somewhat special distributions, while not the random distribution that is tested
in this paper. Since the candidate detector in the heuristic detector generation algo-
rithm is generated according to the self individuals, the performance of h-NSA is
expected to be more competitive in practical applications.
Acknowledgements.
This work is partly supported by the National Natural Science
Foundation of China (NO. 60404004) and Nature Science Major Foundation from
Anhui Education Bureau (NO. 2004kj360zd).
References
1.
Emma Hart, Jonathan Timmis. Application Areas of AIS: the Past, the Present and the Fu-
ture. Proc. the 4th International Conference on Artificial Immune Systems (ICARIS), Lec-
ture Notes in Computer Science, Vol. 3627, Springer-Verlag (2005) 483-497
2.
D. Dasgupta, Z. Ji, et al. Artificial Immune System(AIS) Research in the Last Five Years.
Proc. the IEEE Congress on Evolutionary Computation (CEC), Canberra, Australia (2003)
123-130
3.
L. N. de Castro, J. Timmis. Artificial Immune Systems: a New Computational Intelligence
Approach. Springer-Verlag, London (2002)
