Abstract
Original language | English |
---|---|
Title of host publication | WCCI 2010: IEEE World Congress on Computational Intelligence |
Place of Publication | Piscataway, N.J., USA |
Publisher | IEEE |
Pages | 623-629 |
Number of pages | 7 |
ISBN (Print) | 9781424481262 |
DOIs | |
Publication status | Published - 2010 |
Event | 2010 IEEE World Congress on Computational Intelligence (FUZZ-IEEE 2010) - Barcelona, Barcelona, Spain Duration: 18 Jul 2010 → 23 Jul 2010 |
Conference
Conference | 2010 IEEE World Congress on Computational Intelligence (FUZZ-IEEE 2010) |
---|---|
Country | Spain |
City | Barcelona |
Period | 18/07/10 → 23/07/10 |
Fingerprint
Cite this
}
A practical study on shape space and its occupancy in negative selection. / Ma, Wanli; Tran, Dat; Sharma, Dharmendra.
WCCI 2010: IEEE World Congress on Computational Intelligence. Piscataway, N.J., USA : IEEE, 2010. p. 623-629.Research output: A Conference proceeding or a Chapter in Book › Conference contribution
TY - GEN
T1 - A practical study on shape space and its occupancy in negative selection
AU - Ma, Wanli
AU - Tran, Dat
AU - Sharma, Dharmendra
PY - 2010
Y1 - 2010
N2 - The success of a negative selection algorithm depends on its detectors. A shape space, conceptually, is where selves, detectors, and antigens reside. These detectors are expected to fully cover the whole shape space. The better the coverage; the better the detection rate. However, this assumption brings a major challenge to negative selection experiments - the scalability problem, where the experiments cannot process real life datasets in a timely manner. On the other hand, with any real life dataset, due to arbitrary antibody/antigen representing formats, the shape space actually cannot be fully occupied. The unoccupied locations sometimes constitute a significant, or even overwhelm, portion in a shape space. In this paper, we first briefly review the theoretic model of the shape space and then study the impact of the unoccupied locations, under the term shape space occupancy. Based on the study outcomes, we suggest the heuristics for generating detectors. We demonstrate shape space occupancy, detector generation by antigen feedback mechanism, and negative selection experiments on 4 different datasets, which cover the data presentation formats in both strings and real number valued vectors.
AB - The success of a negative selection algorithm depends on its detectors. A shape space, conceptually, is where selves, detectors, and antigens reside. These detectors are expected to fully cover the whole shape space. The better the coverage; the better the detection rate. However, this assumption brings a major challenge to negative selection experiments - the scalability problem, where the experiments cannot process real life datasets in a timely manner. On the other hand, with any real life dataset, due to arbitrary antibody/antigen representing formats, the shape space actually cannot be fully occupied. The unoccupied locations sometimes constitute a significant, or even overwhelm, portion in a shape space. In this paper, we first briefly review the theoretic model of the shape space and then study the impact of the unoccupied locations, under the term shape space occupancy. Based on the study outcomes, we suggest the heuristics for generating detectors. We demonstrate shape space occupancy, detector generation by antigen feedback mechanism, and negative selection experiments on 4 different datasets, which cover the data presentation formats in both strings and real number valued vectors.
U2 - 10.1109/CEC.2010.5586266
DO - 10.1109/CEC.2010.5586266
M3 - Conference contribution
SN - 9781424481262
SP - 623
EP - 629
BT - WCCI 2010: IEEE World Congress on Computational Intelligence
PB - IEEE
CY - Piscataway, N.J., USA
ER -