We found a match
Your institution may have rights to this item. Sign in to continue.
- Title
Improving MLP classification accuracy for breast cancer detection through evolutionary computation, partially connectivity and feature selection.
- Authors
Belciug, Smaranda; Gorunescu, Florin; Serbanescu, Mircea-Sebastian
- Abstract
Breast cancer is among the deadliest cancers that kills the most people worldwide. A computer-based model to discriminate between benign and malign cases would be helpful in cancer research. We explore in this paper the feasibility of using a feature selection algorithm for removing the redundant information, and a partially connected MLP using a GA routine to find optimum weights, targeting the breast cancer detection. Experimental results obtained on two publicly-available datasets show that using a maximum relevance and minimum redundancy-based strategy for feature selection, and an evolutionary-trained partially connected MLP for patients' classification, can achieve reasonably high performance. The outcomes demonstrate that this procedure is effective, and that it is feasible to apply computational classification techniques in automatic breast cancer detection.
- Subjects
MULTILAYER perceptrons; BREAST cancer diagnosis; EVOLUTIONARY computation; CANCER research; FEATURE selection; GENETIC algorithms
- Publication
Egyptian Computer Science Journal, 2013, Vol 37, Issue 5, p1
- ISSN
1110-2586
- Publication type
Article