A Novel Approach for Classification of Malignant Neoplasm Using Non-Linear Dualist Optimization Algorithm
Prachi Vijayeeta1, M. N. Das2, B. S. P. Mishra3
1Prachi Vijayeeta, Department of Computer Engineering, Kalinga Institute of Industrial Technology, Deemed to Be University, Bhubaneswar (Odisha), India.
2M. N. Das, Department of Computer Engineering, Kalinga Institute of Industrial Technology, Deemed to Be University, Bhubaneswar (Odisha), India.
3B. S. P. Mishra, Department of Computer Engineering, Kalinga Institute of Industrial Technology, Deemed to Be University, Bhubaneswar (Odisha), India.
Manuscript received on 07 April 2019 | Revised Manuscript received on 20 April 2019 | Manuscript published on 30 April 2019 | PP: 1771-1778 | Volume-8 Issue-6, April 2019 | Retrieval Number: F3175048619/19©BEIESP
Open Access | Ethics and Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract: In the previous centuries, our deterministic view of a decisive disease like cancer led us to imagine that only clinical therapy seems to be the ultimate source for prediction of a disease. But the emergence of Machine learning techniques has contributed an astonishingly intellectually efficient mechanism for solving numerous complex biological problems with minimal complexity. Keeping an eye on the need of preliminary detection of initial stages of cancer development, we have reshaped our perception for prediction by implementing certain interdisciplinary fields like Game theory, numerical linear algebraic methods and statistics to reach a valid conclusion. In our work, a model is formulated to determine the existence of malignant neoplasm in the patients sample set using Gauss-Newton method. Our model aims at predicting the class label of an unknown new sample that enters into the system during the runtime. An apodictic approach for construction of an optimized RFE (Recursive Feature Elimination) feature selection model using Dualist Algorithm is incorporated. Furthermore, the optimized features are subjected to a non-linear classification using Decision Tree, K-Nearest Neighbor and Logistic Regression on Wisconsin Breast cancer dataset. In addition, we have applied Euclidean distance and Manhattan Distance to select the most contributing features in a sparsely represented dataset. The simulations carried out using Gauss-Newton method, Logistic Regression, K-nearest neighbor, Decision Tree and Random Forest. Dualist Optimization algorithm with Recursive feature elimination is combined with five different classifier models are experimented. Features with minimum Euclidean distance and Manhattan distance from the sample data are also chosen for training the model and is further optimized using the suggested algorithm.
Keyword: Classification Models, Game theory, Dual-RFE, k-fold cross validations.
Scope of the Article: Classification