Veratridine br Taking into consideration the advantages
● Taking into consideration the advantages of IG and GAW, then proposing a IGSAGAW hybrid feature selection approach, which can remove redundant and irrelevant feature from the feature space, thus improving the classification accuracy and reducing the computational cost.
● Taking into consideration the unequal misclassification costs and classification accuracy (ACC), we propose a novel intelligent classification model to distinguish benign breast tumors from malignant ones. The effectiveness of our proposed method are verified on WBC and WDBC data sets, the empirical results demonstrate that our proposed method can achieve good perfor-mances.
The rest of this paper is organized as follows. Section 2 presents related works on breast cancer diagnosis. Section 3 presents the research objective of our study. Section 4 introduces the backgrounds and preliminaries of our method. Section 5 proposes the framework of our proposed approach. Section 6 presents the experimental analysis of our proposed model. Section 7 presents the discussion of our proposed model. Finally, Section 8 presents the conclusions of our research.
2. Related works
In this section, we summarize the previous studies of the breast cancer diagnosis over 10 years. Existing studies primarily adopt artificial neural networks (ANNs), decision tree analysis (DTA), Naïve Bayes (NB), Support Veratridine Machines (SVM) and so on. As Table 1 present the previous studies of breast cancer intelligent diagnosis in recent 10 years.
Due to the neural network has the advantages of capturing the correlations between attributes, therefore it has been widely utilized for breast cancer diagnosis (Lundin et al., 1999; Ravdin & Clark, 1992; Yao, 1999). Liu, Wang, and Zhang (2009) designed a decision tree prediction model for breast cancer survivability and adopt under-sampling method to balance the training data, the results has shown that when the ratio is equal to 15%, the AUC of the model is 0.7484. On top of the decision tree algorithm, Quinlan (1996) introduced MDL-inspired penalty and designed an improved C4.5 decision tree algorithm for breast cancer prediction, and attained the prediction accuracy of 94.74%. However, the performance of single learning classification algorithm can't reflect the
Summary of typical previous research for breast cancer intelligent diagnosis.
Author Year Methods Results
Ahn et al. 2009 Novel CBR training-test partition
Zheng et al. 2014 K-means and SVM. training-test partition
SVM based ensemble learning
interactive factors of the breast cancer survival and recurrence rate (Wang, Zheng, Yoon, & Ko, 2018), Therefore, In order to overcome the drawbacks bring by single algorithm, numerous hybrid algorithms have been proposed. Akay (2009) presents F-score method for feature selection and SVM for breast cancer prediction. On top of that, another hybrid algorithm presented by Chen et al. (2011) which designed an hybrid classifier with rough set for feature selection and SVM for classification. Zheng, Yoon, and Lam
(2014) proposed K-means and SVM hybrid algorithm for breast cancer diagnosis, K-means method for breast tumor feature extraction and SVM for classification, In another study, Onan (2015) designed a hybrid intelligent classification model for breast cancer di-agnosis, which consist of fuzzy-rough approach for instance selection, consistency-based for feature selection and fuzzy-rough nearest neighbor algorithm for breast tumor classification. In addition, Sheikhpour, Sarram, and Sheikhpour (2016) proposed PSO and non-parametric kernel density estimation (KDE) based classifier to diagnose breast cancer. To summarize, their results shown that the proposed hybrid models have achieved high classification accuracy with fewer feature variables. In 2018, Wang et al designed an ensemble algorithm fusion SVM for breast cancer diagnosis which emphasis on model structures, and the results shown that the model achieves a higher accuracy compared to other ensemble models. To sum up, the main disadvantages of previous studies in breast cancer diagnosis is that they only pursuit of high classification accuracy, ignoring the unequal misclassification cost. Never-theless, to the best of our knowledge, in medical diagnosis, comparing the cost of misclassifying a cancerous patient as a non-cancerous to misclassifying a non-cancerous patient as cancerous, the consequences may vary greatly. Therefore, this study constructs a hybrid intelligent classification model which has a competitive performance compared to other existing methods. The main ad-vantage of our proposed classification model is that Minichromosome can not only achieve the minimum misclassification cost, but also obtain the maximum classification accuracy with fewer input features.