Subscribe by RSS

Graduate Seminar Series

Wed., Nov. 2, 2022 3:30 p.m.

Location: CL 305 and Live Stream

Speaker: W. Madusha De Silva

Title: A Novel Stacking-based Ensemble Learning Model for Cancer Diagnosis (610 kB)

Abstract:

According to the high variability among various breast cancer datasets, widely-used machine learning models are applied to breast cancer diagnosis. However, the robustness and generalization of these models to assist clinical diagnosis are debated recently. Therefore, a stacking-based ensemble learning model is proposed in this study. This model consists of a two-layer learning structure, and its classifier combination is determined by the proposed stacking method. This model will apply to three different breast cancer datasets including Breast Cancer Ultrasound (BCU), Wisconsin Breast Cancer (WBC), and Mammographic Mass (MM), and will evaluated using classification accuracy and robustness in this study. In this seminar, I will perform the feature selection method to identify relevant features and remove irrelevant and redundant features from three datasets. And I will evaluate some base models to construct the stacking model with a 10-fold cross-validation strategy.

Live Stream: