Page 72 - Big Data Analytics for Intelligent Healthcare Management

4.4 PROPOSED MODEL 65

Class 1
Training instance
K=3 Class 2
Distance K=1

?
New example
to classify

FIG. 4.5
K-NN to classify new instance.
Data from Medium, A Quick Introduction to K-Nearest Neighbors Algorithm, 2018. Available from: https://medium.com/@adi.
bronshtein/a-quick-introduction-to-k-nearest-neighbors-algorithm-62214cea29c7, Accessed 10 June 2018.

ConvNets
as Feature Dimension
BreakHis feature vectors reduction by
extractor PCA
SVM

ResNet50 Inception K-NN
40× 200× Resnet
100× 400× Inception V2 Xception
V3 Classification
Magnification factor
Logistic
regression

Performance
analysis

FIG. 4.6
Proposed model.

depending on the explained variance ratio, the dimension of the feature vector can be reduced. Then the
reduced feature set is passed to the classifiers to perform binary classification to automate the classi-
fication of benign and malignant images. Classification is performed by three different classifiers are
LR, SVM, K-NN. All the feature extraction, dimension reduction, and classification is done per
magnification.