CAMELYON 17 Challenge: A Comparison of Traditional Machine Learning (SVM) with the Deep Learning Method

Table 3

A summary table of the comparison between SVM and random forest.

Classifiers	Positives	Negatives
SVM	(1) In nonlinear separable classes, the Gaussian basis function and polynomials are valuable kernel functions; (2) Works well with datasets with a large number of characteristics; (3) Be good at linear classifier and data separation	(1) When dealing with unlabeled datasets is computationally costly; (2) A speed limit when selecting the kernel function settings; (3) Speed and size limitations during both the training and testing phases
Random forest	(1) Even as the number of trees grows, the generalization error converges; (2) Difficult to overfit a single feature	(1) Longer classification times will result from larger input datasets; (2) Could overfit training data