Automatic multiclass classification of laryngeal cancer using deep convolution neural networks

In this work, the classification of laryngeal cancer is attempted using deeply learned features obtained using Inception V3, Squeezenet, and VGG-16 embedders in the Orange toolbox. Machine learning algorithms such as KNN, SVM, random forest, decision tree, and neural network classifiers are employed to classify the stages or categories of laryngeal cancer. The ranking of deep learning feature values is carried out using state-of-the-art metrics such as information gain, information gain ratio, chi-square, and reliefF. It is observed that the performance of the algorithms is affected by the cross-validation. In this work, the classification of laryngeal cancer is attempted using deeply learned features obtained using Inception V3, Squeezenet, and VGG-16 embedders in the Orange toolbox. Machine learning algorithms such as KNN, SVM, random forest, decision tree, and neural network classifiers are employed to classify the stages or categories of laryngeal cancer.image