The authors attempt to compare their own method to other conventional cancer classification methods and try to benchmark results. The baseline was calculated using logistic penalized linear regression and linear support vector machines (widely known methods for cancer classification). Gradient Boosting Decision Tree (GBDT) and Multiple Layer Perceptron (MLP) models were also evaluated.
At MNM, quality and high performance are our driving forces. Therefore, we are developing artificial intelligence methodologies, such as gene convolutional autoencoders and the random similarity forest, to address the AI 'black box' problem, so that our models are interpretable and explainable.