The original dataset SECOM is obtained from the the UC Irvine Machine Learning Repository (https://archive.ics.uci.edu/ml/datasets/secom). Then, each
sample is transformed to an image, with each pixel representing a feature. Therefore, image processing mechanisms such as convolutionary neural networks can be utilized for classification.