This study investigates the applicability of the Auto Machine Learning-based approach (AutoML) for analyzing microscopic printed document images to attribute that document to its source printer. In this perspective, AutoML, a new rising star of machine learning in practice, has shone brightly as it can satisfy the demand of Machine Learning practitioner communities. In this work, three candidates from popular Machine Learning models and two representatives from AutoML are nominated for a competition. The challenges of traditional methods and the merits of applying AutoML are highlighted through the experiments. Especially the power of ensemble methods to achieve the best possible model for our experimental dataset. Furthermore, the learnability of AutoML to the different levels of uncertainties of printed patterns is also recognized.
CITATION STYLE
Vo, P. Q., Dang, N. T., Nguyen, Q. P., Mai, A., Nguyen, L. T. T., Nguyen, Q. T., & Nguyen, N. T. (2022). Auto Machine Learning-Based Approach for Source Printer Identification. In Communications in Computer and Information Science (Vol. 1716 CCIS, pp. 668–680). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-981-19-8234-7_52
Mendeley helps you to discover research relevant for your work.