Extreme Machine Learning: When the average model knows best

Abstract

Machine Learning and Artificial Intelligence are presented as the fix-all for all current day problems. Also in research it is experiencing a golden age. However, before a Machine Learning model can be used, it needs to be trained, which takes enormous quantities of training data. This stands in stark contrast to general academic datasets resulting from research projects. The latter give rise to small or even extremely small data sets (< 50 samples). In this seminar, we address the question: "Is it possible to train an ML model with 30 samples instead of 30.000.000?" To do so we consider simple linear and polynomial regression models, and show how these provide access to analytical models in the context of small data ML. We present a strategy to always obtain the best model and highlight caveats and ways to deal with them. These ideas are applied on small experimental data sets, as an example of what can be expected. The three examples cover modelling adhesive coatings, ink formulations, and optimal setting for spray coating applications. [1] "A machine learning approach for the design of hyperbranched polymeric dispersing agents based on aliphatic polyesters for radiation curable inks",