A series of automated processes facilitate data preparation and algorithm and feature selection. Furthermore, to overcome difficulties in integrating different sources and scales, NAUTILUSÂŪ is capable of exchanging, filtering, and merging the various input data into a Unified “Smart Table” suitable for the input of ML algorithms, taking into account the spatial relationship between them.
NAUTILUSÂŪ incorporates several standard pointwise Machine Learning algorithms (SVM, XGBoost, Neural Networks, etc..), as well as 1D/2D/3D Convolutional Neural Networks. All combinations of Algorithms/Features can be tested to advise the best solution for a given prediction. Furthermore, the platform saves the entire workflow with its parametrization, which can be deployed over the whole database when a model achieves satisfactory criteria.
