Mcfly: Automated deep learning on time series
Dafne van Kuppevelt, Christiaan Meijer, Florian Huber, Atze van der Ploeg, Sonja Georgievska, Vincent T. van Hees
Abstract
Deep learning is receiving increasing attention in the scientific community, but for researchers with no or limited machine learning experience it can be difficult to get started. In particular, the so-called hyperparameter selection, which is critical to successfully train a model, requires a good understanding of deep learning and some experience training models. We present mcfly, a Python package for deep learning for time series classification, designed to ease training by providing automated hyperparameter selection. We evaluated mcfly by organizing two workshops.
Topics & Concepts
Python (programming language)Deep learningComputer scienceHyperparameterArtificial intelligenceMachine learningSelection (genetic algorithm)Programming languageTime Series Analysis and ForecastingAnomaly Detection Techniques and ApplicationsMusic and Audio Processing