Brantner, H. (2021). Neural network arena: Investigating long-term dependencies in deep models [Diploma Thesis, Technische Universität Wien]. reposiTUm. https://doi.org/10.34726/hss.2021.88580
This work should help to objectively compare various machine learning models used to process regularly sampled time-series data. It should outline the weaknesses and strengths of the benchmarked models and determine their primary domain of use. Moreover, as there are many models benchmarked, their relative expressivity across various application domains can be compared reasonably well. Another aim is to provide an overview of what architectures are currently available and how they can be implemented. Furthermore,the implemented benchmark suite should be reusable for future projects in the machinelearning domain.
