Efficient simulation experiments for large-scale parameter optimisation of machine learning approaches in natural language processing (E-SPar)
Third party funded individual grant
Acronym: E-SPar
Start date : 01.10.2016
End date : 30.09.2017
Project details
Scientific Abstract
Our aim is to develop optimized implementations of memory-intensive machine learning approaches in natural language processing, which can be deployed on HPC clusters in order to carry out simulation experiments for the systematic optimization of model parameters. This case study focuses on matrix factorization and deep learning methods in distributional semantics.
