Project Details

Description

The objective is to find ways to explore the space of hyperparameters in a more intelligent way. The idea is to take into account the dependencies between these pa-rameters to sample the space in a more targeted way without (too) sacrificing the performance (e.g., precision, error) of the model. We will also incorporate some domain knowledge in the process to avoid high infrastructure costswill thus be reduced and innovations as well as productivity brought by deep learning will be accessible to structures of all sizes.
Short titleSmartTune
StatusActive
Effective start/end date1/07/2128/06/25

Keywords

  • Deep Learning
  • Software Engineering
  • Variability