Work on paper stories

[thinkh]

• Generalization of ML
• Hyper parameter influence on model selection
• Replication of notable other papers (Sharp Minima, Small Batch Learning, ...)

[NMO13]

@gfrogat What is generalization of ML? And what is Sharp Minima? And which papers are available that we can replicate?

[mstreit]

On Large-Batch Training for Deep Learning: Generalization Gap and Sharp Minima https://arxiv.org/abs/1609.04836

[gfrogat]

• Generalization of ML should be "model generalization error"
• sharp minima: have a look at the paper Marc linked

[gfrogat]

Paper story: Tool allows for comparison of performance between models and across multiple dataset folds.

Main focus is generalization error of model.