Skip to content

Investigating inductive bias of various training methods

Notifications You must be signed in to change notification settings

avisemler/InductiveBias

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

4 Commits
 
 
 
 
 
 

Repository files navigation

InductiveBias

Investigating inductive bias of various training methods.

Planned training types:

  • random features + linear classifcation: use an untrained neural network with a random intialisation to extract features from the data and then train a linear classifier on those features.
  • SGD with typical cost function
  • SGD optimising for orthogonality/usefullness of features

Plan for comparing inductive bias:

  • probability that various generalisations are made (discretising by clasification for selected examples)

About

Investigating inductive bias of various training methods

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages