Skip to content

jhoffman/MaxMarginDomainTransforms

Repository files navigation

MaxMarginDomainTransforms

Author Judy Hoffman (jhoffman@eecs.berkeley.edu)

Package with code and demo usage for the paper: "Efficient Learning of Domain-invariant Image Representations" J. Hoffman, E. Rodner, J. Donahue, K. Saenko, T. Darrell International Conference on Learning Representations (ICLR), 2013.

Setup:

Run:

  • To reproduce results from the paper run "demo.m" and pick the source and target domains by changing the Config.m file.
  • To run on new data you should look at TrainMmdt.m
    • [model, A] = TrainMmdt(labels, data, param)
    • Will return a model file that can be used with the predict() function from liblinear-weights
    • Must input a struct of the form labels.source, labels.target, data.source data.target. param should be a struct that contains the MMDT related parameters from Config.m (Namely: C_s, C_t, mmdt_iter)

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages