## Expected Model Output Change (EMOC)

Source code for methods described in the following papers:

- Active learning and discovery of object categories in the presence of unnameable instances,
  C Käding, A Freytag, E Rodner, P Bodesheim, J Denzler,
  Computer Vision and Pattern Recognition (CVPR), 2015

- Large-Scale Active Learning with Approximations of Expected Model Output Changes,
  C Käding, A Freytag, E Rodner, A Perino, J Denzler,
  German Conference on Pattern Recognition (GCPR), 2016

- Watch, Ask, Learn, and Improve: A Lifelong Learning Cycle for Visual Recognition,
  C Käding, E Rodner, A Freytag, J Denzler,
  European Symposium on Artificial Neural Networks (ESANN), 2016

If you use parts of the code, please cite the corresponding papers.


##### Dependencies

- Python 2.7
- numpy
- scipy
- scikit-learn


##### Usage

1. define setup (see example_setup.cfg)
2. precompute setup (run evaluation/PrecomputeExperimentalSetup.py setup.cfg)
3. start experiment (run evaluation/RunExperiment.py setup.cfg)
4. see results (stored in results.mat)