Created by Philipp Krähenbühl.

This code implements the initialization presented in our arXiv tech report, which is under submission at ICLR 2016.

This is a reimplementation and currently work in progress. Use at your own risk.

This code is released under the BSD License (refer to the LICENSE file for details).

If you find our initialization useful in your research, please consider citing:

@article{krahenbuhl2015data,
  title={Data-dependent Initializations of Convolutional Neural Networks},
  author={Kr{\"a}henb{\"u}hl, Philipp and Doersch, Carl and Donahue, Jeff and Darrell, Trevor},
  journal={arXiv preprint arXiv:1511.06856},
  year={2015}
}

Checkout the project and create a symlink to caffe in the magic_init directory:

ln -s path/to/caffe/python/caffe caffe

Here is a quick example on how to initialize alexnet:

python magic_init.py path/to/alexnet/deploy.prototxt path/to/output.caffemodel -d "path/to/some/images/*.png" -q -nit 10 -cs

Here -d flag allows you to initialize the network using your own images. Feel free to use imagenet, Pascal, COCO or whatever you have at hand, it shouldn't make a big difference. The -q (queit) flag suppresses all the caffe logging, -nit controls the number of batches used (while -bs controls the batch size). Finally -cs rescales the gradients accross layers. This rescaling currently works best for feed-forward networks, and might not work too well for DAG structured networks (we are working on that).

To run the k-means initialization use:

python magic_init.py path/to/alexnet/deploy.prototxt path/to/output.caffemodel -d "path/to/some/images/*.png" -q -nit 10 -cs -t kmeans

Finally, python magic_init.py -h should provide you with more help.

If you're numpy implementation is based on openblas, try disabeling threading export OPENBLAS_NUM_THREADS=1, it can improve the runtime performance a bit.