- Numpy (
pip install numpy
) - gfortran (
brew install gfortran
) - Scipy (
pip install scipy
) - Freetype (
brew install freetype
) - Matplotlib (
pip install matplotlib
) - iPython (
pip install ipython
) - Readline (
pip install readline
) - SVM Light (Link to project)
- Scikit-learn (pip install scikit-learn)
- Theano (pip install theano)
The following programs are assumed to be defined in your PATH:
svm_hmm_learn
svm_hmm_classify
-
ICML_BIRD_DATA_PATH
should be defined as the absolute path to the data directory. E.g.,export ICML_BIRD_DATA_PATH="/var/data/icml-bird/"
This is where all the training data is stored on your computer. See the below section on how to organize it.
-
ICML_BIRD_MODEL_PATH
should be defined as the absolute path to the model directory. E.g.,export ICML_BIRD_MODEL_PATH="/var/models/icml-bird/"
This is where any special pre-processing data or model parameters will be saved, as well as any predictions / evaluation for a particular model. See below for more information about its organization.
It seems easiest to just keep the data directory out of the repository since
it holds a lot of binary data that shouldn't change too much. The only details
mentioned here are those that the codebase makes use of. You can place your
data directory wherever you wish, so long as you set the path variable
ICML_BIRD_DATA_PATH
to point to the correct location, and set up your data
directory in the proper way.
The top-level directory. Contains:
phylogenetic_data.txt
- This file can be found here. It contains information about the relative phylogenetic distance between bird species.weather.txt
- This file can be found here. It contains information about the weather during each of the test recordings.species_numbers.csv
- The official map between species and ID number. Link.
Contains the directories:
Contains the directories:
This is where we contain our submission files (saved as text files).
The top-level directory.