pyAL is an python library that implements common active learning strategies. The project is currently developed by the members of the Machine Learning Lab at IIT: (http://ml.cs.iit.edu).
This work is supported by NSF CAREER Award #1350337. http://nsf.gov/awardsearch/showAward?AWD_ID=1350337
Currently supported strategies:
- Random sampling
- Uncertainty sampling by Lewis and Gale, 1994. A sequential algorithm for training text classifiers. Proceedings of the 17th annual international ACM SIGIR conference on Research and development in information retrieval.
- Query-by-committee by Seung et al, 1992. Query by committee. Proceedings of the fifth annual workshop on Computational learning theory.
- Expected error reduction by Roy and McCallum, 2001. Toward optimal active learning through monte carlo estimation of error reduction. Proceedings of the Eighteenth International Conference on Machine Learning.
- Machine Learning Lab @ IIT (http://cs.iit.edu/~ml)
- Documentation: (http://iitml.github.io/AL/)
pyAL is tested to work under Python 2.7. The required dependencies to build the software are scikit-learn >= 0.15, NumPy >= 1.6.2, and SciPy >= 0.9.
This package uses distutils, which is the default way of installing python modules. To install in your home directory, use:
python setup.py install --user To install for all users on Unix/Linux:
$ python setup.py build
$ sudo python setup.py install
This repository is organized as follows:
al: contains the codedocumentation: contains the documentation of the classesfront_end: containts the GUIutils: contains the utilities
You can check the latest sources with the command:
git clone git@github.com:iitml/AL.git
- Example about how to use the GUI: (http://iitml.github.io/AL/code.html#id1)
from al.learning_curve import LearningCurve
learning_api = LearningCurve()
training_size, avg_accu, avg_auc = learning_api.run_trials(X_pool, y_pool, X_test, y_test, strategy, classifier, alpha, boot_strap_size, step_size, budget, num_trials)