active learning in python
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.

alp: active learning in python

This is a python implementation of some popular active learning techniques, including uncertainty sampling and query-by-committee. It is built on top of numpy, scipy, and sklearn. I wrote this for my own learning purposes; it is not particularly efficient.


from active_learning.active_learning import ActiveLearner
from sklearn.model_selection import train_test_split
from sklearn.linear_model import LogisticRegression
from sklearn.datasets import make_classification

X, X_unlabeled, y, y_oracle = train_test_split(*make_classification())
clf = LogisticRegression().fit(X, y)

AL = ActiveLearner(strategy='entropy')
AL.rank(clf, X_unlabeled, num_queries=5)

I did this while working through Burr Settles' excellent literature survey. If you're interested in this topic you should read it.