a thin scikit-learn
style wrapper for tensorflow
framework
The wrapper takes care of training (fit()
) and prediction(predict()
) infrastructure.
What is left to do is to specify the network topology in _create_network()
method and
loss function in _create_loss()
method.
There are two subclasses of the tflearn
class
which come with ready to use _create_loss()
method:
rtflearn
for regressionctflearn
for classification (in progress)
tf_lasso.py
: Lasso regression, a simplest example
tf_factorization_machine.ipynb
: Factorization machines
gan_dense.py
: Generative adversarial network with dense perceptron layers