An implementation of Deep Canonical Correlation Analysis (DCCA or Deep CCA) with Keras and Tensorflow. It is based on VahidooX's Theano implementation.
Because the loss function of VahidooX's implementation is written with Theano, and does not work on the Tensorflow backend. I rewrote the loss function in Tensorflow. And modify some other details to run the code on Tensorflow. Other than that, most of the code are borrowed from VahidooX's implementation.
See VahidooX/DeepCCA(https://github.com/VahidooX/DeepCCA) for more information.
Note: Tensorflow version needs to be 1.2+