Learning Representation Of Multi-View Data