Robustifying Algorithms of Learning Latent Trees with Vector Variables

We consider learning the structures of Gaussian latent tree models with vector observations when a subset of them are arbitrarily corrupted.