Armin Arjmand

icon indicating a website link arminarj.github.io

icon company Northeastern Univeristy icon location Boston, MA icon location Mehdi (Armin) Arjmand, Research Assistant, Enthusiastic in Multimodal Interaction and Conversational Agents.

Results 4 comments of Armin Arjmand

Model unstable when using transformation

Hi @hilmar05, Thanks for reaching out. Have you tried adding some small values to Tval Matrix prior to using the SVD operation, just as it is implemented in this [line](https://github.com/Michaelvll/DeepCCA/blob/2fa648e444fa6ae9135c52cceee1b653cc1d44d0/objectives.py#L80)...

Calculation of trace norm when using all singular values

Hi @jameschapman19, Thank you for opening a new issue and Thank you for your suggestion. I have made a new branch based on your valuable suggestion, please double-check it before...

Calculation of trace norm when using all singular values

Hi @Eason24, This type of error usually happens when you feed a zero value to the ``torch.sqrt()`` function. As for a short-term solution, I suggest you use the ``self.use_all_singular_values`` as...

the loss value dosen't change using GCCA_loss

Hi @cyx669521, Thanks for your interest in this repo. But it seems the loss objective is not always working and I am still working to understand its bottleneck and conditions....