I believe that your interpretation is not correct. Based on my brief reading of the paper, the model contains 1) some known architecture for embedding the modality, and 2) the feature reconstruction transformer network,…
I believe that your interpretation is not correct. Based on my brief reading of the paper, the model contains 1) some known architecture for embedding the modality, and 2) the feature reconstruction transformer network,…