HlEvag/DMCM

Data problem

Opened this issue · 3 comments

Dawn-LLL commented

Hello, I learned a lot after reading the paper you wrote. I would like to ask if the source data has been dimensionally reduced (eg: dimensionality reduction from 128 to 100)? I saw that the paper said "Thenumber of bands D and window size S of the input hyperspectral patch are set to 100 and 9” but the band of the source data in the code is 128. Is the source data input in 128 dimensions or 100 dimensions?
Thank you so much

HlEvag commented
Dawn-LLL commented

Thanks for your answer, for example the PU data, shown in the code is,
"parser.add_argument("-c","--src_input_dim",type = int, default = 128)
parser.add_argument("-d","--tar_input_dim",type = int, default = 103) "
My understanding is that dimensionality reduction was not used when the data input. But the dimensionality reduction is carried out in
"self.target_mapping = Mapping(TAR_INPUT_DIMENSION, N_DIMENSION)
self.source_mapping = Mapping(SRC_INPUT_DIMENSION, N_DIMENSION)".
Is this understanding correct?
Thanks again!!

HlEvag commented