Dear researchers

I have a hyperspectral image from FTIR (.mat)

Content and shape of the file

--- Content of key: 'dY' --- 480

--- Content of key: 'dX' --- 480

--- Content of key: 'wn' ---

0 0 952.0 1 956.0 2 960.0 3 964.0 4 968.0 .. ... 208 1784.0 209 1788.0 210 1792.0 211 1796.0 212 1800.0

[213 rows x 1 columns]

--- Content of key: 'r' --- 0 1 2 3 4 5 6 \ 0 0.002171 0.002171 0.001737 0.001303 0.000434 0.000000 -0.000434 1 0.002606 0.002606 0.002171 0.001303 0.000869 0.000000 -0.000434 2 0.003040 0.002606 0.00260

[230400 rows x 213 columns]

r shape: (230400, 213), wn shape: (213,)

After preprocessing,( normalization, baseline correction)

--- Keys and Shapes ---

ab | type: ndarray | shape: (212, 230401)

wn | type: ndarray | shape: (212,)

wh | type: ndarray | shape: (2,)

I want to perform DL on this data for binary classification.

I previously tried with ResNet models and faced poor performance.

Any basic guidance on this would be of great help.

Additional information:

There is a class imbalance.

Best

Rahul