Classification task using 2D sequences input

layers = [ ...
    sequenceInputLayer([48 3 1]])
convolution2dLayer([3 3], 1)
bilstmLayer(20)
fullyConnectedLayer(2)
classificationLayer
];

trainNetwork(feature, label, layers, options)

`feature` is a 600x1 cell and it's element is 48x3x800 matrix

`label` is 600x1 cell and it's elements is 1x800

I want to map a vector of size 48x3 to label

But got this error: training sequence have dim 48 3 800 but input layer have dim 48 3 1, should I create a new data store or use `arrayDatastore`

Answer 1

[Xie Shipley]

change `feature` from 48x3x800 to 48x3x1x800