Error using trainNetwork (line 191) Invalid network.

I receive the following error for my function given below. .Would you please help me to correct the function?

for the following input

YPred = trainAndPredictHADEL(XTrain, YTrain, XTest);

++

% Function to train and predict using HADEL model
function YPred = trainAndPredictHADEL(XTrain, YTrain, XTest)
% Define HADEL model architecture
numFeatures = size(XTrain, 2);
numClasses = numel(categories(YTrain));

% Feature-Level Attention
featureAttention = [
    fullyConnectedLayer(64, 'Name', 'fc_feature_attention')
    reluLayer('Name', 'relu_feature_attention')
    fullyConnectedLayer(1, 'Name', 'fc_feature_weights')
    softmaxLayer('Name', 'feature_attention_weights')
    ];

% Temporal Attention (not used for Iris dataset, but included for completeness)
temporalAttention = [
    sequenceInputLayer(numFeatures, 'Name', 'input_sequence')
    lstmLayer(64, 'OutputMode', 'sequence', 'Name', 'lstm_temporal_attention')
    fullyConnectedLayer(1, 'Name', 'fc_temporal_weights')
    softmaxLayer('Name', 'temporal_attention_weights')
    ];

% Combine into Hierarchical Attention
%%Bu kısmı hata verdi aşağıdaki gibi koydum
% hierarchicalAttention = [                           
%     featureAttention
%     temporalAttention
%     dotProductLayer(1, 'Name', 'weighted_output')
%     ];
hierarchicalAttention = [
    featureAttention
    temporalAttention
    ];

% Add classification layers
layers = [
    hierarchicalAttention
    fullyConnectedLayer(64, 'Name', 'fc_final')
    reluLayer('Name', 'relu_final')
    fullyConnectedLayer(numClasses, 'Name', 'fc_output')
    softmaxLayer('Name', 'softmax_output')
    classificationLayer('Name', 'output')
    ];

% Train the network
options = trainingOptions('adam', ...
    'MaxEpochs', 50, ...
    'MiniBatchSize', 8, ...
    'Verbose', false);
net = trainNetwork(XTrain, YTrain, layers, options);

% Predict on test data
YPred = classify(net, XTest);
end

++

Error

++

Error using trainNetwork (line 191)

Invalid network.

Error in trainAndPredictHADEL (line 54)

net = trainNetwork(XTrain, YTrain, layers, options);

Caused by:

Layer 'fc_feature_attention': Unconnected input. Each layer input must be connected to the output of another layer.

Layer 'input_sequence': An input layer must be first in the layer array.

++

Answer 1

[Cris LaPierre]

You are missing an input layer for your model. See the final error message: " Layer 'input_sequence': An input layer must be first in the layer array."

Your first layer is fullyConnectedLayer(64, 'Name', 'fc_feature_attention')

Perhaps the inputLayer page is helfpul.

[Fatih]

Thanks I managed that part. But now I have naother issue. In the following code, I cant connect concatenationLayer = concat to featureAttention & temporalAttention. Would you please help?

Error Message

Caused by:

Layer 'concat': Unconnected input. Each layer input must be connected to the output of another layer.

++

numFeatures = size(XTrain, 2);
numClasses = numel(categories(YTrain));
% Feature-Level Attention
featureAttention = [
    fullyConnectedLayer(64, 'Name', 'fc_feature_attention')
    reluLayer('Name', 'relu_feature_attention')
    fullyConnectedLayer(1, 'Name', 'fc_feature_weights')
    softmaxLayer('Name', 'feature_attention_weights')
    ];
% Temporal Attention (not used for Iris dataset, but included for completeness)
temporalAttention = [
    fullyConnectedLayer(numFeatures, 'Name', 'input_sequence')
    lstmLayer(64, 'OutputMode', 'sequence', 'Name', 'lstm_temporal_attention')
    fullyConnectedLayer(1, 'Name', 'fc_temporal_weights')
    softmaxLayer('Name', 'temporal_attention_weights')
    ];
% Combine into Hierarchical Attention
hierarchicalAttention = [
    featureInputLayer(numFeatures, 'Name', 'input_features') % Input layer for features
    featureAttention
    temporalAttention
    concatenationLayer(1, 2, 'Name', 'concat') % Concatenate feature and temporal attention outputs
    ];

[Cris LaPierre]

I think the issue here is that you have defined the concatenation layer to have 2 inputs, but have only provided one. See the concatenationLayer doc page.

Perhaps this layer is not doing what you think. Try removing it.

[Fatih]

Thanks a lot. Once I analyzed it now works.