CIC IDS2017

Splitting Data

Scaled Training Data

Scaled Testing Data

Input Layer

Dense Embedding

ExpandDims (seq=1)

TabTransformer Block × N (MHA + FFN + Norm)

Flatten

Dense (128, ReLU)

Dropout (0.3)

Dense (Softmax, num_classes)

Predicting Attacks (Evaluation: Accuracy, Precision, Recall, F1, Confusion, ROC, PR)