A data scientist implements a convolutional neural network in SAS DLPy for image classification. After adding Conv2D and pooling layers, they add a Flatten layer. What is the purpose of the Flatten layer?
Select an answer to reveal the explanation.
Short Explanation
Here's the deal — the correct answer is the one that converts the 3D feature map tensor into a 1D vector for fully connected layers. A Flatten layer reshapes the convolutional/pooling output so Dense layers can receive it.
Full Explanation
The correct option converts the 3D feature map tensor (height × width × channels) into a 1D vector for fully connected layers, because a Flatten layer reshapes convolutional/pooling output into a single 1D feature vector for Dense layers. Normalizing pooling output to zero mean and unit variance describes Batch Normalization. Averaging across spatial dimensions describes Global Average Pooling. Applying dropout regularization describes Dropout layers.