NN-Visuals: Neural Network Architectures
Fully Connected NN
Classic feedforward network where each neuron is connected to every neuron in the next layer.
$$y = f(\mathbf{W}x + b)$$
LeNet
Pioneering convolutional neural network for image recognition, using convolution and pooling layers.
$$\text{Conv} \rightarrow \text{Pool} \rightarrow \text{FC}$$
AlexNet
Deep CNN that revolutionized computer vision with ReLU, dropout, and data augmentation.
$$\text{Conv} \rightarrow \text{ReLU} \rightarrow \text{Pool} \rightarrow \text{FC}$$