2. CNN Architecture

2. CNN Architecture. . Convl Conv2 Conv3 æ3x3 Conv4 64@3x3 64@3x3 s M ax-poolhg M ax-pooing Max-poolhg Input Layer Feature Extractbn Layer M ax-poolhg 256 Ckssificatbn Layer.

1-Convolutional layer 2-pooling layer 3-ReLU layer 3.1-The sigmoid function 3.2-hyperbolic tangent 4-Fully connected layer/output layer(the classification layer).

2. CNN Architecture. 1-Convolutional layer output=.

2. CNN Architecture. 2-pooling layer 2.1-MAX Pooling 2.2 Average Pooling.

2. CNN Architecture. 3-ReLU layer Some of the ReLU variants Softplus(SmoothReLU) Noisy ReLU Parametric ReLU ExponentialReLU(ELU).

2. CNN Architecture. 4-Fully connected layer/output layer (the classification layer) 4.1 softmax.

2. CNN Architecture. Overview of some models 1-LeNet-5 CNN_based model 2-Stacked Denoising Autoencoders network(SDAE) 3- Deep Belief Network (DBN) 4-Artificial neural network(ANN) 5-Custom CNN 6-VGG 19 NETWORK.