https://medium.com/machine-learning-world/convolutional-neural-networks-for-all-part-ii-b4cb41d424fd

University of Michigan - Ann Arbor

https://www.coursera.org/learn/convolutional-neural-networks?specialization=deep-learning

Convolutional Neural Networks

just padding: (n*n) * (f*f) --> (n+2p-f+1) * (n+2p-f+1)
with padding and stride: (n*n) * (f*f)  --> floor(((n+2p-f)/s) + 1) *  floor(((n+2p-f)/s) + 1) 
multiple filters: (n*n*nc) * (f*f*fc)  --> (n-f_1) * (n-f+1) * nc 
where nc = number of filters 

Dimensions of Output

An efficient method that is used for object detection. First, train a convolutional neural network (CNN) on closely cropped pictures of the object. Next, create a window frame of a smaller size than the image and place it in a corner over the image. Run a CNN to determine if this window frame matches the cropped version of the object. Move the window to the next part of the image and detect the object again. Repeat this procedure for the entire image. Object detection is very computing intense and is prone to miss the object if the sizes of the window and the object don’t match.

Sliding window

Different CNN Methods

Examples of filter matrices 
Vertical helps detect vertical edges 
horizontal helps detect horizontal edges 
sobel is more robust

Filter matrices

Which of the following do you typically see as you move to deeper layers in a ConvNet? 

In order to be able to build very deep networks, we usually only use pooling layers to downsize the height/width of the activation volumes while convolutions are used with “valid” padding. Otherwise, we would downsize the input of the model too quickly.

Training a deeper network (for example, adding additional layers to the network) allows the network to fit more complex functions and thus almost always results in lower training error. For this question, assume we’re referring to “plain” networks.

Suppose you have an input volume of dimension 64x64x16. How many parameters would a single 1x1 convolutional filter have (including the bias)?

The TensorFlow Keras Sequential API can be used to build simple models with layer operations that proceed in a sequential order. If the model is non-linear or contains layers with multiple inputs or outputs, a Sequential model is not a good choice.

model = tf.keras.Sequential(\[
tensorflow.keras.layers.layer_name_1(arguments),
tensorflow.keras.layers.layer_name_2(arguments),
...
\])

Also, we can use .add() method to add layers incrementally to a Sequential model. To remove them, we can use .pop() method.


Learn Before

Related