Convolutional Neural Network Theoretical Course (Course VIII)July 17, 2020 2020-08-04 10:54
Convolutional Neural Network Theoretical Course (Course VIII)
- Welcome to Course VIII!
- Introduction to Convolutional Neural Networks
- The Convolution Operation
- Stride and Calculation of Output Size
- The Pooling Operation
- The Convolution/Pooling Operation for RGB images
- Padding an Image
- Building a Convolutional Neural Network
- Training a Convolutional Neural Network
- End of Course
Training a Convolutional Neural Network
The training process of a Convolutional Neural Network is also similar to that of Training a Deep Neural Network. The training process goes through multiple finite iterations where data is feed-forwarded into the network and then, weights are adjusted using back-propagation (gradient descent) until the loss of the network reaches a certain threshold. Once the model reaches the threshold, training is stopped.
However, the parameters that CNNs learn during training are slightly different than that of ordinary Deep Neural Network. In CNN, the training objective is to optimize the pixel values of the kernels of the convolutional layers along with the connections of the fully connected layers. In other words, the weights of the convolution filter/kernel is also a learnable parameter in CNNs.
The pooling layer doesn’t have any weight assigned to it since it is just taking the maximum or average value of the output of the convolutional layer.