understanding keras 's layer



dot - computes a dot product between 2 tensors. Just like doing multiplication.

conv2d - creates a convulsion - a square box that you hover over a 2d image matrix.

maxPool2d - taking the max values out of a stride after we iterate our image with a predefined filter size. Small little window walks through our image to produce a matrix - which made up of biggest value of a filter.

To see more please read from this link here.

https://www.quora.com/What-is-max-pooling-in-convolutional-neural-networks

dense layer - it is a linear function which maps input directly to output based on a predefined weight. sometimes weight can be a softmax function.



dropout is a techniques used to prevent over fitting in neural network. Performs averaging and prevent co-adaptions.





Comments

Popular posts from this blog

The specified initialization vector (IV) does not match the block size for this algorithm