Convolution layer

This is the main layer of feature extraction part of the convolution neural networks.

Convolution layer

They had a filter of shape (x,x), generally x=3 so shape of filter is (3,3).

Kernel

2	2	1
3	3	3
3	2	1

Image

2	3	5
3	3	3 (22)+(32)+(51)+(33)+(33)+(33)+(33)+(22)+(1*1)=56
3	2	1 Every pixel will change by 56 of image

Comments

Gradient Descent

In previous class we learn what is MLP? but we listen a term called gradient descent today we study gradient descent. Gradient descent In mathematics, gradient descent is a first-order iterative optimization algorithm for finding a local minimum of a differentiable function. The idea is to take repeated steps in the opposite direction of the gradient of the function at the current point, because this is the direction of steepest descent. Here it is explained by image

MLP( Multi Layer Perceptron )

Multilayer Perceptron Multilayer Perceptron is collection of Perceptron ANN(MULTILAYER PERCEPTRON) Backpropagation Backpropagation is a process to find weights and baises Step 1 : Initialize all weight and biases Step 2:Choose a Loss function Step 3: Apply Gradient Descent Technique to find correct weights and biases

Flatten

Flatten can convert 2D image to a 1D tensor Example-Image - [ 3 1 5 1 1 5] Flatten- [ 3 1 5 1 1 5 ]

Deep Learning

Search This Blog