Convolutional Neural Networks

Which one is a non-linear activation function?

• ReLu

• Sigmoid

• Tanh

• All of the above

For an image recognition problem, which neural network architecture would be better suited to solve the problem?

• Multilayer Perceptron

• Convolutional Neural Network

• Single-Layer Perceptron

• None of the above

Ten neurons in the hidden layer and five neurons in the output layer; how many weights and biases parameters?

• More than 50

• Less than 50

• 50

• None of the above

Given an image of size 27 x 27 and a filter of size 5 x 5 with a stride of 2 and no padding. What is the output size?

• 12 x 12

• 13 x 13

• 23 x 23

• 27 x 27

RGB image size = 64 x 64 x 3, 10 filters, each of shape 5 x 5 x 3 with stride 1 and pad 2. What is the output size?

• 60 x 60 x 3

• 60 x 60 x 10

• 64 x 64 x 3

• 64 x 64 x 10

Input size = 27 x 27, filter size = 7 x 7, stride = 1. How many layers of zeros are added on each side so that the output size remains 27 x 27?

• One

• Two

• Three

• Four

                 _____ produces a discrete probability distribution vector.

• Softmax

• ReLu

• Linear

• Tanh

RGB image size = 32 x 32 x 3, 10 filters, each of shape 5 x 5 x 3 with stride 1 and pad 0. How many weights and biases for the convolutional operation?

• 225

• 320

• 456

• 760

Given the shown input matrix of shape 4 x 4. What will be the output on applying a max pooling of size 2 x 2 with a stride of 2?

• 6723

• 6733

• 6834

• 7834

What value would be in place of the question mark?

• 3

• 4

• 5

• 6