wayground logo

Free Printable Worksheets

Font size

S
M
L
XL
Worksheets

ImageNet Classification Reading PoW

Total questions: 10

Worksheet time: 4mins

Name
Class
Date
1.

Who is the primary author of the paper "ImageNet Classification with Deep Convolutional Neural Networks"

a)

Ilya Sutskever

b)

Alex Krizhevsky

c)

Yann LeCun

d)

Geoffrey Hinton

2.

Which activation function did the AlextNet architecture use?

a)

f(x) = tanh(x)f\left(x\right)\ =\ \tanh\left(x\right)

b)

f(x) = (1 + ex)1f\left(x\right)\ =\ \left(1\ +\ e^{-x}\right)^{-1}

c)

f(x) = max(0, x)f\left(x\right)\ =\ \max\left(0,\ x\right)

3.

Choose the max pooling configuration used in the AlexNet architecture.

a)
b)
c)
4.

AlexNet uses a technique for reducing complex co-adaptation of neurons. Which of the following do they correspond to?

a)

50% of all activations of the network were set to zero and also did not participate in backpropagation

b)

Local response normalization

c)

Combining prediction of different models

d)

Dropout of 0.5

5.

What type of neural network does AlextNet use?

a)

Feedforward Neural Network

b)

Perceptron

c)

Convolutional Neural Network

d)

Long Short Term Memory

6.

What approaches of data augmentation did AlexNet opt?

a)

Adding random crops of images, their rotations and flips in training

b)

Light conditioning

c)

Test time augmentation by prediction on corner and center patches of image

d)

PCA color augmentation

7.

On what did the AlextNet train on?

a)

CPU

b)

1 GPU

c)

2 GPU

d)

3 GPU

8.

Which of the following statements are true based on the graph?

a)

Implies the reason why the network uses saturating activations

b)

The dotted lines corresponds to tanh and solid line to ReLU

c)

Use of ReLU increased training speed of the network by six times

d)

The dotted lines corresponds to ReLU and solid line to tanh

9.

How many layers did AlexNet have?

a)

8 convolutional layers

b)

5 convolutional and 3 fully connected

c)

8 fully connected layers

d)

3 convolutional layers and 5 fully connected

10.

Which of these do you consider the most crucial to the success of the architecture.

a)

Down sampling images to fixed resolution of 256 x 256

b)

ReLU non linearity

c)

GPU cross parallelization

d)

Normalization layer

e)

Overlapping pooling