WorksheetsConvolution Neural Network
Total questions: 12
Worksheet time: 6mins
What is the primary purpose of the convolution operation in image processing?
To reduce the size of an image
To extract features from an image using a defined kernel matrix
To increase the color intensity of an image
To convert an image from grayscale to color
What is the primary purpose of adding padding in a convolution operation?
To increase the color intensity of the image edges
To shrink the original image size during convolution
To prevent the image from overlapping during convolution
To preserve the size of the original image and ensure edge features are utilized
What does the term "stride" refer to in the context of convolution operations in image processing?
The size of the kernel used for the convolution
The number of pixels the kernel shifts over the input matrix
The amount of padding added to the image
The resolution of the input image
What is the primary function of a pooling layer in a Convolutional Neural Network (CNN)?
To increase the resolution of the image representation
To progressively reduce the spatial size of the representation, reducing network complexity and computational cost
To apply convolutional operations with larger kernel sizes
To add padding to the input image
What is the primary purpose of Max Pooling in a Convolutional Neural Network (CNN)?
To take the average of a region in the image
To select the maximum value from a region, highlighting the most important features
To increase the resolution of the image
To darken the background of the image
What is the primary difference between Average Pooling and Max Pooling in a Convolutional Neural Network (CNN)?
selects the minimum value from a region, while Max Pooling selects the maximum value.
Average Pooling blends all values in a region, while Max Pooling only retains the maximum value.
Average Pooling increases the resolution, while Max Pooling decreases it.
Average Pooling is used for color images, while Max Pooling is used for grayscale images.
What is the primary purpose of Convolutional Neural Networks (CNNs) in deep learning?
To perform natural language processing tasks.
To process grid-like data such as images and video frames.
To model sequential data like time series and speech.
To improve the efficiency of search algorithms.
What is the primary application of the LeNet-5 architecture?
Object detection in real-time video.
Speech recognition.
Handwritten digit recognition.
Natural language processing.
Which of the following statements about AlexNet is TRUE?
introduced by Yann LeCun and has 50 million parameters with 7 layers.
first CNN architecture to use ReLUs as activation functions and won the 2012 ImageNet ILSVRC challenge with a top-5 error rate of 17%
introduced by Geoffrey Hinton, has 80 million parameters, and uses sigmoid activation functions.
the first CNN to use CPUs for training, achieving a top-5 error rate of 26% in the 2012 ImageNet ILSVRC challenge.
Which of the following statements about GoogLeNet is TRUE?
created by Yann LeCun and has approximately 60 million parameters, achieving a top-5 error rate of around 17% in the ILSVRC 2014 challenge.
developed by Christian Szegedy from Google Research, has roughly 6 million parameters and achieved a top-5 error rate of below 7% in the ILSVRC 2014 challenge.
introduced by Geoffrey Hinton, has more parameters than AlexNet, and uses ReLU activation functions to achieve a top-5 error rate of around 10%.
uses inception modules to reduce the depth of the network and was the first CNN to win the ILSVRC challenge with a top-5 error rate of 26%.
Which of the following statements about ResNet is TRUE?
developed by Yann LeCun, achieved a top-5 error rate of 6.5% with a network composed of 50 layers, and it does not use skip connections.
introduced by Kaiming He, won the ILSVRC 2015 challenge with a top-5 error rate of 3.6% and is known for its use of skip connections to facilitate training of very deep networks.
developed by Geoffrey Hinton, has 152 layers, and relies on dropout layers to improve its performance, achieving a top-5 error rate of 5%.
uses a shallower architecture compared to earlier CNNs and achieved a top-5 error rate of 10% in the ILSVRC 2015 challenge.
Which of the following statements about DenseNet is TRUE?
introduced a sparse connection scheme, where each layer is only connected to the preceding layer, which helps in mitigating the vanishing and exploding gradients problem.
uses a densely connected network design where each layer’s output is connected to the input of every subsequent layer, addressing issues related to vanishing and exploding gradients.
developed to have fewer layers than traditional CNNs, achieving better performance by avoiding the vanishing gradient problem through reduced depth.
relies on dropout techniques to combat the vanishing gradients issue, resulting in a significant reduction in the number of parameters compared to other architectures.
