WorksheetsUnderstanding Convolutional Neural Networks
Total questions: 18
Worksheet time: 9mins
What is the principle of convolutional layers compared to fully-connected layers?
Convolutional layers use a floating window to process pixels.
Fully-connected layers are more efficient in computation.
Convolutional layers are dependent on the shape of images.
Fully-connected layers have fewer parameters.
What is the objective of the convolution operation?
To extract high-level features from the input image.
To increase the number of parameters.
To reduce the dimensions of the image.
To apply filters to the image.
What is the purpose of zero padding in convolutional layers?
To preserve the size of the original image.
To increase the number of parameters.
To reduce the computational efficiency.
To enhance the quality of the image.
What does stride control in convolutional layers?
The size of the filter.
How the filter convolves through the input layer.
The number of channels in the input layer.
The dimensions of the output image.
What is the role of pooling layers in CNNs?
To reduce the dimensions of feature maps.
To increase the number of parameters.
To apply convolution operations.
To flatten the output from convolutional layers.
What is the main advantage of max pooling?
It selects the maximum pixel value.
It reduces the number of parameters.
It increases the computational efficiency.
It preserves all pixel values.
What is the purpose of regression algorithms?
To predict the value of an output variable.
To classify data into categories.
To reduce the dimensions of data.
To enhance the quality of images.
What does R Squared (R2) indicate in regression models?
The performance of the model.
The loss value of the model.
The accuracy of predictions.
The number of parameters in the model.
Which layer in a CNN is primarily responsible for detecting edges and textures in an image?
Pooling Layer
Convolutional Layer
Fully-Connected Layer
Input Layer
What is the role of padding in a convolutional layer?
To increase the number of features
To shrink the image size
To preserve image dimensions after convolution
To activate more neurons
Which technique helps reduce the number of parameters in a CNN?
Fully-connected layers
Input resizing
Pooling layers
Adding more channels
Why is Mean Absolute Error (MAE) used in regression model evaluation?
It penalizes large errors more than small ones
It shows percentage errors
It gives the average magnitude of prediction errors
It explains variance in data
Which regression algorithm is best suited for modeling a linear relationship between variables?
Decision Tree Regression
Support Vector Regression
Linear Regression
Polynomial Regression
A company observes that the sales data forms a curve when plotted against advertising spend. Which regression technique is most appropriate?
Linear Regression
Support Vector Regression
Polynomial Regression
Ridge Regression
Which metric gives the average of the absolute errors between predicted and actual values?
Mean Squared Error
Mean Absolute Error
Root Mean Squared Error
R2 Squared
Why is MSE sometimes preferred over MAE?
It penalizes small errors more
It penalizes larger errors more
It is less sensitive to outliers
It is always lower than MAE
Which metric is best when you need to express error as a percentage?
MAE
RMSE
MSE
MAPE
What does a high R² value (close to 1) indicate?
High error
Poor model performance
Model explains most of the variability in target
Model is underfitting
