In linear algebra, a scalar is a single number, while a vector is an array of numbers. What is a tensor?

• A tensor is a vector with more than three dimensions.

• A tensor is a data structure representing a multi-dimensional array.

• A tensor is an array of vectors.

• A tensor is a special type of matrix with only positive values.

The goal of gradient-based optimization is to minimize a function by taking steps proportional to the negative of the gradient. This is because the gradient points in the direction of the function's steepest _________________.

• convexity

• decrease

• increase

• linearity

Which of the following is an example of a hyperparameter?

• Learning rate

• Weights of a neural network

• The input data

• Bias of a neural network

Which of the following describes the relationship between bias and variance?

• A high-bias model is likely to have low variance.

• High bias and high variance occur simultaneously.

• Bias and variance are independent of each other.

• A high-variance model is likely to have low bias.

Stochastic gradient descent (SGD) differs from standard gradient descent in that it computes the gradient on a _______________ instead of the entire dataset.

• validation set

• randomly selected subset

• larger batch of data

• single data point or small batch

A fully-connected feedforward network is called 'deep' when it has more than one _______________ layer between the input and output layers.

• regularization

• hidden

• activation

• softmax

Regularization is a technique used to prevent _________________ in a model.

• instability

• underfitting

• convergence

• overfitting

The convolution operation in a CNN extracts features by applying a small filter or kernel to the input data. What is the key advantage of this operation that allows it to reduce the number of parameters?

• Batch normalization

• Stochastic gradient descent

• Max pooling

• Parameter sharing

In a Convolutional Neural Network (CNN), what is the purpose of a pooling layer?

• To learn new features from the input.

• To increase the spatial dimensions of the feature map.

• To apply a non-linear transformation to the feature map.

• To reduce the spatial size of the representation, which reduces the number of parameters and computation in the network.

The 'Equivariance' property of a CNN means that if the input image changes, the output representation changes in a ______________ way.

• similar or predictable

• magnified

• opposite

• random