Artificial Neural Networks QUIZ (MCQ QUESTIONS AND ANSWERS)

The number of neurons in the input layer

The number of layers in the neural network

One complete pass through the entire training dataset during training

The number of forward and backward passes in a single training iteration

To add noise to the input data

To calculate gradients

To adjust the learning rate during training

To accelerate convergence by accumulating past gradient updates

Momentum

Batch normalization

Learning rate annealing

AdaGrad

To calculate the model's accuracy

To compute the mean squared error between predictions and actual values

To measure the difference between predicted class probabilities and true class labels

To initialize the weights of neurons

The model's weights are not being updated during training.

The model performs well on the training data but poorly on unseen data.

The model's loss function is not decreasing during training.

The model is too simple to capture the underlying patterns in the data.

To initialize the model's weights

To reduce the learning rate

To calculate gradients

To optimize the model's weights efficiently

Linear regression

Random forest

Gradient descent

K-means clustering

To increase the number of weights in the network

To set all weights to zero

To avoid convergence during training

To provide suitable initial values for weights

Convolutional layer

Recurrent layer

Fully connected layer

Pooling layer

Forward propagation

Hyperparameter tuning

Backward propagation

Model inference

To increase the complexity of the network

To normalize the input data

To accelerate training convergence and reduce internal covariate shift

To reduce the number of layers in the network

To increase the number of neurons in each layer

To add noise to the input data

To reduce overfitting by randomly deactivating neurons during training

To speed up training by skipping certain layers

It introduces non-linearity into the network.

It computes the gradient of the loss function.

It normalizes input data.

It initializes the weights of neurons.

Recurrent Neural Network (RNN)

Radial Basis Function Neural Network (RBFNN)

Convolutional Neural Network (CNN)

Feedforward Neural Network (FNN)

FNNs are deeper than RNNs.

FNNs can handle sequential data, while RNNs cannot.

FNNs have feedback connections, while RNNs do not.

FNNs do not have hidden layers.

To initialize the weights of the neurons

To compute the gradient during backpropagation

To introduce non-linearity into the model

To determine the learning rate

To control the batch size during training

To specify the number of training epochs

To adjust the step size for weight updates during optimization

To set the initial weights of the neurons

Gradient clipping

Learning rate annealing

Weight initialization

Batch normalization

To introduce non-linearity

To convert the network's outputs into probability scores for each class

To reduce the impact of outliers in the data

To increase the network's capacity

Convolutional layer

Pooling layer

Recurrent layer

Fully connected layer

It occurs when the model's loss function becomes too large.

It happens when gradients become too small during backpropagation, causing slow or stalled learning in deep networks.

It refers to the rapid increase in gradients, leading to instability during training.

It is a synonym for gradient descent convergence.

To increase the complexity of the network

To reduce overfitting by randomly deactivating neurons during training

To speed up training by skipping certain layers

To add noise to the input data

Forward propagation

Backward propagation (Backpropagation)

Gradient descent

Weight initialization

To calculate the gradient of the loss function

To initialize the weights of neurons

To introduce non-linearity into the network

To determine the learning rate

Convolutional Neural Network (CNN)

Recurrent Neural Network (RNN)

Feedforward Neural Network (FNN)

Radial Basis Function Neural Network (RBFNN)

To determine the network architecture

To measure the difference between predictions and actual values

To initialize the weights of neurons

To perform forward propagation

They reduce the number of parameters in the network

They perform element-wise matrix multiplication

They enable the network to learn spatial hierarchies of features

They replace the need for fully connected layers

The process of updating weights during training

The forward pass through the network

The selection of the activation function

The process of normalizing input data

Increasing the learning rate

Reducing the number of neurons in each layer

Adding more hidden layers

Using dropout layers during training

Performs mathematical operations on input data

Passes the input data to the output layer

Contains the neurons responsible for making predictions

Normalizes the input data