Neural Networks CIE AS2

Artificial Neural Networks are inspired by the structure and function of:

• The human brain

• Computer processors

• Complex algorithms

• Random number generators

The process of adjusting the weights in a neural network to improve its performance is called:

• Hidden layer activation

• Backpropagation

• Network convergence

• Input normalization

The role of weights in a neural network is to:

• Determine the number of hidden layers

• Control the flow of information between neurons

• Define the activation function

• Store the training data

Hidden layers in a neural network allow the model to:

• Increase the speed of processing

• Learn complex relationships in the data

• Reduce the number of training iterations

• Simplify the output representation

The input layer of a neural network receives:

• The final prediction of the model

• The pre-processed data for the task

• The error signal during backpropagation

• The weights of the network connections

The output layer of a neural network provides:

• The raw training data

• The calculated errors in the network

• The final classification or prediction

• The weights used for hidden layer neurons

A neural network with high accuracy on the training data but poor performance on unseen data is likely suffering from:

• Underfitting

• Overfitting

• Insufficient training iterations

• Incorrect activation function choice

Techniques to improve the accuracy of a neural network can include:

• Increasing the learning rate

• Using a simpler network architecture

• Applying data augmentation techniques

• Removing dropout layers

During backpropagation, the error signal is propagated backwards through the network to:

• Fix the weights of the output layer neurons

• Update the weights of all neurons

• Calculate the activation function

• Generate new training data

Training a neural network requires a large amount of:

• Processing power

• Network bandwidth

• Pre-processed data

• User interaction