Deep Learning for Computer Vision

What is a Perceptron in the context of neural networks?

• A type of activation function

• A single-layer neural network

• A deep learning framework

• A machine learning algorithm

Which of the following is a key characteristic of the Multi-layer Perceptron (MLP)?

• It has only one layer of neurons

• It can have multiple hidden layers

• It uses linear activation functions

• It is primarily used for regression tasks

Which activation function is commonly used to introduce non-linearity in neural networks?

• Linear activation function

• ReLU (Rectified Linear Unit)

• Sigmoid activation function

• Step function

What is the primary purpose of the feed-forward process in a neural network?

• To calculate the error between predicted and actual outputs

• To update the model's weights and biases

• To propagate gradients during backpropagation

• To make predictions based on input data

Which of the following is an example of an error function commonly used in training neural networks?

• Mean Squared Error (MSE)

• Gradient Descent

• Sigmoid Function

• ReLU Activation

What is the main objective of optimization algorithms in neural network training?

• To minimize the training time

• To maximize the number of hidden layers

• To find the best initial weights and biases

• To minimize the error or loss function

Backpropagation is a technique used for:

• Forward pass in neural networks

• Initializing weights in a neural network

• Calculating gradients and updating weights in a neural network

• Regularizing the weights in a neural network

In classification tasks using a Multi-layer Perceptron (MLP), the output layer typically uses which activation function?

• ReLU

• Sigmoid

• Tanh

• Linear

Which of the following is a common problem associated with overfitting in neural networks?

• High bias

• Low variance

• Generalization

• Poor performance on unseen data

What is the purpose of the softmax activation function in the output layer of a neural network used for multi-class classification?

• It converts the output values into probabilities

• It introduces non-linearity in the output layer

• It accelerates the training process

• It reduces the risk of vanishing gradients