{"description":"Test your understanding of the backpropagation algorithm, including forward and backward passes, error gradients, and weight updates in neural networks.","questions":[{"answer":"To adjust weights and biases to minimize error","number":1,"options":["To generate activation functions","To adjust weights and biases to minimize error","To initialize the neural network structure","To increase data size"],"question":"What is the main purpose of backpropagation in neural networks?"},{"answer":"Forward propagation and backward propagation","number":2,"options":["Training and testing","Prediction and visualization","Forward propagation and backward propagation","Data scaling and normalization"],"question":"Backpropagation algorithm consists of which two major steps?"},{"answer":"Sigmoid","number":3,"options":["ReLU","Sigmoid","Softmax","Tanh"],"question":"What function is most commonly used as an activation in backpropagation?"},{"answer":"It measures the error between the predicted output and the actual target","number":4,"options":["It prevents overfitting","It calculates the distance between input and output layers","It measures the error between the predicted output and the actual target","It selects the best activation function"],"question":"What is the role of the cost or loss function in backpropagation?"},{"answer":"Chain rule of calculus","number":5,"options":["Linear transformation","Chain rule of calculus","Bayes theorem","Matrix inversion"],"question":"Which mathematical technique is used to compute the gradient in backpropagation?"},{"answer":"Rate of change of output with respect to input","number":6,"options":["Loss function values","Input normalization","Rate of change of output with respect to input","Maximum value of input"],"question":"What does the derivative of the sigmoid function help calculate during backpropagation?"},{"answer":"By subtracting the learning rate times the gradient","number":7,"options":["By adding the input values","By applying the derivative of the activation","By subtracting the learning rate times the gradient","By normalizing the inputs"],"question":"How are the weights updated in backpropagation?"},{"answer":"Mean squared error","number":8,"options":["Cross-entropy","Mean squared error","Cosine similarity","Hinge loss"],"question":"What is typically used as the cost function in simple neural networks?"},{"answer":"The partial derivative of the loss function with respect to each weight","number":9,"options":["A graphical curve of error","The learning rate","The partial derivative of the loss function with respect to each weight","The number of input features"],"question":"In backpropagation, what does the term 'gradient' represent?"},{"answer":"It maps any input to a value between 0 and 1 and is differentiable","number":10,"options":["It normalizes the dataset","It generates binary outputs only","It maps any input to a value between 0 and 1 and is differentiable","It helps initialize weights"],"question":"Why is the sigmoid function commonly used in neural networks?"}],"title":"Backpropagation"}