{"description":"Evaluate your understanding of multi-layer neural networks, forward and backpropagation, and their role in modeling complex patterns.","questions":[{"answer":"It consists of multiple layers of neurons, not just one","number":1,"options":["It uses no activation function","It contains only a single neuron","It consists of multiple layers of neurons, not just one","It does not use weights"],"question":"What distinguishes a multi-layer neural network from a perceptron?"},{"answer":"To introduce non-linearity into the model","number":2,"options":["To scale input values","To introduce non-linearity into the model","To measure prediction accuracy","To normalize gradients"],"question":"What is the main purpose of the activation function in a neural network?"},{"answer":"Weighted sum of inputs and bias","number":3,"options":["Gradient","Error","Weighted sum of inputs and bias","Prediction accuracy"],"question":"In forward propagation, what is calculated first for each neuron?"},{"answer":"Backpropagation","number":4,"options":["Normalization","Cross-validation","Backpropagation","Input encoding"],"question":"Which process allows a multi-layer network to learn from its errors?"},{"answer":"It compares predictions to actual outputs to compute error","number":5,"options":["It creates training batches","It compares predictions to actual outputs to compute error","It initializes neuron weights","It determines the output activation"],"question":"How is the loss function used in backpropagation?"},{"answer":"Chain rule of calculus","number":6,"options":["Gradient descent rule","Sigmoid rule","Chain rule of calculus","Forward propagation rule"],"question":"Which rule is used to compute gradients across multiple layers during backpropagation?"},{"answer":"Applying an output activation function","number":7,"options":["Output error computation","Activation of hidden layers","Applying an output activation function","Calculating weights"],"question":"What is the final step of the forward propagation process?"},{"answer":"They control how inputs are transformed and adjusted","number":8,"options":["They determine the batch size","They measure activation output","They control how inputs are transformed and adjusted","They are used for visualization only"],"question":"What role do weights and biases play in neural networks?"},{"answer":"It highlights the need for a non-linear model with hidden layers","number":9,"options":["It is a linear problem easily solved with a perceptron","It highlights the need for a non-linear model with hidden layers","It uses multiple activation functions","It trains faster than other problems"],"question":"Why is the XOR problem significant in neural network training?"},{"answer":"Gradients for weights in all layers","number":10,"options":["Loss values across neurons","Learning rate adjustments","Gradients for weights in all layers","Normalization factors"],"question":"What does the chain rule help compute in backpropagation?"}],"title":"Multi-layer Neural Network"}