Neural Network Foundations

Certified Deep Learning Specialist (CDLS) · 50 questions

• What is the primary function of a convolutional layer in a CNN?
• In a very deep neural network, what problem does a residual (skip) connection primarily address?
• A CNN has already extracted features through several convolutional and pooling layers. What does the fully connected layer near the end of the network typically do?
• Which statement best describes backpropagation in a neural network?
• What is the main objective behind using a deep neural network rather than a shallow one?
• A neuron computes a weighted sum of its inputs plus a bias term before applying an activation function. What role does the bias play?
• How does a recurrent neural network (RNN) fundamentally differ from a standard feedforward network?
• A machine learning engineer notices that after several convolutional layers, the feature maps have shrunk considerably. Which technique is designed to prevent this shrinkage by preserving the spatial size of the input?
• What is a perceptron in the context of neural networks?
• What core mechanism gives the Transformer architecture its ability to model relationships across an entire input sequence?
• Why is a convolutional neural network generally preferred over a fully connected network for image classification tasks?
• A team training a generative adversarial network (GAN) notices that the generator keeps producing only a handful of nearly identical outputs, no matter how long training continues. What is this well-known GAN training problem called?
• What specific architectural property allows a CNN to efficiently capture local spatial features in an image?
• How does a dense (fully connected) layer differ structurally from a convolutional layer?
• A team is choosing between a CNN and an RNN for two different projects: one involves classifying photographs, and the other involves predicting the next word in a sentence. Which statement correctly matches architecture to task?
• What does the learning rate hyperparameter primarily control during neural network training?
• What is a key distinction between deep learning and traditional machine learning approaches?
• In the context of deep learning frameworks, what is a tensor?
• What is the primary role of a generative model, such as a GAN?
• What is the main objective of using the Adam optimizer during neural network training?
• A model achieves very high accuracy on training data but performs noticeably worse on new, unseen data. Which of the following is a common technique to address this problem?
• What is the purpose of a cost (loss) function in training a neural network?
• In a convolutional layer, what does the stride parameter control?
• What does the process of hyperparameter tuning involve?
• What is a distinguishing property of the tanh (hyperbolic tangent) activation function compared to sigmoid?
• In a generative adversarial network, what is the primary job of the generator network?
• In a convolutional layer, what best describes the kernel (also called a filter)?
• A colleague asks you to explain the difference between a shallow neural network and a deep neural network. What is the key distinction?
• Which of the following is a classic example of a supervised learning task?
• A network's final layer produces one raw score per class for a 10-class image classification problem. What is the purpose of applying softmax to these scores?
• What is a learning-rate schedule in the context of training a neural network?
• A machine learning engineer chooses ReLU as the activation function for the hidden layers of a new network. Which property of ReLU is she relying on?
• In the context of an artificial neural network, what is a neuron?
• A team notices their deep network performs very well on training data but poorly on validation data. They add a dropout layer to help. What does a dropout layer actually do?
• What is the main purpose of an autoencoder?
• Why did the Transformer architecture become so influential for natural language processing tasks?
• A team is building a system that takes an English sentence as input and produces its French translation as output. Which type of model architecture is best suited to this task?
• During training of a multi-class classifier, why does the output layer typically apply the softmax function rather than leaving the raw scores untouched?
• A team splits their labeled data into three groups before training a neural network. What is the main reason they carve out a validation set separate from the training and test sets?
• In the context of training a neural network, what does the term 'epoch' refer to?
• A model's final layer needs to output something that can be interpreted as the probability of belonging to the positive class in a two-class problem. Which activation function is the classic choice for that output neuron, and why?
• A deep network is training unstably, with loss swinging wildly between mini-batches, and convergence is painfully slow. An engineer suggests inserting batch normalization layers between the linear layers and activations. What problem is batch normalization primarily designed to address?
• A small startup has only 2,000 labeled medical images, nowhere near enough to train a deep CNN from scratch. Instead, they start from a network already trained on millions of general-purpose photos, then fine-tune its later layers on their small medical dataset. What is this strategy called?
• A team is choosing an architecture to classify tens of thousands of product photos into categories like 'shoes,' 'bags,' and 'watches.' Which type of network is generally the best default choice for this kind of image classification task?
• In an NLP model, raw text is first tokenized into integer IDs and then passed through an embedding layer before reaching the rest of the network. What role does that embedding layer serve?
• What key architectural characteristic sets a Recurrent Neural Network (RNN) apart from a standard feedforward network when processing sequential data like sentences or time series?
• Before training even begins, an engineer must decide on values like the learning rate, the number of layers, and the batch size. What is the general term for these kinds of values?
• Standard RNNs struggle to learn dependencies across long sequences because gradients shrink to near zero as they're propagated back through many time steps. Which component of an LSTM (Long Short-Term Memory) unit is primarily responsible for mitigating this vanishing gradient problem?
• Inside a convolutional layer, a learned filter slides across the input image and produces a 2D grid of output values. What is this resulting grid of values called?
• During the training loop, after the network makes a prediction for a batch of examples, something has to quantify how wrong that prediction was so the optimizer knows which direction to adjust the weights. What is the name of that component, and what role does it play?