DeepLearning

Which architecture is best suited for handling sequential data like text or time series?

• CNN
• MLP
• RNN
• GAN

What type of layer in CNNs is primarily responsible for reducing spatial dimensions?

• ReLU Layer
• Convolution Layer
• Max Pooling Layer
• Embedding Layer

What does a softmax activation function output?

• A set of binary values
• A discrete probability distribution
• Normalized text
• Raw logits

Which model includes gates such as input, forget, and output?

• RNN
• CNN
• LSTM
• MLP

What is the primary benefit of the ReLU activation function?

• Non-differentiability
• Saturates quickly
• Solves vanishing gradients
• Slower training

What does the 'Flatten' layer do in a CNN?

• Converts vectors into matrices
• Flattens the sequence order
• Reduces overfitting
• Converts matrix input into 1D array

Which method is used to tokenize text into sequences of integers?

• CNN
• Tokenizer
• Embedding
• One-hot Encoding

What is the main role of the embedding layer in NLP?

• Increase dimensionality
• Convert sequences to text
• Transform words to dense vectors
• Downsample text

Which function is typically used in binary classification loss?

• Mean Squared Error
• Binary Cross-Entropy
• Hinge Loss
• Huber Loss

How does max pooling contribute to CNNs?

• Expands the feature map
• Increases model complexity
• Reduces spatial dimensions
• Removes activation functions