True or False Flashcards
(17 cards)
In a CNN, feature maps remain the same size throughout all layers of the network.
False
RNNs are only used for text-based tasks and cannot be applied to other domains like time-series forecasting or speech recognition.
False
Expert systems are a type of intelligent system designed to simulate human expertise in specific domains, such as medical diagnosis.
True
Neural networks in intelligent systems process information in a way that is structurally similar to the human brain.
True
Using too many convolutional layers in a CNN always results in better performance.
False
Machine Learning (ML) and Deep Learning (DL) are both subsets of Artificial Intelligence (AI), but ML does not require human intervention.
False
The primary purpose of max-pooling in CNNs is to extract the most important features while reducing computation.
True
Batch normalization helps in training deep CNNs by normalizing input activations, reducing internal covariate shift.
True
RNNs are specifically designed to handle sequential data by maintaining memory of previous inputs.
True
Pooling layers in CNNs help in reducing computational complexity but do not impact feature extraction.
False
Gradient-based optimization techniques, such as stochastic gradient descent (SGD), are the only way to train CNNs.
False
A fully connected layer in a CNN is responsible for final classification by converting extracted features into output probabilities.
True
Recurrent Neural Networks (RNNs) and CNNs are fundamentally the same, as both process sequential data.
False
Bias in AI models only occurs due to poor algorithm design, not because of biased training data.
False
A well-designed intelligent system should always follow fixed rules to ensure accuracy and consistency.
False
RNNs are the best choice for all sequence-based tasks and should always be preferred over other architectures like CNNs or Transformers.
False
Using larger kernel sizes in CNNs always leads to better performance in feature extraction.
False
