Deep Learning Models Flashcards
What is the primary function of a Convolutional Neural Network (CNN)?
CNNs are designed for processing grid-like data, such as images. They use convolutional layers to automatically extract spatial features and reduce the need for manual feature engineering.
How do RNNs differ from traditional neural networks?
RNNs use the same NN each time with a hidden state the persists to the future. They are using the same NN over and over again allowing to produce results on varying sequences. traditional NNs are forward feed options for prediction
What is the role of Fully Connected layers in deep learning models?
Fully Connected layers connect every neuron from one layer to every neuron in the next, enabling the network to learn complex non-linear relationships.
What makes LSTMs more effective than standard RNNs for long sequences?
LSTMs (Long Short-Term Memory networks) include a gating mechanism that helps retain long-term dependencies while mitigating vanishing gradient issues in RNNs.
What are Transformers commonly used for?
Transformers are used for tasks involving sequential data, such as natural language processing, by using self-attention mechanisms to model dependencies between tokens.
What is the purpose of Autoencoders in deep learning?
Autoencoders are used for unsupervised learning to encode data into a compressed representation and then reconstruct the input, often for dimensionality reduction or anomaly detection.
What distinguishes Reinforcement Learning from supervised learning?
Reinforcement Learning involves an agent learning to make decisions by interacting with an environment and receiving feedback in the form of rewards, rather than relying on labeled data.
What is a Generative Adversarial Network (GAN)?
GANs consist of a generator and a discriminator competing in a zero-sum game, where the generator creates data and the discriminator tries to distinguish it from real data.
How in GraphML information moves from edges to nodes in vice versa?
Using a pooling layer from edges to nodes
How do CNNs handle spatial hierarchies in images?
CNNs use pooling layers to progressively reduce the spatial dimensions, capturing hierarchical features from edges to complex patterns.
What is the main drawback of standard RNNs?
Standard RNNs struggle with long-term dependencies due to vanishing or exploding gradient problems. + Due to sequential nature they can’t be parallelised.
What is the input-output structure of Fully Connected networks?
Fully Connected networks take a fixed-size vector as input and produce a fixed-size vector as output, making them suitable for dense data representations.
What problem do LSTMs address compared to vanilla RNNs?
LSTMs address the issue of vanishing gradients in RNNs by introducing gates that control the flow of information.
What is the role of the self-attention mechanism in Transformers?
Compute relationships between data doesn’t matter their distance in the input.
Use the data to get attention and figure out what to focus on
What is the bottleneck layer in Autoencoders?
The bottleneck layer is the compressed, low-dimensional representation of the input data, capturing its most essential features.
