Learning Path
Question & Answer
Choose the Best Answer
By using a single activation function for all nodes
By selectively forgetting irrelevant information and maintaining important features
By increasing the number of hidden layers in the network
By processing all data at once without any temporal consideration
Understanding the Answer
Let's break down why this is correct
GRUs use update and reset gates that decide which past information to keep and which to discard. Other options are incorrect because Using only one activation function does not control memory flow; Adding more hidden layers increases capacity but does not give the network a way to choose what to keep.
Key Concepts
Recurrent Neural Networks (RNN)
medium level question
understand
Deep Dive: Recurrent Neural Networks (RNN)
Master the fundamentals
Definition
Recurrent neural networks, including LSTM and gated recurrent networks, have been widely used for sequence modeling and transduction tasks. These networks factor computation along symbol positions and generate hidden states sequentially, limiting parallelization and efficiency.
Topic Definition
Recurrent neural networks, including LSTM and gated recurrent networks, have been widely used for sequence modeling and transduction tasks. These networks factor computation along symbol positions and generate hidden states sequentially, limiting parallelization and efficiency.
Ready to Master More Topics?
Join thousands of students using Seekh's interactive learning platform to excel in their studies with personalized practice and detailed explanations.