A Theoretical Analysis on Feature Learning in Neural Networks: Emergence from Inputs and Advantage over Fixed Features.
Zhenmei ShiJunyi WeiYingyu LiangPublished in: ICLR (2022)
Keyphrases
- theoretical analysis
- neural network
- feature set
- image features
- learning algorithm
- learning process
- prior knowledge
- feature values
- feature vectors
- low level
- selecting features
- numerical simulations
- feature interactions
- significant features
- irrelevant features
- feature maps
- invariant features
- training process
- learning systems
- feature extraction
- unsupervised learning
- supervised learning
- multiple features
- knowledge acquisition
- learning capabilities
- additional features
- reinforcement learning
- binary features
- support vector
- feature relevance
- pattern recognition