The Geometry of Neural Nets' Parameter Spaces Under Reparametrization.
Agustinus KristiadiFelix DangelPhilipp HennigPublished in: NeurIPS (2023)
Keyphrases
- neural nets
- parameter space
- feed forward
- high dimensional
- neural network
- starting points
- multi layer
- back propagation
- artificial neural networks
- backpropagation neural networks
- parameter values
- three dimensional
- learning tasks
- single layer
- search space
- artificial intelligence
- real time
- counter propagation
- riemannian manifolds
- starting point
- image processing
- knowledge base
- real world