A Deep Neural Network's Loss Surface Contains Every Low-dimensional Pattern.
Wojciech Marian CzarneckiSimon OsinderoRazvan PascanuMax JaderbergPublished in: CoRR (2019)
Keyphrases
- low dimensional
- neural network
- high dimensional
- manifold learning
- dimensionality reduction
- pattern matching
- artificial neural networks
- back propagation
- associative memory
- input space
- d objects
- three dimensional
- surface reconstruction
- feed forward neural networks
- high dimensional data
- euclidean space
- neural network model
- fuzzy logic
- nonlinear dimensionality reduction
- feature space
- neural nets
- genetic algorithm
- object surface
- multi layer
- neural network is trained
- pattern discovery
- feed forward
- self organizing maps
- fault diagnosis
- principal component analysis
- data points
- pattern recognition
- hidden layer
- network architecture
- knn