Dense neural networks as sparse graphs and the lightning initialization.
Thomas PircherDominik HaspelEberhard SchlückerPublished in: CoRR (2018)
Keyphrases
- neural network
- densely connected
- dense stereo
- pattern recognition
- dense optical flow
- graph matching
- motion field estimation
- graph mining
- artificial neural networks
- neural network model
- high dimensional
- directed acyclic
- fault diagnosis
- fuzzy logic
- multilayer perceptron
- gaussian graphical models
- graph theoretic
- graph representation
- genetic algorithm
- neural nets
- feed forward
- sparse coding
- dense motion estimation
- k means
- quasi cliques
- dense sampling
- graph structures
- sparse data
- recurrent neural networks
- fuzzy systems
- subgraph isomorphism
- weighted graph
- self organizing maps
- sparse representation
- associative memory
- back propagation
- graph model