Activation Landscapes as a Topological Summary of Neural Network Performance.
Matthew WheelerJose BouzaPeter BubenikPublished in: CoRR (2021)
Keyphrases
- neural network
- back propagation
- artificial neural networks
- network architecture
- fuzzy artmap
- fault diagnosis
- information processing
- data sets
- multi layer
- neural network model
- neural network is trained
- activation function
- auto associative
- topological spaces
- text summarization
- image reconstruction from projections
- hidden layer
- hybrid neural network
- self organizing maps
- pattern recognition
- reinforcement learning
- genetic algorithm
- neural nets
- network model
- feed forward
- connected components
- fuzzy neural network
- medical images
- feed forward neural networks
- competitive learning
- expert systems
- activation patterns
- search engine
- real time