Understanding and mitigating exploding inverses in invertible neural networks.
Jens BehrmannPaul VicolKuan-Chieh WangRoger B. GrosseJörn-Henrik JacobsenPublished in: CoRR (2020)
Keyphrases
- neural network
- pattern recognition
- artificial neural networks
- evolutionary algorithm
- neural network model
- back propagation
- sufficient conditions
- learning algorithm
- website
- training data
- information systems
- computer vision
- real time
- fuzzy logic
- genetic algorithm
- information retrieval
- fuzzy systems
- network architecture
- radon transform
- rule extraction
- deeper understanding
- competitive learning