ART: Abstraction Refinement-Guided Training for Provably Correct Neural Networks.
Xuankang LinHe ZhuRoopsha SamantaSuresh JagannathanPublished in: CoRR (2019)
Keyphrases
- provably correct
- neural network
- training algorithm
- training process
- adaptive resonance theory
- formal methods
- feed forward neural networks
- situation calculus
- pattern recognition
- backpropagation algorithm
- back propagation
- neural network model
- artificial neural networks
- feedforward neural networks
- training set
- artificial intelligence
- recurrent networks
- error back propagation
- neural network training
- multi layer perceptron
- multi layer
- feed forward
- radial basis function
- training examples
- training samples
- supervised learning
- error estimates
- fuzzy logic