Squashing activation functions in benchmark tests: towards eXplainable Artificial Intelligence using continuous-valued logic.
Daniel ZeltnerBenedikt SchmidGábor CsiszárOrsolya CsiszárPublished in: CoRR (2020)
Keyphrases
- activation function
- benchmark tests
- artificial intelligence
- valued logic
- continuous functions
- neural network
- fuzzy logic
- artificial neural networks
- back propagation
- classical logic
- feed forward
- complex valued
- hidden layer
- propositional logic
- expert systems
- neural nets
- learning rate
- first order logic
- multilayer perceptron
- computational intelligence
- radial basis function
- truth values
- training phase
- machine learning
- basis functions
- logic programs
- network architecture
- multi layer perceptron
- genetic algorithm
- rbf neural network
- real time
- model checking
- fuzzy sets
- knowledge representation