Morphology dictates learnability in neural controllers.
Joshua P. PowersRyan GrindleSam KriegmanLapo FratiNick CheneyJosh C. BongardPublished in: ALIFE (2020)
Keyphrases
- evolutionary robotics
- network architecture
- neural network
- finite automata
- neural model
- bio inspired
- control system
- mathematical morphology
- boolean functions
- vapnik chervonenkis dimension
- pattern languages
- nonlinear predictive control
- neural computation
- neural fuzzy
- morphological operators
- uniform distribution
- image processing
- structuring elements
- agnostic learning
- hebbian learning
- reinforcement learning
- regular expressions
- control strategy
- inductive logic programming
- neural network model
- real robot
- image analysis
- dnf formulas
- learning algorithm
- machine learning
- linear separability
- data sets