The Interpolation Phase Transition in Neural Networks: Memorization and Generalization under Lazy Training.
Andrea MontanariYiqiao ZhongPublished in: CoRR (2020)
Keyphrases
- phase transition
- neural network
- training algorithm
- training process
- constraint satisfaction
- np complete
- satisfiability problem
- combinatorial problems
- multi layer perceptron
- random instances
- artificial neural networks
- hard problems
- average degree
- back propagation
- pattern recognition
- randomly generated
- graph coloring
- sat problem
- multilayer perceptron
- relational learning
- genetic algorithm
- np complete problems
- hamiltonian cycle
- constraint satisfaction problems
- random graphs
- sat instances
- hidden layer
- orders of magnitude
- dynamic programming
- special case
- deep structure
- computational complexity
- random constraint satisfaction problems