SGD learning on neural networks: leap complexity and saddle-to-saddle dynamics.
Emmanuel AbbeEnric Boix-AdseràTheodor MisiakiewiczPublished in: CoRR (2023)
Keyphrases
- neural network
- learning algorithm
- learning process
- active learning
- neural nets
- online learning
- learning problems
- stochastic gradient descent
- reinforcement learning
- computational complexity
- supervised learning
- inductive inference
- connectionist systems
- recurrent networks
- recurrent neural networks
- back propagation
- knowledge acquisition