ADMM Training Algorithms for Residual Networks: Convergence, Complexity and Parallel Training.
Jintao XuYifei LiWenxun XingPublished in: CoRR (2023)
Keyphrases
- computational complexity
- computational cost
- training process
- early stopping
- worst case
- training algorithm
- convergence rate
- lower complexity
- stochastic approximation
- space complexity
- neural network
- training examples
- theoretical analysis
- supervised learning
- depth first search
- linear svm
- significant improvement
- optimal solution
- recurrent networks
- parallel implementations
- learning algorithm