Faster Gradient-based NAS Pipeline Combining Broad Scalable Architecture with Confident Learning Rate.
Zixiang DingYaran ChenNannan LiDongbin ZhaoPublished in: CoRR (2020)
Keyphrases
- learning rate
- learning algorithm
- convergence rate
- error function
- hidden layer
- pipeline architecture
- memory efficient
- parallel architecture
- rapid convergence
- convergence speed
- training algorithm
- activation function
- convergence theorem
- weight vector
- adaptive learning rate
- multilayer neural networks
- bp neural network algorithm
- delta bar delta
- network architecture
- reinforcement learning
- middle layer
- model free
- fuzzy neural network
- dynamic programming