An Optimal Control Approach to Deep Learning and Applications to Discrete-Weight Neural Networks.
Qianxiao LiShuji HaoPublished in: ICML (2018)
Keyphrases
- optimal control
- deep learning
- neural network
- global exponential stability
- control problems
- dynamic programming
- unsupervised learning
- control strategy
- unsupervised feature learning
- machine learning
- reinforcement learning
- pattern recognition
- control law
- lyapunov function
- mental models
- neural nets
- optimal control problems
- lyapunov stability
- weakly supervised
- hopfield neural network
- linear quadratic
- activation function
- data mining