Differentiable Trust Region Layers for Deep Reinforcement Learning.
Fabian OttoPhilipp BeckerNgo Anh VienHanna Carolin ZiescheGerhard NeumannPublished in: CoRR (2021)
Keyphrases
- variational inequalities
- trust region
- newton method
- reinforcement learning
- function approximation
- objective function
- global optimum
- machine learning
- state space
- column generation
- dynamic programming
- line search
- optimization methods
- loss function
- hessian matrix
- learning problems
- step size
- optimization method
- mean shift
- optimal solution
- log likelihood
- function approximators
- simulated annealing
- evolutionary algorithm