-
Adaptive Smoothing Path Integral Control
In Path Integral control problems a representation of an optimally contr...
read it
-
Smoothed functional-based gradient algorithms for off-policy reinforcement learning
We consider the problem of control in an off-policy reinforcement learni...
read it
-
Robust Reinforcement Learning using Least Squares Policy Iteration
This paper addresses the problem of model-free reinforcement learning fo...
read it
-
Logistic Q-Learning
We propose a new reinforcement learning algorithm derived from a regular...
read it
-
Reinforcement Learning via Fenchel-Rockafellar Duality
We review basic concepts of convex duality, focusing on the very general...
read it
-
Optimal Network Control in Partially-Controllable Networks
The effectiveness of many optimal network control algorithms (e.g., Back...
read it
-
On the convergence of cycle detection for navigational reinforcement learning
We consider a reinforcement learning framework where agents have to navi...
read it
Smoothed Dual Embedding Control
We revisit the Bellman optimality equation with Nesterov's smoothing technique and provide a unique saddle-point optimization perspective of the policy optimization problem in reinforcement learning based on Fenchel duality. A new reinforcement learning algorithm, called Smoothed Dual Embedding Control or SDEC, is derived to solve the saddle-point reformulation with arbitrary learnable function approximator. The algorithm bypasses the policy evaluation step in the policy optimization from a principled scheme and is extensible to integrate with multi-step bootstrapping and eligibility traces. We provide a PAC-learning bound on the number of samples needed from one single off-policy sample path, and also characterize the convergence of the algorithm. Finally, we show the algorithm compares favorably to the state-of-the-art baselines on several benchmark control problems.
READ FULL TEXT
Comments
There are no comments yet.