Reward-balancing methods normalize RL reward functions to enable greedy optimal policies, reformulated as optimal control with stochastic sampling for uncertainty and shown to improve performance in MPC simulations.
2, 7392–7399
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
math.OC 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
On Reward-Balancing Methods for Reinforcement Learning
Reward-balancing methods normalize RL reward functions to enable greedy optimal policies, reformulated as optimal control with stochastic sampling for uncertainty and shown to improve performance in MPC simulations.