An overview of reinforcement learning techniques

Stochastic calculus-guided reinforcement learning: A probabilistic framework for optimal decision-making

“…Reinforcement Learning is a machine learning paradigm where an agent learns to make decisions through interactions with an environment. The agent takes action, observes the environment's response, and receives rewards or penalties, enabling it to learn a strategy to maximize cumulative rewards over time [ 22 ]. Critical components of RL include states (s), actions (a), policies (π), rewards (r), and value functions.…”

Command filter based event-triggered adaptive control for strict-feedback nonlinear systems with full state constraints

“…The ADP method has the capability to solve a wider variety of real-world practical issues, such as autonomous driving [9], robust control [10], and robot control [11], greatly expanding the application domains of the ADP method. Reinforcement learning (RL) (or ADP) is a technique applied to train machine learning models to take actions in specific scenarios to maximize expected returns [12]. In recent years, utilizing the consistent approximation and adaptive capability of neural networks (NNs) [13], NN-based RL techniques have successfully developed various effective optimal control strategies (such as [14][15][16]).…”