Online learning control by association and reinforcement

By Jennie Si, Yu-Tsung Wang. 2000

659 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computational Theory and Mathematics, Reinforcement Learning

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computational Theory and Mathematics

In Reinforcement Learning

Citations in the corpus

Adaptive Dynamic Programming for Control: A Survey and Recent Advances (2021, background): “Direct HDP was presented in [57] and the convergence guarantee was provided for this design in [58].”
Sparse Q-learning with Mirror Descent (2012, methodology): “The triple-link inverted pendulum [SW01] is a highly nonlinear time-variant under-actuated system, which is a standard benchmark testbed in the control community.”
Data-Driven Optimal Consensus Control for Discrete-Time Multi-Agent Systems With Unknown Dynamics Using Reinforcement Learning Method (2017, methodology): “Furthermore, RL methods are able to find out the optimal action from unknown environment [9], which motivates the promising applications of RL technique on the control field for real systems.”
Policy Iteration Adaptive Dynamic Programming Algorithm for Discrete-Time Nonlinear Systems (2014, methodology): “Therefore, the gradient-based weight update rule [24] for the critic network is given by”
Safe Exploration Algorithms for Reinforcement Learning Controllers (2018, background): “RL has also proved its worth in combination with neural networks (NNs) in the development of neurodynamic programming [7], [8], where the approximation power of NNs is utilized to efﬁciently represent both the value and the policy, thus reducing the “curse of dimensionality” [9].”
Optimal control of unknown nonaffine nonlinear discrete-time systems based on adaptive dynamic programming (2012, neutral): “…‘‘approximate dynamic programming’’ (Al-Tamimi et al., 2008; Werbos, 1992), ‘‘neuro-dynamic programming’’ (Bertsekas & Tsitsiklis, 1996), ‘‘neural dynamic programming’’ (Si & Wang, 2001), ‘‘adaptive critic designs’’ (Prokhorov & Wunsch, 1997), and ‘‘reinforcement learning’’ (Watkins & Dayan, 1992).”
Adaptive Neural Network Control of AUVs With Control Input Nonlinearities Using Reinforcement Learning (2017, result): “this measure is also similar to the standard Bellman equation [63], [64].”
Energy-Storage-Based Low-Frequency Oscillation Damping Control Using Particle Swarm Optimization and Heuristic Dynamic Programming (2014, methodology): “3) Action Network Training: The procedure of adapting the action network in this architecture is similar to the classic ADP approach to implicity backpropagate the error between the desired ultimate object and the approximate function from the critic network [27].”
Model-free Q-learning designs for discrete-time zero-sum games with application to H-infinity control (2007, background): “See also Bertsekas and Tsitsiklis (1996); He and Jagannathan (2005), Si and Wang (2001) and Cao (2002).”
Adaptive Critic Nonlinear Robust Control: A Survey (2017, background): “Among them, complex industrial systems, such as power systems [30], [44], [67], [84], [129]–[133], [158], [181]–[186], mechanical systems [32], [36], [41], [44], [71], [123], [125], [131], [187], [188], and intelligent transportation systems [189], [190] are the most common application areas.”
Reinforcement learning in feedback control (2011, background): “This basic idea of propagating the gradient of a neural Q-function through the policy network can also be found in earlier approaches (Jordan and Jacobs 1990; Prokhorov and Wunsch 1997; Wang and Si 2001).”
Regularized Off-Policy TD-Learning (2012, background): “The triple-link inverted pendulum [18] is a highly nonlinear under-actuated system with 8dimensional state space and discrete action space.”