Learning to Control an Unstable System with Forward Modeling

By Michael I. Jordan, Robert A. Jacobs. 1989

Subfields: Reinforcement Learning, Robotics & Control

How researchers see this paper

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

In Reinforcement Learning

In Robotics & Control

Citations in the corpus

Continual learning in reinforcement environments (1995, methodology): “This technique has been used both for supervised learning tasks (training the controller to produce speci c results in the environment) [44, 46, 69], and for reinforcementlearning tasks (training the controller to maximize the predicted reinforcement) [45, 68, 91, 93, 116].”
Making the world differentiable: on using self supervised fully recurrent neural networks for dynamic reinforcement learning and planning in non-stationary environments (1990, result): “In this sense the approach resembles the approaches of Watkins [40] and Jordan and Jacobs [9].”
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).”
Handbook of Intelligent Control: Neural, Fuzzy, and Adaptive Approaches (1992, background): “So far, there have only been two true neural net implementations of ADHDP, implementations in which both the Critic and the Action network are neural networks [10,11].”
Locally Weighted Learning for Control (1997, methodology): “Nonlinear parametric models can be trained by 1) exposing the model to a new data point only once (e.g., (Jordan and Jacobs, 1990; Kuperstein, 1988)), or 2) by storing the data in a database and cycling through the training data repeatedly.”
A comparison of direct and model-based reinforcement learning (1997, background): “A number of researchers have focused on this case (Gullapalli, 1990; Jordan and Jacobs, 1990; Williams, 1990; Millington, 1991; Baird and Klopf, 1993; Bradtke, 1993; Bradtke et al., 1994; Doya, 1996).”
Forward Models: Supervised Learning with a Distal Teacher (1992, methodology): “Jordan and Jacobs (1990) illustrated this approach by using distal learning to solve the problem of learning to balance an inverted pendulum on a moving cart.”
Learning to Act Using Real-Time Dynamic Programming (1995, background): “, Jacobson and Mayne [31], Jordan and Jacobs [33], Werbos [83; 88], White and Jordan [89]).”
Adaptive Synchronization of Neural and Physical Oscillators (1991, background): “Back-propagation through forward models of physical systems can also be applied to the learning of sensory feedback (Jordan and Jacobs, 1990).”
Networks adjusting networks (1990, result): “Since MAC does not evaluate just a state but a state/action pair, it is similar to the approaches described in [16] and [5].”
Continuous-in-Depth Neural Networks (2020, result): “This form also resembles a general class of temporal difference models which predate the interpretations of ODE-Nets or residual networks [72, 38].”