Locally Weighted Learning for Control

By Christopher G. Atkeson, Andrew W. Moore, Stefan Schaal. 1997

How researchers see this paper

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

Citations in the corpus

Reinforcement learning in robotics: A survey (2013, methodology): “In fact, many of the methods thus far demonstrated on difficult problems have been modelbased (Atkeson et al., 1997; Abbeel et al., 2007; Deisenroth and Rasmussen, 2011) and robot learning systems often employ policy search methods rather than value functionbased approaches (Gullapalli et al.”
Few-Shot Image Recognition by Predicting Parameters from Activations (2017, background): “But this requires careful designs of the distance metrics [1], which can be difficult and sometimes empirical.”
An Algorithmic Perspective on Imitation Learning (2018, background): “Actionstate space abstraction Chambers and Michie [1969], Widrow and Smith [1964], Pomerleau [1988], Ross and Bagnell [2010], Ross et al. [2011], Boularias et al. [2011], Kalakrishnan et al. [2013], Ho and Ermon [2016] Abbeel and Ng [2004], Ratliff et al.”
Making AI Forget You: Data Deletion in Machine Learning (2019, methodology): “Laziness Lazy learning methods delay computation until inference time [88, 11, 6], resulting in trivial deletions.”
Learning Multimodal Transition Dynamics for Model-Based Reinforcement Learning (2017, methodology): “Most work in model-based RL has either focussed on tabular learning [3], linear function approximation [1] with uni-modal Gaussian noise [20], or other deterministic function approximators like random forests [14].”
Autonomous Helicopter Aerobatics through Apprenticeship Learning (2010, methodology): “To construct an accurate non-linear model to predict ztþ1 from zt, using the aligned data, one could use locally weighted linear regression (Atkeson et al. 1997), where a linear model is learned based on a weighted dataset.”
A Review and Empirical Evaluation of Feature Weighting Methods for a Class of Lazy Learning Algorithms (1997, methodology): “A diierent distance function could be applied at each query (query-based) (Atkeson et al., 1996a) or each instance (point-based) (Aha & Goldstone, 1992).”
Robot Learning System Based on Adaptive Neural Control and Dynamic Movement Primitives (2019, methodology): “With this formulation, the LWR can be employed to learn the model.”
Learning and Using Models (2012, methodology): “One of the earliest model-based RL methods used locally weighted regression (LWR) to learn models (Schaal and Atkeson, 1994; Atkeson et al, 1997).”
Manifold Embeddings for Model-Based Reinforcement Learning under Partial Observability (2009, background): “Impressive performance and scalability of local modelbased approaches [1, 2] and global model-free approaches [6, 17] have been achieved by exploiting the locality of dynamics in fully observable state-space representations of challenging real-world problems.”
Local and Piecewise Affine Approaches to System Identification (2003, methodology): “The local modelling approach and similar ideas have occurred in many contexts [18, 47, 61, 122, 141, 155] under names such as Model on Demand [146], lazy learning and least commitment learning [5, 6, 17].”
Robust Supervised Learning (2005, background): “(Moore et al. 1995) These models are intuitively appealing because local models do not experience “catastrophic interference” where learning in one part of an input space adversely affects things already learned in another part of the space.”
A hybrid machine learning approach to network anomaly detection (2007, methodology): “Cross validation tests were performed using a 3-fold cross validation method on 3000 normal packets divided into three subsets, with the holdout method [4] repeated three times.”
Learning robot motion control with demonstration and advice-operators (2008, methodology): “Our specific implementation employs a form of Locally Weighted Learning [2].”
Task-Specific Generalization of Discrete and Periodic Dynamic Movement Primitives (2010, unknown): “sufficiently efficient for online control [33].”
Variational Bayesian learning of nonlinear hidden state-space models for model predictive control (2009, methodology): “Dynamic programming can be used to solve complex control problems in two different ways: Atkeson et al. [4] [5] used dynamic programming in the state space by discretising the state space into cells and solving locally optimal trajectories for each cell.”
Improved empirical methods in reinforcement-learning evaluation (2015, methodology): “better underlying function approximators (Atkeson et al., 1997).”