Learning action models from plan examples using weighted MAX-SAT

By Qiang Yang, Kangheng Wu, Yunfei Jiang. 2007

264 citations recorded by the source corpus.

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

Citations in the corpus

Learning First-Order Symbolic Representations for Planning from the Structure of the State Space (2019, background): “Object-oriented MDPs [13] and similar work in classical planning [37, 2, 1], build ﬁrst-order model representations but starting with a ﬁrst-order symbolic language, or with information about the actions and their arguments [11].”
Automated Planning and Acting (2016, methodology): “Learning planning operators and domain models from plan examples and solution traces has been addressed with logic-based techniques, for example, in classical planning [617, 594, 627] and HTN planning [626, 283].”
Inventing Relational State and Action Abstractions for Effective and Efficient Bilevel Planning (2022, background): “A separate line of work studies learning symbolic planning models from plan traces [82, 83, 84]; we are also interested in learning planning models, but in our setting, the data and environment are continuous, and our symbolic models are used as abstractions for efﬁcient planning.”
Classical Planning in Deep Latent Space: Bridging the Subsymbolic-Symbolic Boundary (2017, methodology): “However, action model learning from a limited set of action instances is a nontrivial area of active research [Cresswell et al. , 2013; Gregory and Cresswell, 2015; Konidaris et al. , 2014; Mour˜ao et al. , 2012; Yang et al. , 2007; Celorrio et al. , 2012].”
Deliberation for autonomous robots: A survey (2017, methodology): “It has also been used to improve the quality of plans with respect to cost, success rate [143, 204], or user’s preferences [6], and to learn domain models and planning operators [235, 220].”
Unsupervised Grounding of Plannable First-Order Logic Representation from Images (2019, methodology): “Extending or leveraging the existing Action Model Acquisition methods (e.g., AMA 1 , AMA 2 , Action Learner (Amado et al. 2018), (Yang, Wu, and Jiang 2007; Mour˜ao et al. 2012; Cresswell, McCluskey, and West 2013)) is an important avenue for future work.”
A review of learning planning action models (2018, methodology): “The ARMS (Action-Relation Modelling System) (Yang et al. (2007)) system learns action models from traces which consist of action sequences occasionally interleaved with intermediate states.”
Maintaining Evolving Domain Models (2016, background): “Knowledge engineering for planning (Cresswell et al., 2013, Vaquero et al., 2013b, Yang et al., 2007) can be greatly enhanced with systems like Marshal.”
Model Recognition as Planning (2019, background): “While most approaches can learn from empty or partially speciﬁed models (Yang, Wu, and Jiang 2007; Amir and Chang 2008; Cresswell, McCluskey, and West 2013), model recognition requires the ability of editing a possibly inconsistent model M so that the resulting model M (cid:48) is consistent with a…”
Robust planning with incomplete domain models (2017, background): “Similarly, although there exist efforts (Amir and Chang, 2008; Yang et al., 2007; Zhuo and Yang, 2014; Cresswell et al., 2013) that attempt to either learn models from scratch or revise existing ones, their operation is contingent on the availability of successful plan traces or access to execution ”
Learning Features and Abstract Actions for Computing Generalized Plans (2018, methodology): “In this sense, the work is different than action model learning (Yang, Wu, and Jiang 2007) and model-based reinforcement learning and closer in aims to work on learning general policies from examples or experience (Mart´ın and Geffner 2004; Fern, Yoon, and Givan 2006; Sukhbaatar et al. 2015; Zhang…”
Leveraging Pre-trained Large Language Models to Construct and Utilize World Models for Model-based Task Planning (2023, unknown): “In classical planning, the community has explored numerous learning-based methods [59, 61, 9, 25, 4] and interactive editor-based methods [49] for acquiring symbolic domain models.”