Learning Hierarchical Task Networks (HTNs)

This project aims at developing an unified architecture for learning and reusing Hierarchical Task Networks (HTNs): learning of Hierarchical Task Network (HTNs), Applying learned HTNs for problem-solving and an educational component where secondary-school students experience knowledge engineering activities (e.g., to encode HTNs).

Activities:

The IJCAI-09 Workshop on Learning Structural Knowledge From Observations Struck09, organized in part as a result of the disemination activities for this project.
A summer 2007 Invited talk at National Taiwan University about HTNs, Learning and Computer games. Variants of this talk were given also at National Tsing Hua University and at SET Corporation. The latter took place in the summer 2008.
A summer 2008 Invited talk at Hong Kong University of Science and Technology about our work on learning HTNs.
We built a prototype of a Hierarchical Task Network Learner, called HTN-MAKER. HTN-Maker is an offline and incremental algorithm for learning the structural relations between tasks in a Hierarchical Task Network (HTN). HTN-MAKER receives as input a STRIPS domain model, a collection of STRIPS plans, and a collection of task definitions, and produces an HTN domain model. This work was presented at AAAI-08.
- We are studying ways to augment partial HTNs with episodic knowledge (i.e., cases).
We are studying ways to enhance Hierarchical Task Network Learners, all of which currently learn domains for deterministic domains, to non-deterministic domains. We developed techniques to make such an enhancement and used on HTN-Maker to obtain the first HTN learner in non-determinisitc domaisn, which we call ND-HTNMaker. This work was presented at IJCAI-09.
We are currently enhancing experiments on RepairSHOP. RepairSHOP is a plan adaptation algorithm build on top of the HTN planner SHOP, which implements a variant of HTN planning called Ordered Task Decomposition. In this variant tasks are totally ordered and conditions are evaluated relative to the current state of the world, which is updated during planning.
We are exploring using reinforcement learning techniques to learn HTNs in domains under uncertainty.
We implemented an instructional game aimed K-12 students that provides a hands-on experience on aspects of building a knowledge base (e.g., HTN's methods).
A course on Automated Planning developed under this project.

Publications:

Learning Hierarchical Task Networks for Nondeterministic Planning Domains.

PDF

Hankz Hankui Zhuo, Derek Hao Hu, Chad Hogg Qiang Yang, and Hector Munoz-Avila (2009) Learning HTN Method Preconditions and Action Models from Partial Observations. To appear in Proceedings of the Twenty-first International Joint Conference on Artificial Intelligence (IJCAI-09). AAAI Press.
PDF

Hankz Hankui Zhuo, Derek Hao Hu, Qiang Yang, Chad Hogg and Hector Munoz-Avila (2009) Learning Model Structures in AI Planning from Partial Observations. To appear in Proceedings of the IJCAI-09 Workshop on Learning Structural Information from Traces (STRUCK-09). AAAI Press.
PDF

Chad Hogg, Ugur Kuter, and Hector Munoz-Avila (2009) From Plan Traces to Hierarchical Task Networks Using Reinforcements: A Preliminary Report. To appear in Proceedings of the IJCAI-09 Workshop on Learning Structural Information from Traces (STRUCK-09). AAAI Press.
PDF

Munoz-Avila, Cox, M.T. (2008) Case-Based Plan Adaptation: An Analysis and Review. IEEE Intelligent Systems. IEEE inc.
PDF file

Hogg, Chad, Munoz-Avila, Hector, and Ugur Kuter. (2008) HTN-MAKER: Learning HTNs with Minimal Additional Knowledge Engineering Required. To appear in Proceedings of the Twenty-Third AAAI Conference on Artificial Intelligence (AAAI-08). AAAI Press.
PDF

Hogg, Chad, Lee-Urban, Stephen, Auslander, Bryan, and Munoz-Avila, Hector. (2008) Discovering Feature Weights for Feature-Based Indexing of Q-Tables. To appear in Proceedings of the Uncertainty and Knowledge Discovery in CBR Workshop at the 9th European Conference on Case-Based Reasoning (ECCBR-08).
PDF

Auslander, Bryan, Lee-Urban, Stephen, Hogg, Chad, and Munoz-Avila, Hector. (2008) Recognizing The Enemy: Combining Reinforcement Learning with Strategy Selection using Case-Based Reasoning. To appear in Proceedings of the 9th European Conference on Case-Based Reasoning (ECCBR-08).
PDF

Kuchibatla, V., and Muñoz-Avila, H. (2008) An Analysis on Transformational Analogy: General Framework and Complexity To appear in Proceedings of the Twenty-Third AAAI Conference on Artificial Intelligence (AAAI-08). Nectar track. AAAI Press.
PDF

Hogg, C. & Munoz-Avila, H. (2007) Learning of Tasks Models for HTN Planning. Proceedings of the ICAPS-07 Workshop on AI Planning and Learning (AIPL). AAAI Press.
PDF

Vasta, M., Lee-Urban S. & Munoz-Avila, H. (2007) RETALIATE: Learning Winning Policies in First-Person Shooter Games. Proceedings of the Seventeenth Innovative Applications of Artificial Intelligence Conference (IAAI-07). AAAI Press.
PDF

Sanchez Ruiz-Granados, A., Lee-Urban, S. & Munoz-Avila, H., Gonzalez Calero, P. A., Diaz Agudo, B. (2007) Game AI for a Turn-based Strategy Game with Plan Adaptation and Ontology-based retrieval. Proceedings of the ICAPS-07 Workshop on ICAPS 2007 Workshop on Planning in Games. AAAI Press.
PDF

Lee-Urban, S., Parker, A., Kuter, U., Munoz-Avila, H., & Nau, D. (2007) Transfer Learning of Hierarchical Task-Network Planning Methods in a Real-Time Strategy Game. Proceedings of the ICAPS-07 Workshop on ICAPS 2007 Workshop on Planning and Learning (AIPL). AAAI Press.
PDF

Warfield, I., Hogg, C., Lee-Urban, S., Munoz-Avila, H. (2007) Adaptation of Hierarchical Task Network Plans. Proceedings of the Twentieth Flairs International Conference (FLAIRS-07). AAAI Press.
PDF

Contact author: Hector Munoz-Avila

This material is based upon work supported by the National Science Foundation under Grant No. NSF 0642882. Any opinions, findings and conclusions or recomendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the National Science Foundation (NSF)

Last updated: Thur. Nov. 20 11:49:51 EST 2008