Nonlinear Hybrid Planning with deep net ltearned transition models and mixed-integer linear programming

Buser Say, Ga Wu, Yu Qing Zhou, Scott Sanner

Research output: Chapter in Book/Report/Conference proceedingConference PaperResearchpeer-review

25 Citations (Scopus)

Abstract

In many real-world hybrid (mixed discrete continuous) planning problems such as Reservoir Control, Heating, Ventilation and Air Conditioning (HVAC), and Navigation, it is difficult to obtain a model of the complex nonlinear dynamics that govern state evolution. However, the ubiquity of modern sensors allow us to collect large quantities of data from each of these complex systems and build accurate, nonlinear deep network models of their state transitions. But there remains one major problem for the task of control - how can we plan with deep network learned transition models without resorting to Monte Carlo Tree Search and other black-box transition model techniques that ignore model structure and do not easily extend to mixed discrete and continuous domains? In this paper, we make the critical observation that the popular Rectified Linear Unit (ReLU) transfer function for deep networks not only allows accurate nonlinear deep net model learning, but also permits a direct compilation of the deep network transition model to a Mixed-Integer Linear Program (MILP) encoding in a planner we call Hybrid Deep MILP Planning (HDMILP-PLAN). We identify deep net specific optimizations and a simple sparsification method for HD-MILP-PLAN that improve performance over a naïve encoding, and show that we are able to plan optimally with respect to the learned deep network.

Original languageEnglish
Title of host publicationProceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence
EditorsCarles Sierra
Place of PublicationMarina del Rey CA USA
PublisherAssociation for the Advancement of Artificial Intelligence (AAAI)
Pages750-756
Number of pages7
ISBN (Electronic)9780999241103
ISBN (Print)9780999241110
DOIs
Publication statusPublished - 2017
Externally publishedYes
EventInternational Joint Conference on Artificial Intelligence 2017 - Melbourne, Australia
Duration: 19 Aug 201725 Aug 2017
Conference number: 26th
https://ijcai-17.org/
https://www.ijcai.org/Proceedings/2017/ (Proceedings)

Conference

ConferenceInternational Joint Conference on Artificial Intelligence 2017
Abbreviated titleIJCAI 2017
Country/TerritoryAustralia
CityMelbourne
Period19/08/1725/08/17
Internet address

Cite this