Weighted model integration (WMI) is a recent formalism generalizing weighted model counting (WMC) to run probabilistic inference over hybrid domains, characterized by both discrete and continuous variables and relationships between them. WMI is computationally very demanding as it requires to explicitly enumerate all possible truth assignments to be integrated over. Component caching strategies which proved extremely effective for WMC are difficult to apply in this formalism because of the tight coupling induced by the arithmetic constraints. In this paper we present a novel formulation of WMI, which allows to exploit the power of SMT-based predicate abstraction techniques in designing efficient inference procedures. A novel algorithm combines a strong reduction in the number of models to be integrated over with their efficient enumeration. Experimental results on synthetic and real-world data show drastic computational improvements over the original WMI formulation as well as existing alternatives for hybrid inference. © 2019 Elsevier B.V. All rights reserved.

Advanced SMT techniques for weighted model integration / Morettin, Paolo; Passerini, Andrea; Sebastiani, Roberto. - In: ARTIFICIAL INTELLIGENCE. - ISSN 0004-3702. - 2019, 275:(2019), pp. 1-27. [10.1016/j.artint.2019.04.003]

Advanced SMT techniques for weighted model integration

Morettin, Paolo;Passerini, Andrea;Sebastiani, Roberto
2019-01-01

Abstract

Weighted model integration (WMI) is a recent formalism generalizing weighted model counting (WMC) to run probabilistic inference over hybrid domains, characterized by both discrete and continuous variables and relationships between them. WMI is computationally very demanding as it requires to explicitly enumerate all possible truth assignments to be integrated over. Component caching strategies which proved extremely effective for WMC are difficult to apply in this formalism because of the tight coupling induced by the arithmetic constraints. In this paper we present a novel formulation of WMI, which allows to exploit the power of SMT-based predicate abstraction techniques in designing efficient inference procedures. A novel algorithm combines a strong reduction in the number of models to be integrated over with their efficient enumeration. Experimental results on synthetic and real-world data show drastic computational improvements over the original WMI formulation as well as existing alternatives for hybrid inference. © 2019 Elsevier B.V. All rights reserved.
2019
Morettin, Paolo; Passerini, Andrea; Sebastiani, Roberto
Advanced SMT techniques for weighted model integration / Morettin, Paolo; Passerini, Andrea; Sebastiani, Roberto. - In: ARTIFICIAL INTELLIGENCE. - ISSN 0004-3702. - 2019, 275:(2019), pp. 1-27. [10.1016/j.artint.2019.04.003]
File in questo prodotto:
File Dimensione Formato  
AIJ19.pdf

Solo gestori archivio

Descrizione: articolo principale
Tipologia: Versione editoriale (Publisher’s layout)
Licenza: Tutti i diritti riservati (All rights reserved)
Dimensione 1.3 MB
Formato Adobe PDF
1.3 MB Adobe PDF   Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11572/234592
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus 17
  • ???jsp.display-item.citation.isi??? 7
social impact