• Corpus ID: 57189502

Differentiable Satisfiability and Differentiable Answer Set Programming for Sampling-Based Multi-Model Optimization

@article{Nickles2018DifferentiableSA,
  title={Differentiable Satisfiability and Differentiable Answer Set Programming for Sampling-Based Multi-Model Optimization},
  author={Matthias Nickles},
  journal={ArXiv},
  year={2018},
  volume={abs/1812.11948}
}
We propose Differentiable Satisfiability and Differentiable Answer Set Programming (Differentiable SAT/ASP) for multi-model optimization. Models (answer sets or satisfying truth assignments) are sampled using a novel SAT/ASP solving approach which uses a gradient descent-based branching mechanism. Sampling proceeds until the value of a user-defined multi-model cost function reaches a given threshold. As major use cases for our approach we propose distribution-aware model sampling and expressive… 
1 Citations

Figures from this paper

Proceedings 36th International Conference on Logic Programming (Technical Communications)

  • Matthias Nickles
  • Computer Science
    Electronic Proceedings in Theoretical Computer Science
  • 2020

References

SHOWING 1-10 OF 22 REFERENCES

Differentiable SAT/ASP

This work presents an enhancement of the CDNL/CDCL algorithm as primary implementation approach and introduces alternative algorithms which use an unmodified ASP solver and map the optimization task to conventional answer set optimization or use so-called propagators.

Sampling-Based SAT/ASP Multi-model Optimization as a Framework for Probabilistic Inference

This paper enhances a state-of-the-art SAT/ASP solving algorithm with Gradient Descent as branching literal decision approach, and optionally a cost backtracking mechanism, with multi-model optimization through SAT witness or answer set sampling.

A general stochastic approach to solving problems with hard and soft constraints

This work shows how both hard and soft constraints can be handled by encoding problems as instances of weighted MAX-SAT, and presents experimental results on encodings of a set of well-studied network Steiner-tree problems.

Distribution-Aware Sampling and Weighted Model Counting for SAT

Anovel approach is presented that works with a black-box oracle for weights of assignments and requires only an NP-oracle to solve both the counting and sampling problems when the tilt is small, and scales toproblems involving several thousand variables.

Conflict-Driven Answer Set Solving

A new approach to computing answer sets of logic programs, based on concepts from constraint processing (CSP) and satisfiability checking (SAT), to view inferences in answer set programming (ASP) as unit propagation on no-goods to provide a uniform constraint-based framework for the different kinds of inferences.

ASSAT: computing answer sets of a logic program by SAT solvers

Sound and Efficient Inference with Probabilistic and Deterministic Dependencies

MC-SAT is an inference algorithm that combines ideas from MCMC and satisfiability, based on Markov logic, which defines Markov networks using weighted clauses in first-order logic and greatly outperforms Gibbs sampling and simulated tempering over a broad range of problem sizes and degrees of determinism.

Probability logic and optimization SAT: The PSAT and CPA models

  • D. Pretolani
  • Mathematics, Computer Science
    Annals of Mathematics and Artificial Intelligence
  • 2005
It is proved that Max SAT with unrestricted weights is NP-hard for the class of graph formulas, where Min SAT can be solved in polynomial time, and that PSAT isNP-complete for ideal formulas.

Weight Learning in a Probabilistic Extension of Answer Set Programs

The concept of weight learning in LPMLN and learning algorithms forLPMLN derived from those for Markov Logic are presented and the method to learn the parameters for probabilistic abductive reasoning about actions is applied.

Distribution-Aware Sampling of Answer Sets

Algorithms for the sampling of answer sets under given probabilistic constraints allow for the specification of probability distributions over stable models using probabilistically weighted facts and rules as constraints for an approximate sampling task with specifiable accuracy.