Exploit ILP to learn symmetry breaking constraints of ASP programs.

Last update: Apr 13, 2022

Overview

ILP Symmetry Breaking

Overview

This project aims to exploit inductive logic programming to lift symmetry breaking constraints of ASP programs.

Given an ASP file, we use the system SBASS (symmetry-breaking answer set solving) to infer its graph representation and then detect the symmetries as a graph automorphism problem (performed by the system SAUCY). SBASS returns a set of (irredundant) graph symmetry generators, which are used in our framework to compute the positive and negative examples for the ILP system ILASP.

Note: the files of Active Background Knowledge (active_BK/active_BK_sat) contain the constraints learned for the experiments. To test the framework, remove the constraints and follow the files' instructions to obtain the same result.

Project Structure

.
├── \Experiments              # Directory with experiments results 
│   ├── experiments.csv         # CSV file with results
│   └── experiments             # Script to compare the running-time performance     
│
├── \Instances              # Directory with problem instances
│   ├── \House_Configuration     # House-Configuration Problem     
│   ├── \Pigeon_Owner            # Pigeon-Hole Problem with colors and owners extension   
│   ├── \Pigeon_Color            # Pigeon-Hole Problem with colors extension
│   └── \Pigeon_Hole             # Pigeon-Hole Problem  
│
├── \src                    # Sources  
│   ├── \ILASP4                  # ILASP4 
│   ├── \SBASS                   # SBASS 
│   ├── file_names.py            # Python module with file names
│   ├── parser.py                # Main python file: create the positive and negative examples from SBASS output
│   ├── remove.py                # Auxiliary python file to remove duplicate in smodels file
│   └── permutations.lp          # ASP file which computes the (partial) non symmetric 
│                                  permutations of atoms
│
├── .gitignore 
├── .gitattributes
├── ILP_SBC                 # Script that runs SBASS and lift the SBC found using ILASP
└── README.md

Prerequisites

Usage

1) Create default positive examples

Create the default positive examples for Pigeon_Hole problem: each instance in the directory Gen generate a positive example.

$ .\ILP_SBC -g .\Instances\Pigeon_Hole

2) Create positive and negative examples

Default mode: each non-symmetric answer set defines a positive example

 $ .\ILP_SBC -d .\Instances\Pigeon_Hole

Satisfiable mode: define a single positive example with empty inclusions and exclusions

 $ .\ILP_SBC -s .\Instances\Pigeon_Hole

3) Run ILASP to extend the active background knowledge

 $ .\ILP_SBC -i .\Instances\Pigeon_Hole

Citations

C. Drescher, O. Tifrea, and T. Walsh, “Symmetry-breaking answer set solving” (SBASS)

@article{drescherSymmetrybreakingAnswerSet2011,
	title = {Symmetry-breaking answer set solving},
	volume = {24},
	doi = {10.3233/AIC-2011-0495},
	number = {2},
	journal = {AI Commun.},
	author = {Drescher, Christian and Tifrea, Oana and Walsh, Toby},
	year = {2011},
	pages = {177--194}
}

M. Law, A. Russo, and K. Broda, “The {ILASP} System for Inductive Learning of Answer Set Programs” (ILASP)

@article{larubr20b,
     title = {The {ILASP} System for Inductive Learning of Answer Set Programs},
     author = {M. Law and A. Russo  and K. Broda},
     journal = {The Association for Logic Programming Newsletter},
     year = {2020}
}
@misc{ilasp,
     author = {M. Law and A. Russo  and K. Broda},
     title = {Ilasp Releases},
     howpublished = {\url{www.ilasp.com}},
     note = {Accessed: 2020-10-01},
     year={2020}
}

Exploit ILP to learn symmetry breaking constraints of ASP programs.

Related tags

Overview

ILP Symmetry Breaking

Overview

Project Structure

Prerequisites

Usage

1) Create default positive examples

2) Create positive and negative examples

Default mode: each non-symmetric answer set defines a positive example

Satisfiable mode: define a single positive example with empty inclusions and exclusions

3) Run ILASP to extend the active background knowledge

Citations

Owner

Research Group Production Systems

A PyTorch implementation of QANet.

Datasets for new state-of-the-art challenge in disentanglement learning

An end-to-end regression problem of predicting the price of properties in Bangalore.

Source code for our CVPR 2019 paper - PPGNet: Learning Point-Pair Graph for Line Segment Detection

Causal Influence Detection for Improving Efficiency in Reinforcement Learning

Beyond Image to Depth: Improving Depth Prediction using Echoes (CVPR 2021)

Rethinking Portrait Matting with Privacy Preserving

Official code for the ICLR 2021 paper Neural ODE Processes

A font family with a great monospaced variant for programmers.

Collapse by Conditioning: Training Class-conditional GANs with Limited Data

It helps user to learn Pick-up lines and share if he has a better one

Adversarial Self-Defense for Cycle-Consistent GANs

NIMA: Neural IMage Assessment

This is our ARTS test set, an enriched test set to probe Aspect Robustness of ABSA.

A Python Package for Portfolio Optimization using the Critical Line Algorithm

Code for "Learning the Best Pooling Strategy for Visual Semantic Embedding", CVPR 2021

Evolving neural network parameters in JAX.

Official Repository for our ICCV2021 paper: Continual Learning on Noisy Data Streams via Self-Purified Replay

An Implicit Function Theorem (IFT) optimizer for bi-level optimizations

pyhsmm - library for approximate unsupervised inference in Bayesian Hidden Markov Models (HMMs) and explicit-duration Hidden semi-Markov Models (HSMMs), focusing on the Bayesian Nonparametric extensions, the HDP-HMM and HDP-HSMM, mostly with weak-limit approximations.