End-to-End Learning of LTLf Formulae by Faithful LTLf Encoding
Authors: Hai Wan, Pingjia Liang, Jianfeng Du, Weilin Luo, Rongzhen Ye, Bo Peng
AAAI 2024 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Experimental results demonstrate that our approach achieves state-of-the-art performance with up to 7% improvement in accuracy, highlighting the beneļ¬ts of introducing the faithful LTLf encoding. |
| Researcher Affiliation | Collaboration | 1School of Computer Science and Engineering, Sun Yat-sen University, Guangzhou 510006, P.R.China 2Guangdong University of Foreign Studies, Guangzhou 510006, P.R.China 3Bigmath Technology, Shenzhen 518063, P.R.China |
| Pseudocode | Yes | Algorithm 1: Interpreting LTLf Formula |
| Open Source Code | Yes | All the proofs of lemmas/theorems are provided in the technical report available at https://github.com/a79461378945/TLTLf.git. |
| Open Datasets | Yes | We reused the datasets that are provided by (Luo et al. 2022). |
| Dataset Splits | No | For each dataset, there is a formula with kf operators, and there are 250/250 positive/negative traces for this formula constituting the training set and 500/500 positive/negative traces for this formula constituting the test set. The paper specifies training and test sets but does not mention a separate validation set. |
| Hardware Specification | Yes | All experiments were conducted on a Linux system equipped with an Intel(R) Xeon(R) Gold 6248R processor with 3.0 GHz and 126 GB RAM. |
| Software Dependencies | No | The paper mentions using 'Adam (Kingma and Ba 2015) to optimize the parameters in our model' but does not specify version numbers for Adam or any other software or libraries used in the implementation. |
| Experiment Setup | Yes | Settings. All experiments were conducted on a Linux system equipped with an Intel(R) Xeon(R) Gold 6248R processor with 3.0 GHz and 126 GB RAM. The time limit is set to 1 hour and the memory limit set to 10 GB for each instance. |