Forgetting to Learn Logic Programs

Authors: Andrew Cropper3676-3683

AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Our experimental results show that Forgetgol outperforms the alternative approaches when learning from over 10,000 tasks.
Researcher Affiliation Academia Andrew Cropper University of Oxford andrew.cropper@cs.ox.ac.uk
Pseudocode Yes Algorithm 1 Forgetgol
Open Source Code Yes All the experimental data are available at https://github.com/andrewcropper/aaai20-forgetgol.
Open Datasets Yes All the experimental data are available at https://github.com/andrewcropper/aaai20-forgetgol.
Dataset Splits No The paper uses a multi-task learning setup where tasks are generated. It measures performance on the percentage of tasks solved and learning times, but does not specify traditional train/validation/test splits for each task's underlying examples, nor does it specify a validation split for the overall set of tasks.
Hardware Specification No The paper does not specify the hardware used for the experiments.
Software Dependencies No The paper mentions Metagol, a MIL system based on a Prolog meta-interpreter, but does not provide specific version numbers for any software dependencies.
Experiment Setup Yes We enforce a timeout of 60 seconds per task per search depth. We set the maximum program size to 6 clauses.