Learning logic programs by explaining their failures
<p>Scientists form hypotheses and experimentally test them. If a hypothesis fails (is refuted), scientists try to <em>explain</em> the failure to eliminate other hypotheses. The more precise the failure analysis the more hypotheses can be eliminated. Thus inspired...
| Main Authors: | , |
|---|---|
| Format: | Journal article |
| Language: | English |
| Published: |
Springer
2023
|
| _version_ | 1826313022531960832 |
|---|---|
| author | Morel, R Cropper, A |
| author_facet | Morel, R Cropper, A |
| author_sort | Morel, R |
| collection | OXFORD |
| description | <p>Scientists form hypotheses and experimentally test them. If a hypothesis fails (is refuted), scientists try to <em>explain</em> the failure to eliminate other hypotheses. The more precise the failure analysis the more hypotheses can be eliminated. Thus inspired, we introduce failure explanation techniques for inductive logic programming. Given a hypothesis represented as a logic program, we test it on examples. If a hypothesis fails, we explain the failure in terms of failing sub-programs. In case a positive example fails, we identify failing sub-programs at the granularity of literals. We introduce a failure explanation algorithm based on analysing branches of SLD-trees. We integrate a meta-interpreter based implementation of this algorithm with the test-stage of the POPPER ILP system. We show that fine-grained failure analysis allows for learning fine-grained constraints on the hypothesis space. Our experimental results show that explaining failures can drastically reduce hypothesis space exploration and learning times.</p> |
| first_indexed | 2024-09-25T04:04:23Z |
| format | Journal article |
| id | oxford-uuid:ef44dde5-3671-4945-963f-7b97fbad8a7f |
| institution | University of Oxford |
| language | English |
| last_indexed | 2024-09-25T04:04:23Z |
| publishDate | 2023 |
| publisher | Springer |
| record_format | dspace |
| spelling | oxford-uuid:ef44dde5-3671-4945-963f-7b97fbad8a7f2024-05-13T11:33:45ZLearning logic programs by explaining their failuresJournal articlehttp://purl.org/coar/resource_type/c_dcae04bcuuid:ef44dde5-3671-4945-963f-7b97fbad8a7fEnglishSymplectic ElementsSpringer2023Morel, RCropper, A<p>Scientists form hypotheses and experimentally test them. If a hypothesis fails (is refuted), scientists try to <em>explain</em> the failure to eliminate other hypotheses. The more precise the failure analysis the more hypotheses can be eliminated. Thus inspired, we introduce failure explanation techniques for inductive logic programming. Given a hypothesis represented as a logic program, we test it on examples. If a hypothesis fails, we explain the failure in terms of failing sub-programs. In case a positive example fails, we identify failing sub-programs at the granularity of literals. We introduce a failure explanation algorithm based on analysing branches of SLD-trees. We integrate a meta-interpreter based implementation of this algorithm with the test-stage of the POPPER ILP system. We show that fine-grained failure analysis allows for learning fine-grained constraints on the hypothesis space. Our experimental results show that explaining failures can drastically reduce hypothesis space exploration and learning times.</p> |
| spellingShingle | Morel, R Cropper, A Learning logic programs by explaining their failures |
| title | Learning logic programs by explaining their failures |
| title_full | Learning logic programs by explaining their failures |
| title_fullStr | Learning logic programs by explaining their failures |
| title_full_unstemmed | Learning logic programs by explaining their failures |
| title_short | Learning logic programs by explaining their failures |
| title_sort | learning logic programs by explaining their failures |
| work_keys_str_mv | AT morelr learninglogicprogramsbyexplainingtheirfailures AT croppera learninglogicprogramsbyexplainingtheirfailures |