AgraSSt: approximate graph Stein statistics for interpretable assessment of implicit graph generators
We propose and analyse a novel statistical procedure, coined AgraSSt, to assess the quality of graph generators which may not be available in explicit forms. In particular, AgraSSt can be used to determine whether a learned graph generating process is capable of generating graphs which resemble a gi...
| Main Authors: | , |
|---|---|
| Format: | Conference item |
| Language: | English |
| Published: |
Curran Associates
2023
|
| _version_ | 1826312051394347008 |
|---|---|
| author | Xu, W Reinert, G |
| author_facet | Xu, W Reinert, G |
| author_sort | Xu, W |
| collection | OXFORD |
| description | We propose and analyse a novel statistical procedure, coined AgraSSt, to assess the quality of graph generators which may not be available in explicit forms. In particular, AgraSSt can be used to determine whether a learned graph generating process is capable of generating graphs which resemble a given input graph. Inspired by Stein operators for random graphs, the key idea of AgraSSt is the construction of a kernel discrepancy based on an operator obtained from the graph generator. AgraSSt can provide interpretable criticisms for a graph generator training procedure and help identify reliable sample batches for downstream tasks. We give theoretical guarantees for a broad class of random graph models. Moreover, we provide empirical results on both synthetic input graphs with known graph generation procedures, and real-world input graphs that the state-of-the-art (deep) generative models for graphs are trained on. |
| first_indexed | 2024-03-07T08:21:52Z |
| format | Conference item |
| id | oxford-uuid:912b01de-0cc8-4e81-b721-0c3de67a1653 |
| institution | University of Oxford |
| language | English |
| last_indexed | 2024-03-07T08:21:52Z |
| publishDate | 2023 |
| publisher | Curran Associates |
| record_format | dspace |
| spelling | oxford-uuid:912b01de-0cc8-4e81-b721-0c3de67a16532024-01-30T11:32:39ZAgraSSt: approximate graph Stein statistics for interpretable assessment of implicit graph generatorsConference itemhttp://purl.org/coar/resource_type/c_5794uuid:912b01de-0cc8-4e81-b721-0c3de67a1653EnglishSymplectic ElementsCurran Associates2023Xu, WReinert, GWe propose and analyse a novel statistical procedure, coined AgraSSt, to assess the quality of graph generators which may not be available in explicit forms. In particular, AgraSSt can be used to determine whether a learned graph generating process is capable of generating graphs which resemble a given input graph. Inspired by Stein operators for random graphs, the key idea of AgraSSt is the construction of a kernel discrepancy based on an operator obtained from the graph generator. AgraSSt can provide interpretable criticisms for a graph generator training procedure and help identify reliable sample batches for downstream tasks. We give theoretical guarantees for a broad class of random graph models. Moreover, we provide empirical results on both synthetic input graphs with known graph generation procedures, and real-world input graphs that the state-of-the-art (deep) generative models for graphs are trained on. |
| spellingShingle | Xu, W Reinert, G AgraSSt: approximate graph Stein statistics for interpretable assessment of implicit graph generators |
| title | AgraSSt: approximate graph Stein statistics for interpretable assessment of implicit graph generators |
| title_full | AgraSSt: approximate graph Stein statistics for interpretable assessment of implicit graph generators |
| title_fullStr | AgraSSt: approximate graph Stein statistics for interpretable assessment of implicit graph generators |
| title_full_unstemmed | AgraSSt: approximate graph Stein statistics for interpretable assessment of implicit graph generators |
| title_short | AgraSSt: approximate graph Stein statistics for interpretable assessment of implicit graph generators |
| title_sort | agrasst approximate graph stein statistics for interpretable assessment of implicit graph generators |
| work_keys_str_mv | AT xuw agrasstapproximategraphsteinstatisticsforinterpretableassessmentofimplicitgraphgenerators AT reinertg agrasstapproximategraphsteinstatisticsforinterpretableassessmentofimplicitgraphgenerators |