GICDM: Mitigating Hubness for Reliable Distance-Based Generative Model Evaluation
Fuente:
arXiv
Salvato in:
| Autori principali: | , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2026
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866917543634010112 |
|---|---|
| author | Salvy, Nicolas Talbot, Hugues Thirion, Bertrand |
| author_facet | Salvy, Nicolas Talbot, Hugues Thirion, Bertrand |
| contents | Generative model evaluation commonly relies on high-dimensional embedding spaces to compute distances between samples. We show that dataset representations in these spaces are affected by the hubness phenomenon, which distorts nearest-neighbor relationships and biases distance-based metrics. Building on the classical Iterative Contextual Dissimilarity Measure (ICDM), we introduce Generative ICDM (GICDM), a method to correct neighborhood estimation for both real and generated data. We introduce a multi-scale extension to improve empirical behavior. Extensive experiments on synthetic and real benchmarks demonstrate that GICDM resolves hubness-induced failures, restores reliable metric behavior, and improves alignment with human assessment. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_16449 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | GICDM: Mitigating Hubness for Reliable Distance-Based Generative Model Evaluation Salvy, Nicolas Talbot, Hugues Thirion, Bertrand Machine Learning Artificial Intelligence Generative model evaluation commonly relies on high-dimensional embedding spaces to compute distances between samples. We show that dataset representations in these spaces are affected by the hubness phenomenon, which distorts nearest-neighbor relationships and biases distance-based metrics. Building on the classical Iterative Contextual Dissimilarity Measure (ICDM), we introduce Generative ICDM (GICDM), a method to correct neighborhood estimation for both real and generated data. We introduce a multi-scale extension to improve empirical behavior. Extensive experiments on synthetic and real benchmarks demonstrate that GICDM resolves hubness-induced failures, restores reliable metric behavior, and improves alignment with human assessment. |
| title | GICDM: Mitigating Hubness for Reliable Distance-Based Generative Model Evaluation |
| topic | Machine Learning Artificial Intelligence |
| url | https://arxiv.org/abs/2602.16449 |