Salvato in:
Dettagli Bibliografici
Autori principali: Lee, Su Hyeong, Zhang, Qingqi, Kondor, Risi
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:https://arxiv.org/abs/2402.06662
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866910452235108352
author Lee, Su Hyeong
Zhang, Qingqi
Kondor, Risi
author_facet Lee, Su Hyeong
Zhang, Qingqi
Kondor, Risi
contents Inner product-based decoders are among the most influential frameworks used to extract meaningful data from latent embeddings. However, such decoders have shown limitations in representation capacity in numerous works within the literature, which have been particularly notable in graph reconstruction problems. In this paper, we provide the first theoretical elucidation of this pervasive phenomenon in graph data, and suggest straightforward modifications to circumvent this issue without deviating from the inner product framework.
format Preprint
id arxiv_https___arxiv_org_abs_2402_06662
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Sign Rank Limitations for Inner Product Graph Decoders
Lee, Su Hyeong
Zhang, Qingqi
Kondor, Risi
Machine Learning
Inner product-based decoders are among the most influential frameworks used to extract meaningful data from latent embeddings. However, such decoders have shown limitations in representation capacity in numerous works within the literature, which have been particularly notable in graph reconstruction problems. In this paper, we provide the first theoretical elucidation of this pervasive phenomenon in graph data, and suggest straightforward modifications to circumvent this issue without deviating from the inner product framework.
title Sign Rank Limitations for Inner Product Graph Decoders
topic Machine Learning
url https://arxiv.org/abs/2402.06662