AiSAQ: All-in-Storage ANNS with Product Quantization for DRAM-free Information Retrieval
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2024
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866915172985077760 |
|---|---|
| author | Tatsuno, Kento Miyashita, Daisuke Ikeda, Taiga Ishiyama, Kiyoshi Sumiyoshi, Kazunari Deguchi, Jun |
| author_facet | Tatsuno, Kento Miyashita, Daisuke Ikeda, Taiga Ishiyama, Kiyoshi Sumiyoshi, Kazunari Deguchi, Jun |
| contents | Graph-based approximate nearest neighbor search (ANNS) algorithms work effectively against large-scale vector retrieval. Among such methods, DiskANN achieves good recall-speed tradeoffs using both DRAM and storage. DiskANN adopts product quantization (PQ) to reduce memory usage, which is still proportional to the scale of datasets. In this paper, we propose All-in-Storage ANNS with Product Quantization (AiSAQ), which offloads compressed vectors to the SSD index. Our method achieves $\sim$10 MB memory usage in query search with billion-scale datasets without critical latency degradation. AiSAQ also reduces the index load time for query search preparation, which enables fast switch between muitiple billion-scale indices.This method can be applied to retrievers of retrieval-augmented generation (RAG) and be scaled out with multiple-server systems for emerging datasets. Our DiskANN-based implementation is available on GitHub. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2404_06004 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | AiSAQ: All-in-Storage ANNS with Product Quantization for DRAM-free Information Retrieval Tatsuno, Kento Miyashita, Daisuke Ikeda, Taiga Ishiyama, Kiyoshi Sumiyoshi, Kazunari Deguchi, Jun Information Retrieval Computation and Language Data Structures and Algorithms Graph-based approximate nearest neighbor search (ANNS) algorithms work effectively against large-scale vector retrieval. Among such methods, DiskANN achieves good recall-speed tradeoffs using both DRAM and storage. DiskANN adopts product quantization (PQ) to reduce memory usage, which is still proportional to the scale of datasets. In this paper, we propose All-in-Storage ANNS with Product Quantization (AiSAQ), which offloads compressed vectors to the SSD index. Our method achieves $\sim$10 MB memory usage in query search with billion-scale datasets without critical latency degradation. AiSAQ also reduces the index load time for query search preparation, which enables fast switch between muitiple billion-scale indices.This method can be applied to retrievers of retrieval-augmented generation (RAG) and be scaled out with multiple-server systems for emerging datasets. Our DiskANN-based implementation is available on GitHub. |
| title | AiSAQ: All-in-Storage ANNS with Product Quantization for DRAM-free Information Retrieval |
| topic | Information Retrieval Computation and Language Data Structures and Algorithms |
| url | https://arxiv.org/abs/2404.06004 |