{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/vector-and-line-quantization-for-billion","title":"Vector and Line Quantization for Billion-scale Similarity Search on GPUs","arxiv_id":"1901.00275","date":"2019-01-02","proceeding":null,"authors":["Wei Chen","Jincai Chen","Fuhao Zou","Yuan-Fang Li","Ping Lu","Qiang Wang","Wei Zhao"],"abstract":"Billion-scale high-dimensional approximate nearest neighbour (ANN) search has\nbecome an important problem for searching similar objects among the vast amount\nof images and videos available online. The existing ANN methods are usually\ncharacterized by their specific indexing structures, including the inverted\nindex and the inverted multi-index structure. The inverted index structure is\namenable to GPU-based implementations, and the state-of-the-art systems such as\nFaiss are able to exploit the massive parallelism offered by GPUs. However, the\ninverted index requires high memory overhead to index the dataset effectively.\nThe inverted multi-index structure is difficult to implement for GPUs, and also\nineffective in dealing with database with different data distributions. In this\npaper we propose a novel hierarchical inverted index structure generated by\nvector and line quantization methods. Our quantization method improves both\nsearch efficiency and accuracy, while maintaining comparable memory\nconsumption. This is achieved by reducing search space and increasing the\nnumber of indexed regions. We introduce a new ANN search system, VLQ-ADC, that\nis based on the proposed inverted index, and perform extensive evaluation on\ntwo public billion-scale benchmark datasets SIFT1B and DEEP1B. Our evaluation\nshows that VLQ-ADC significantly outperforms the state-of-the-art GPU- and\nCPU-based systems in terms of both accuracy and search speed. The source code\nof VLQ-ADC is available at\nhttps://github.com/zjuchenwei/vector-line-quantization.","url_abs":"http://arxiv.org/abs/1901.00275v2","url_pdf":"http://arxiv.org/pdf/1901.00275v2.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"vector-and-line-quantization-for-billion","repo_url":"https://github.com/zjuchenwei/vector-line-quantization","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"none","reach":null}],"tasks":[{"task_slug":null,"task_name":"CPU"},{"task_slug":null,"task_name":"GPU"},{"task_slug":"quantization","task_name":"Quantization"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}