{"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/bolt-accelerated-data-mining-with-fast-vector","title":"Bolt: Accelerated Data Mining with Fast Vector Compression","arxiv_id":"1706.10283","date":"2017-06-30","proceeding":null,"authors":["Davis W. Blalock","John V. Guttag"],"abstract":"Vectors of data are at the heart of machine learning and data mining.\nRecently, vector quantization methods have shown great promise in reducing both\nthe time and space costs of operating on vectors. We introduce a vector\nquantization algorithm that can compress vectors over 12x faster than existing\ntechniques while also accelerating approximate vector operations such as\ndistance and dot product computations by up to 10x. Because it can encode over\n2GB of vectors per second, it makes vector quantization cheap enough to employ\nin many more circumstances. For example, using our technique to compute\napproximate dot products in a nested loop can multiply matrices faster than a\nstate-of-the-art BLAS implementation, even when our algorithm must first\ncompress the matrices.\n  In addition to showing the above speedups, we demonstrate that our approach\ncan accelerate nearest neighbor search and maximum inner product search by over\n100x compared to floating point operations and up to 10x compared to other\nvector quantization methods. Our approximate Euclidean distance and dot product\ncomputations are not only faster than those of related algorithms with slower\nencodings, but also faster than Hamming distance computations, which have\ndirect hardware support on the tested platforms. We also assess the errors of\nour algorithm's approximate distances and dot products, and find that it is\ncompetitive with existing, slower vector quantization algorithms.","url_abs":"http://arxiv.org/abs/1706.10283v1","url_pdf":"http://arxiv.org/pdf/1706.10283v1.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":"bolt-accelerated-data-mining-with-fast-vector","repo_url":"https://github.com/dblalock/bolt","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok","spdx":"MPL-2.0"}}],"tasks":[{"task_slug":"quantization","task_name":"Quantization"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1706.10283","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}