Content deleted Content added
Altered url. URLs might have been anonymized. Add: class, date, title, eprint, issue, pmid, authors 1-9. Removed parameters. Some additions/deletions were parameter name changes. | Use this tool. Report bugs. | #UCB_Gadget |
fix formatting |
||
Line 4:
Product quantization can be used by itself or as a component of more complex ANN search algorithms.<ref name="douze2024faiss">{{cite arXiv |eprint=2401.08281 |last1=Douze |first1=Matthijs |last2=Guzhva |first2=Alexandr |last3=Deng |first3=Chengqi |last4=Johnson |first4=Jeff |last5=Szilvasy |first5=Gergely |last6=Mazaré |first6=Pierre-Emmanuel |last7=Lomeli |first7=Maria |last8=Hosseini |first8=Lucas |last9=Jégou |first9=Hervé |title=The Faiss library |date=2024 |class=cs.LG }}</ref> The combination of product quantization with inverted files is sometimes known as IVFDAC (inverted file system with asymmetric distance computation): this involves dividing the database into coarse buckets and, for a given query, doing distance computations only with the buckets nearest to that query.<ref name="matsui2018survey">{{cite journal |last1=Matsui |first1=Yusuke |last2=Uchida |first2=Yusuke |last3=Jégou |first3=Hervé |last4=Satoh |first4=Shin'ichi |title=A Survey of Product Quantization |journal=ITE Transactions on Media Technology and Applications |date=2018 |volume=6 |issue=1 |pages=2–10 |doi=10.3169/mta.6.2 |url=https://www.jstage.jst.go.jp/article/mta/6/1/6_2/_article/-char/ja/ |access-date=13 April 2025}}</ref>
==
'''Optimized product quantization''' is a widely used enhancement which applies a [[rotation matrix]] before quantizing vectors, in order to better take into account the data distribution.<ref name="matsui2018survey"></ref>
==References==
{{Reflist}}
{{Drafts moved from mainspace|date=March 2025}}
| |||