Approximate Nearest Neighbor Search by Residual Vector Quantization
A recently proposed product quantization method is efficient for large scale approximate nearest neighbor search, however, its performance on unstructured vectors is limited. This paper introduces residual vector quantization based approaches that are appropriate for unstructured vectors. Database v...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2010-12-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | http://www.mdpi.com/1424-8220/10/12/11259/ |
| _version_ | 1811278061079887872 |
|---|---|
| author | Cheng Wang Tao Guan Yongjian Chen |
| author_facet | Cheng Wang Tao Guan Yongjian Chen |
| author_sort | Cheng Wang |
| collection | DOAJ |
| description | A recently proposed product quantization method is efficient for large scale approximate nearest neighbor search, however, its performance on unstructured vectors is limited. This paper introduces residual vector quantization based approaches that are appropriate for unstructured vectors. Database vectors are quantized by residual vector quantizer. The reproductions are represented by short codes composed of their quantization indices. Euclidean distance between query vector and database vector is approximated by asymmetric distance, i.e., the distance between the query vector and the reproduction of the database vector. An efficient exhaustive search approach is proposed by fast computing the asymmetric distance. A straight forward non-exhaustive search approach is proposed for large scale search. Our approaches are compared to two state-of-the-art methods, spectral hashing and product quantization, on both structured and unstructured datasets. Results show that our approaches obtain the best results in terms of the trade-off between search quality and memory usage. |
| first_indexed | 2024-04-13T00:27:59Z |
| format | Article |
| id | doaj.art-8db492e7c4574d51b091b396ce4d5e52 |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-04-13T00:27:59Z |
| publishDate | 2010-12-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-8db492e7c4574d51b091b396ce4d5e522022-12-22T03:10:33ZengMDPI AGSensors1424-82202010-12-011012112591127310.3390/s101211259Approximate Nearest Neighbor Search by Residual Vector QuantizationCheng WangTao GuanYongjian ChenA recently proposed product quantization method is efficient for large scale approximate nearest neighbor search, however, its performance on unstructured vectors is limited. This paper introduces residual vector quantization based approaches that are appropriate for unstructured vectors. Database vectors are quantized by residual vector quantizer. The reproductions are represented by short codes composed of their quantization indices. Euclidean distance between query vector and database vector is approximated by asymmetric distance, i.e., the distance between the query vector and the reproduction of the database vector. An efficient exhaustive search approach is proposed by fast computing the asymmetric distance. A straight forward non-exhaustive search approach is proposed for large scale search. Our approaches are compared to two state-of-the-art methods, spectral hashing and product quantization, on both structured and unstructured datasets. Results show that our approaches obtain the best results in terms of the trade-off between search quality and memory usage.http://www.mdpi.com/1424-8220/10/12/11259/approximate nearest neighbor searchhigh-dimensional indexingresidual vector quantization |
| spellingShingle | Cheng Wang Tao Guan Yongjian Chen Approximate Nearest Neighbor Search by Residual Vector Quantization Sensors approximate nearest neighbor search high-dimensional indexing residual vector quantization |
| title | Approximate Nearest Neighbor Search by Residual Vector Quantization |
| title_full | Approximate Nearest Neighbor Search by Residual Vector Quantization |
| title_fullStr | Approximate Nearest Neighbor Search by Residual Vector Quantization |
| title_full_unstemmed | Approximate Nearest Neighbor Search by Residual Vector Quantization |
| title_short | Approximate Nearest Neighbor Search by Residual Vector Quantization |
| title_sort | approximate nearest neighbor search by residual vector quantization |
| topic | approximate nearest neighbor search high-dimensional indexing residual vector quantization |
| url | http://www.mdpi.com/1424-8220/10/12/11259/ |
| work_keys_str_mv | AT chengwang approximatenearestneighborsearchbyresidualvectorquantization AT taoguan approximatenearestneighborsearchbyresidualvectorquantization AT yongjianchen approximatenearestneighborsearchbyresidualvectorquantization |