Tuning Database-Friendly Random Projection Matrices for Improved Distance Preservation on Specific Data
Random Projection is one of the most popular and successful dimensionality reduction algorithms for large volumes of data. However, given its stochastic nature, different initializations of the projection matrix can lead to very different levels of performance. This paper presents a guided random se...
| Main Authors: | , , , , |
|---|---|
| Other Authors: | |
| Format: | Article |
| Language: | English |
| Published: |
Springer US
2021
|
| Online Access: | https://hdl.handle.net/1721.1/133016 |
| _version_ | 1826217895573585920 |
|---|---|
| author | López-Sánchez, Daniel de Bodt, Cyril Lee, John A. Arrieta, Angélica G. Corchado, Juan M. |
| author2 | Massachusetts Institute of Technology. Media Laboratory |
| author_facet | Massachusetts Institute of Technology. Media Laboratory López-Sánchez, Daniel de Bodt, Cyril Lee, John A. Arrieta, Angélica G. Corchado, Juan M. |
| author_sort | López-Sánchez, Daniel |
| collection | MIT |
| description | Random Projection is one of the most popular and successful dimensionality reduction algorithms for large volumes of data. However, given its stochastic nature, different initializations of the projection matrix can lead to very different levels of performance. This paper presents a guided random search algorithm to mitigate this problem. The proposed method uses a small number of training data samples to iteratively adjust a projection matrix, improving its performance on similarly distributed data. Experimental results show that projection matrices generated with the proposed method result in a better preservation of distances between data samples. Conveniently, this is achieved while preserving the database-friendliness of the projection matrix, as it remains sparse and comprised exclusively of integers after being tuned with our algorithm. Moreover, running the proposed algorithm on a consumer-grade CPU requires only a few seconds. |
| first_indexed | 2024-09-23T17:10:45Z |
| format | Article |
| id | mit-1721.1/133016 |
| institution | Massachusetts Institute of Technology |
| language | English |
| last_indexed | 2024-09-23T17:10:45Z |
| publishDate | 2021 |
| publisher | Springer US |
| record_format | dspace |
| spelling | mit-1721.1/1330162024-06-05T23:44:20Z Tuning Database-Friendly Random Projection Matrices for Improved Distance Preservation on Specific Data López-Sánchez, Daniel de Bodt, Cyril Lee, John A. Arrieta, Angélica G. Corchado, Juan M. Massachusetts Institute of Technology. Media Laboratory Random Projection is one of the most popular and successful dimensionality reduction algorithms for large volumes of data. However, given its stochastic nature, different initializations of the projection matrix can lead to very different levels of performance. This paper presents a guided random search algorithm to mitigate this problem. The proposed method uses a small number of training data samples to iteratively adjust a projection matrix, improving its performance on similarly distributed data. Experimental results show that projection matrices generated with the proposed method result in a better preservation of distances between data samples. Conveniently, this is achieved while preserving the database-friendliness of the projection matrix, as it remains sparse and comprised exclusively of integers after being tuned with our algorithm. Moreover, running the proposed algorithm on a consumer-grade CPU requires only a few seconds. 2021-10-18T13:53:52Z 2021-10-18T13:53:52Z 2021-07 2021-10-17T03:14:38Z Article http://purl.org/eprint/type/JournalArticle 1573-7497 0924-669X https://hdl.handle.net/1721.1/133016 López-Sánchez, D., de Bodt, C., Lee, J.A. et al. Tuning Database-Friendly Random Projection Matrices for Improved Distance Preservation on Specific Data. Appl Intell (2021) en https://doi.org/10.1007/s10489-021-02626-6 Applied Intelligence Creative Commons Attribution https://creativecommons.org/licenses/by/4.0/ The Author(s) application/pdf Springer US Springer US |
| spellingShingle | López-Sánchez, Daniel de Bodt, Cyril Lee, John A. Arrieta, Angélica G. Corchado, Juan M. Tuning Database-Friendly Random Projection Matrices for Improved Distance Preservation on Specific Data |
| title | Tuning Database-Friendly Random Projection Matrices for Improved Distance Preservation on Specific Data |
| title_full | Tuning Database-Friendly Random Projection Matrices for Improved Distance Preservation on Specific Data |
| title_fullStr | Tuning Database-Friendly Random Projection Matrices for Improved Distance Preservation on Specific Data |
| title_full_unstemmed | Tuning Database-Friendly Random Projection Matrices for Improved Distance Preservation on Specific Data |
| title_short | Tuning Database-Friendly Random Projection Matrices for Improved Distance Preservation on Specific Data |
| title_sort | tuning database friendly random projection matrices for improved distance preservation on specific data |
| url | https://hdl.handle.net/1721.1/133016 |
| work_keys_str_mv | AT lopezsanchezdaniel tuningdatabasefriendlyrandomprojectionmatricesforimproveddistancepreservationonspecificdata AT debodtcyril tuningdatabasefriendlyrandomprojectionmatricesforimproveddistancepreservationonspecificdata AT leejohna tuningdatabasefriendlyrandomprojectionmatricesforimproveddistancepreservationonspecificdata AT arrietaangelicag tuningdatabasefriendlyrandomprojectionmatricesforimproveddistancepreservationonspecificdata AT corchadojuanm tuningdatabasefriendlyrandomprojectionmatricesforimproveddistancepreservationonspecificdata |