TUTCRIS - Tampereen teknillinen yliopisto

TUTCRIS

Multiplicative update for fast optimization of information retrieval based neighbor embedding

Tutkimustuotosvertaisarvioitu

Standard

Multiplicative update for fast optimization of information retrieval based neighbor embedding. / Peltonen, Jaakko; Lin, Ziyuan.

2013 IEEE International Workshop on Machine Learning for Signal Processing - Proceedings of MLSP 2013. 2013. 6661899.

Tutkimustuotosvertaisarvioitu

Harvard

Peltonen, J & Lin, Z 2013, Multiplicative update for fast optimization of information retrieval based neighbor embedding. julkaisussa 2013 IEEE International Workshop on Machine Learning for Signal Processing - Proceedings of MLSP 2013., 6661899, Southampton, Iso-Britannia, 22/09/13. https://doi.org/10.1109/MLSP.2013.6661899

APA

Peltonen, J., & Lin, Z. (2013). Multiplicative update for fast optimization of information retrieval based neighbor embedding. teoksessa 2013 IEEE International Workshop on Machine Learning for Signal Processing - Proceedings of MLSP 2013 [6661899] https://doi.org/10.1109/MLSP.2013.6661899

Vancouver

Peltonen J, Lin Z. Multiplicative update for fast optimization of information retrieval based neighbor embedding. julkaisussa 2013 IEEE International Workshop on Machine Learning for Signal Processing - Proceedings of MLSP 2013. 2013. 6661899 https://doi.org/10.1109/MLSP.2013.6661899

Author

Peltonen, Jaakko ; Lin, Ziyuan. / Multiplicative update for fast optimization of information retrieval based neighbor embedding. 2013 IEEE International Workshop on Machine Learning for Signal Processing - Proceedings of MLSP 2013. 2013.

Bibtex - Lataa

@inproceedings{b0f8e84d312b40c9a12dd2186ed229a0,
title = "Multiplicative update for fast optimization of information retrieval based neighbor embedding",
abstract = "Dimensionality reduction of high-dimensional data for visualization has recently been formalized as an information retrieval task where original neighbors of data points are retrieved from the low-dimensional display, and the visualization is optimized to maximize flexible tradeoffs between precision and recall of the retrieval, avoiding misses and false neighbors. The approach has yielded well-performing visualization methods as well as information retrieval interpretations of earlier neighbor embedding methods. However, most of the methods are based on slow gradient search approaches, whereas fast methods are crucial for example in interactive applications. In this paper we propose a fast multiplicative update rule for visualization optimized for information retrieval, and show in experiments it yields equally good results as the previous state of the art gradient based approach but much faster.",
keywords = "dimensionality reduction, information retrieval, multiplicative update, visualization",
author = "Jaakko Peltonen and Ziyuan Lin",
year = "2013",
doi = "10.1109/MLSP.2013.6661899",
language = "English",
isbn = "9781479911806",
booktitle = "2013 IEEE International Workshop on Machine Learning for Signal Processing - Proceedings of MLSP 2013",

}

RIS (suitable for import to EndNote) - Lataa

TY - GEN

T1 - Multiplicative update for fast optimization of information retrieval based neighbor embedding

AU - Peltonen, Jaakko

AU - Lin, Ziyuan

PY - 2013

Y1 - 2013

N2 - Dimensionality reduction of high-dimensional data for visualization has recently been formalized as an information retrieval task where original neighbors of data points are retrieved from the low-dimensional display, and the visualization is optimized to maximize flexible tradeoffs between precision and recall of the retrieval, avoiding misses and false neighbors. The approach has yielded well-performing visualization methods as well as information retrieval interpretations of earlier neighbor embedding methods. However, most of the methods are based on slow gradient search approaches, whereas fast methods are crucial for example in interactive applications. In this paper we propose a fast multiplicative update rule for visualization optimized for information retrieval, and show in experiments it yields equally good results as the previous state of the art gradient based approach but much faster.

AB - Dimensionality reduction of high-dimensional data for visualization has recently been formalized as an information retrieval task where original neighbors of data points are retrieved from the low-dimensional display, and the visualization is optimized to maximize flexible tradeoffs between precision and recall of the retrieval, avoiding misses and false neighbors. The approach has yielded well-performing visualization methods as well as information retrieval interpretations of earlier neighbor embedding methods. However, most of the methods are based on slow gradient search approaches, whereas fast methods are crucial for example in interactive applications. In this paper we propose a fast multiplicative update rule for visualization optimized for information retrieval, and show in experiments it yields equally good results as the previous state of the art gradient based approach but much faster.

KW - dimensionality reduction

KW - information retrieval

KW - multiplicative update

KW - visualization

UR - http://www.scopus.com/inward/record.url?scp=84893243408&partnerID=8YFLogxK

U2 - 10.1109/MLSP.2013.6661899

DO - 10.1109/MLSP.2013.6661899

M3 - Conference contribution

SN - 9781479911806

BT - 2013 IEEE International Workshop on Machine Learning for Signal Processing - Proceedings of MLSP 2013

ER -