Abstract
Spatial centrality, whereby samples closer to the center of a dataset tend to be closer to all other samples, is regarded as one source of hubness. Hubness is well known to degrade k-nearest-neighbor (k-NN) classification. Spatial centrality can be removed by centering, i.e., shifting the origin to the global center of the dataset, in cases where inner product similarity is used. However, when Euclidean distance is used, centering has no effect on spatial centrality because the distance between the samples is the same before and after centering. As described in this paper, we propose a solution for the hubness problem when Euclidean distance is considered. We provide a theoretical explanation to demonstrate how the solution eliminates spatial centrality and reduces hubness. We then present some discussion of the reason the proposed solution works, from a viewpoint of density gradient, which is regarded as the origin of spatial centrality and hubness. We demonstrate that the solution corresponds to flattening the density gradient. Using real-world datasets, we demonstrate that the proposed method improves k-NN classification performance and outperforms an existing hub-reduction method.
| Original language | English |
|---|---|
| Title of host publication | 30th AAAI Conference on Artificial Intelligence, AAAI 2016 |
| Publisher | AAAI press |
| Pages | 1659-1665 |
| Number of pages | 7 |
| ISBN (Electronic) | 9781577357605 |
| Publication status | Published - 2016 |
| Externally published | Yes |
| Event | 30th AAAI Conference on Artificial Intelligence, AAAI 2016 - Phoenix, United States Duration: 2016 Feb 12 → 2016 Feb 17 |
Other
| Other | 30th AAAI Conference on Artificial Intelligence, AAAI 2016 |
|---|---|
| Country | United States |
| City | Phoenix |
| Period | 16/2/12 → 16/2/17 |
ASJC Scopus subject areas
- Artificial Intelligence
Cite this
Flattening the density gradient for eliminating spatial centrality to reduce hubness. / Hara, Kazuo; Suzuki, Ikumi; Kobayashi, Kei; Fukumizu, Kenji; Radovanovíc, Milǒs.
30th AAAI Conference on Artificial Intelligence, AAAI 2016. AAAI press, 2016. p. 1659-1665.Research output: Chapter in Book/Report/Conference proceeding › Conference contribution
}
TY - GEN
T1 - Flattening the density gradient for eliminating spatial centrality to reduce hubness
AU - Hara, Kazuo
AU - Suzuki, Ikumi
AU - Kobayashi, Kei
AU - Fukumizu, Kenji
AU - Radovanovíc, Milǒs
PY - 2016
Y1 - 2016
N2 - Spatial centrality, whereby samples closer to the center of a dataset tend to be closer to all other samples, is regarded as one source of hubness. Hubness is well known to degrade k-nearest-neighbor (k-NN) classification. Spatial centrality can be removed by centering, i.e., shifting the origin to the global center of the dataset, in cases where inner product similarity is used. However, when Euclidean distance is used, centering has no effect on spatial centrality because the distance between the samples is the same before and after centering. As described in this paper, we propose a solution for the hubness problem when Euclidean distance is considered. We provide a theoretical explanation to demonstrate how the solution eliminates spatial centrality and reduces hubness. We then present some discussion of the reason the proposed solution works, from a viewpoint of density gradient, which is regarded as the origin of spatial centrality and hubness. We demonstrate that the solution corresponds to flattening the density gradient. Using real-world datasets, we demonstrate that the proposed method improves k-NN classification performance and outperforms an existing hub-reduction method.
AB - Spatial centrality, whereby samples closer to the center of a dataset tend to be closer to all other samples, is regarded as one source of hubness. Hubness is well known to degrade k-nearest-neighbor (k-NN) classification. Spatial centrality can be removed by centering, i.e., shifting the origin to the global center of the dataset, in cases where inner product similarity is used. However, when Euclidean distance is used, centering has no effect on spatial centrality because the distance between the samples is the same before and after centering. As described in this paper, we propose a solution for the hubness problem when Euclidean distance is considered. We provide a theoretical explanation to demonstrate how the solution eliminates spatial centrality and reduces hubness. We then present some discussion of the reason the proposed solution works, from a viewpoint of density gradient, which is regarded as the origin of spatial centrality and hubness. We demonstrate that the solution corresponds to flattening the density gradient. Using real-world datasets, we demonstrate that the proposed method improves k-NN classification performance and outperforms an existing hub-reduction method.
UR - http://www.scopus.com/inward/record.url?scp=85007152062&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85007152062&partnerID=8YFLogxK
M3 - Conference contribution
AN - SCOPUS:85007152062
SP - 1659
EP - 1665
BT - 30th AAAI Conference on Artificial Intelligence, AAAI 2016
PB - AAAI press
ER -