k-Nearest Neighbors algorithms

Luckily, there are solutions for every problem above – for example, the Locality Sensitive Hashing for solving the expensive computation cost, Distance metric learning[3] for approximating the ‘correct’ distance metric, and Random projection[4], as well as other dimension reduction techniques, for the breaking the curse of dimensionality. These topics will be further exploit, in future blog posts.

References

[1] Cover and Hart (1967), Nearest neighbor pattern recognition, IEEE Transactions on Information Theory 13; p.21-27

[2] Beyer, K.; Goldstein, J.; Ramakrishnan, R.; Shaft, U. (1999), When is “Nearest Neighbor” Meaningful?, Proc. 7th International Conference on Database Theory – ICDT’99. LNCS. 1540: 217235

[3] Yang and Liu (2006), Distance Metric Learning: A Comprehensive Survey.

[4] Dimitris Achlioptas (2003), Database-friendly random projections: Johnson-Lindenstrauss with binary coins, Journay of Computer and System Sciences, 66(4):671687.