Instance variant nearest neighbor using particle swarm optimization for function approximation
| Autor: | Kyoung-Yun Kim, Junheung Park |
|---|---|
| Rok vydání: | 2016 |
| Předmět: |
0209 industrial biotechnology
business.industry Nearest neighbor search Particle swarm optimization Pattern recognition 02 engineering and technology Function (mathematics) k-nearest neighbors algorithm Support vector machine 020901 industrial engineering & automation Function approximation 0202 electrical engineering electronic engineering information engineering 020201 artificial intelligence & image processing Artificial intelligence Multi-swarm optimization business Fixed-radius near neighbors Algorithm Software Mathematics |
| Zdroj: | Applied Soft Computing. 40:331-341 |
| ISSN: | 1568-4946 |
| DOI: | 10.1016/j.asoc.2015.10.055 |
| Popis: | An approach to the function approximation problem using IVNN is presented.IVNN to apply less neighbors for noisy data and more neighbors for non-noisy data.IVNN solved by a PSO such that the prediction performance is maximized.IVNN requires no model/distribution assumption for variant numbers of neighbors.The prediction performance compared to conventional algorithms. In this paper, a new approach called 'instance variant nearest neighbor' approximates a regression surface of a function using the concept of k nearest neighbor. Instead of fixed k neighbors for the entire dataset, our assumption is that there are optimal k neighbors for each data instance that best approximates the original function by fitting the local regions. This approach can be beneficial to noisy datasets where local regions form data characteristics that are different from the major data clusters. We formulate the problem of finding such k neighbors for each data instance as a combinatorial optimization problem, which is solved by a particle swarm optimization. The particle swarm optimization is extended with a rounding scheme that rounds up or down continuous-valued candidate solutions to integers, a number of k neighbors. We apply our new approach to five real-world regression datasets and compare its prediction performance with other function approximation algorithms, including the standard k nearest neighbor, multi-layer perceptron, and support vector regression. We observed that the instance variant nearest neighbor outperforms these algorithms in several datasets. In addition, our new approach provides consistent outputs with five datasets where other algorithms perform poorly. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full Text from ScienceDirect