Learning high-dimensional data

Verleysen, Michel
(2003) Limitations and Future Trends in Neural Computation — p. 141-162, published

Files

6-Learninghigh-dimensionaldata.pdf
  • Restricted Access
  • Adobe PDF
  • 587.53 KB
Request a copy

Details

Authors
Abstract
Observations from real-world problems are often high-dimensional vectors, i.e. made up of many variables. Learning methods, including artificial neural networks, often have difficulties to handle a relatively small number of high-dimensional data. In this paper, we show how concepts gained from our intuition on 2- and 3-dimensional data can be misleading when used in high-dimensional settings. When then show how the "curse of dimensionality" and the "empty space phenomenon" can be taken into account in the design of neural network algorithms, and how non-linear dimension reduction techniques can be used to circumvent the problem. We conclude by an illustrative example of this last method on the forecasting of financial time series.
Affiliations

Citations

  • APA
  • Chicago
  • FWB

Verleysen, M. (2003). Learning high-dimensional data. In S. Ablameyko, L. Goras, M. Gori, V. Piuri (ed.), Limitations and Future Trends in Neural Computation (p. p. 141-162). IOS Press. https://hdl.handle.net/2078.5/253779