An introduction to nonlinear dimensionality reduction by maximum variance unfolding
AAAI'06 proceedings of the 21st national conference on Artificial intelligence - Volume 2
Unsupervised learning of image manifolds by semidefinite programming
CVPR'04 Proceedings of the 2004 IEEE computer society conference on Computer vision and pattern recognition
| Hi-index | 0.00 |
Seeking to identify the constituent parts of the multidimensional auditory attribute that musicians know as timbre, music psychologists have made extensive use of multidimensional scaling (mds), a statistical technique for visualising the geometric spaces implied by perceived dissimilarity. mdsis also well known in the machine learning community, where it is used as a basic technique for dimensionality reduction. We adapt a nonlinear variant of mdsthat is popular in machine learning, Isomap, for use in analysing psychological data and re-analyse three earlier experiments on human perception of timbre. Isomap is designed to eliminate undesirable nonlinearities in the input data in order to reduce the overall dimensionality; our results show that it succeeds in these goals for timbre spaces, compressing the output onto well-known dimensions of timbre and highlighting the challenges inherent in quantifying differences in spectral shape.