Unsupervised Learning by Convex and Conic Coding

@inproceedings{Lee1996UnsupervisedLB,
  title={Unsupervised Learning by Convex and Conic Coding},
  author={Daniel D. Lee and H. Sebastian Seung},
  booktitle={NIPS},
  year={1996}
}
Unsupervised learning algorithms based on convex and conic encoders are proposed. The encoders find the closest convex or conic combination of basis vectors to the input. The learning algorithms produce basis vectors that minimize the reconstruction error of the encoders. The convex algorithm develops locally linear models of the input, while the conic algorithm discovers features. Both algorithms are used to model handwritten digits and compared with vector quantization and principal component… CONTINUE READING
Highly Cited
This paper has 183 citations. REVIEW CITATIONS

3 Figures & Tables

Extracted Numerical Results

  • With r = 25 patterns per digit class, Convex incorrectly classified 113 digits out of the 2007 test examples for an overall error rate of 5.6%.
  • However, scaling up the convex models to r = 100 patterns results in an error rate of 4.4% (89 errors).
  • On the other hand, Conic coding with r = 25 results in an error rate of 6.8% (138 errors).

Topics

Statistics

01020'98'00'02'04'06'08'10'12'14'16'18
Citations per Year

184 Citations

Semantic Scholar estimates that this publication has 184 citations based on the available data.

See our FAQ for additional information.