PageRenderTime 38ms CodeModel.GetById 15ms RepoModel.GetById 0ms app.codeStats 0ms

Unknown | 42 lines | 38 code | 4 blank | 0 comment | 0 complexity | 1674c7179912857a1fc76dd81392d361 MD5 | raw file
  1. \name{}
  2. \alias{}
  3. \title{
  4. Convert a Set of Feature Fingerprints to Binary Fingerprints
  5. }
  6. \description{
  7. Most feature vector style fingerprints (such as circular fingerprints) are
  8. of variable length, with the features being pulled from an implicit (and
  9. very large) universe of features.
  10. While it is possible to convert a single feature fingerprint to a binary vector
  11. via a hashing procedure, one cannot convert the feature representation into 1:1
  12. binary representation.
  13. However, for a collection of feature fingerprints, one can define a "local" universe
  14. as the union of the features contained in the set of fingerprints. With this definition
  15. each feature can be mapped to a single bit and thus each fingerprint can be converted
  16. to a fixed length, keyed fingerprint.
  17. Since circular fingerprints can lead to a very large binary fingerprint, the initial
  18. fingerprint is usually folded multiple times to achieve a desired bit density.
  19. Since the fingerprint folding procedure divides a fingerprint into two equal halves and
  20. OR's the two halves together, the starting fingerprint must be of even length. For circular
  21. fingerprints that may not lead to an even-length initial bit string, an extra bit is added
  22. at the top of the bit string.
  23. }
  24. \usage{
  25., bit.length = 256)
  26. }
  27. \arguments{
  28. \item{fps}{
  29. A list of \code{featvec} objects
  30. }
  31. \item{bit.length}{The length of the bit string to stop folding at}
  32. }
  33. \value{
  34. A list of objects of class \code{fingerprint}
  35. }
  36. \seealso{\code{\link{fold}}}
  37. \keyword{logic}
  38. \author{Rajarshi Guha \email{}}