Methods and Implementations for Storing Sparse Vectors
- Technology Benefits
- Lossless compression
- Technology Application
- Database searching, drug discovery, cheminformatics
- Detailed Technology Description
- None
- Others
-
Additional Technologies by these Inventors
Tech ID/UC Case
21322/2007-793-0
Related Cases
2007-793-0
- *Abstract
-
This invention consists of a set of methods and algorithms to compress the information contained in large vectors of binary or integer variables. These vectors occur in a variety of applications where objects are represented by spectral fingerprints, which by nature tend to be large and sparse. By leveraging the power law distributions often observed in these spaces, researchers at UCI have developed new lossless compression methods using integer entropy coding. In contrast to current compression systems requiring 1024 bits to store each molecule, the UCI methods can achieve lossless compression to a mere 300-400 bits.
- *Principal Investigator
-
Name: Pierre Baldi
Department:
Name: Daniel Hirschberg
Department:
Name: S. Joshua Swamidass
Department:
- Country/Region
- USA