Abstract | ||
---|---|---|
We have described a running system that embodies efficient parallel implementations of several symbolic machine learning induction operators. It functions as an “Induction Assistant” to a domain expert. First we developed an efficient, noise-tolerant, similarity-based parallel matching algorithm. This should apply to other graph-based representations of domains possessing an embedding in which the low-level features (relations or groupings) are mostly local. It was used as infrastructure to construct efficient parallel implementations of several symbolic machine learning induction operators. Finally, the induction operators were sandwiched together with sets of filters (both syntactic and empirical) to compose a crude form of induction scripts, which are invoked by a domain expert. The matching algorithm has very attractive scaling properties as the size of the problem and/or the number of processors increases. Hardware usage is efficient. The results reported in this article were obtained on an 8K CM-2 Connection Machine. The implemented system was used to discover something previously unknown to the domain expert [47]. |
Year | DOI | Venue |
---|---|---|
1993 | 10.1109/HICSS.1991.183931 | System Sciences, 1991. Proceedings of the Twenty-Fourth Annual Hawaii International Conference |
Keywords | Field | DocType |
learning systems,macromolecular configurations,parallel algorithms,proteins,dna polymerases,computational complexity,hardware induction,induction operators,machine learning,protein structure,symbolic induction,time complexity,transcriptional activators,hardware,computational biology,protein engineering,concurrent computing,artificial intelligence,dna polymerase,front end,linear space,matched filters | Common Lisp,Protein structure function,Programming language,Subroutine,Computer science,Parallel algorithm,Massively parallel,Lisp,Computational learning theory,Time complexity | Conference |
Volume | ISBN | Citations |
i | 3-540-56483-7 | 1 |
PageRank | References | Authors |
1.64 | 4 | 4 |
Name | Order | Citations | PageRank |
---|---|---|---|
Richard H. Lathrop | 1 | 1358 | 241.79 |
T A Webster | 2 | 180 | 132.66 |
Temple F. Smith | 3 | 139 | 73.26 |
Patrick H. Winston | 4 | 370 | 559.01 |