Title :
An approximate L1-difference algorithm for massive data streams
Author :
Feigenbaum, J. ; Kannan, S. ; Strauss, M. ; Viswanathan, M.
Author_Institution :
AT&T Res. Labs., Florham Park, NJ, USA
Abstract :
We give a space-efficient, one-pass algorithm for approximating the L1 difference Σi|ai-bi | between two functions, when the function values ai and bi are given as data streams, and their order is chosen by an adversary. Our main technical innovation is a method of constructing families {Vj} of limited independence random variables that are range summable by which we mean that Σj=0 c-1 Vj(s) is computable in time polylog(c), for all seeds s. These random variable families may be of interest outside our current application domain, i.e., massive data streams generated by communication networks. Our L1-difference algorithm can be viewed as a “sketching” algorithm, in the sense of (A. Broder et al., 1998), and our algorithm performs better than that of Broder et al., when used to approximate the symmetric difference of two sets with small symmetric difference
Keywords :
computational complexity; data handling; random processes; set theory; L1-difference algorithm; application domain; approximate L1-difference algorithm; communication networks; function values; limited independence random variables; massive data streams; random variable families; range summable; sketching algorithm; space-efficient one-pass algorithm; symmetric difference; technical innovation; Computerized monitoring; Instruments; Internet; Reactive power; Statistics; Technological innovation;
Conference_Titel :
Foundations of Computer Science, 1999. 40th Annual Symposium on
Conference_Location :
New York City, NY
Print_ISBN :
0-7695-0409-4
DOI :
10.1109/SFFCS.1999.814623