Abstract | ||
---|---|---|
A new algorithm is presented for principal component anal- ysis and subspace tracking, which improves upon classical stochastic gradient based algorithms (SGA) as well as several other related algorithms that have been presented in the liter- ature. The new algorithm is based on and inherits its main properties from a continuous-time algorithm, closely related to the QR flow. It gives the same estimates as classical SGA algorithms but requires only N ·κ operations per update instead of N ·κ2 , where N is the dimension of the in- put vector and κ is the number of principal components to be estimated. A parallel version with κ parallelism (pro- cessors) and throughput N−1 and is straightforwardly de- rived. A fully parallel version, with throughput independent of the problem size ( 1 ), may be obtained at the expense of N2 additional operations. |
Year | Venue | Field |
---|---|---|
1996 | EUSIPCO | Digital signal processing,Algorithm design,Subspace topology,Parallel algorithm,Computer science,Algorithm,Throughput,Principal component analysis,Numerical linear algebra,Signal processing algorithms |
DocType | Citations | PageRank |
Conference | 0 | 0.34 |
References | Authors | |
2 | 3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Dehaene, Jeroen | 1 | 4 | 2.49 |
Marc Moonen | 2 | 377 | 46.79 |
Vandewalle, Joos | 3 | 20 | 20.18 |