Abstract | ||
---|---|---|
An adaptive on-line algorithm extending the learning of learningidea is proposed and theoretically motivated. Relying only on gradientflow information it can be applied to learning continuousfunctions or distributions, even when no explicit loss function is givenand the Hessian is not available. Its efficiency is demonstratedfor a non-stationary blind separation task of acoustic signals.1 IntroductionNeural networks provide powerful tools to capture the structure in data by... |
Year | Venue | DocType |
---|---|---|
1996 | NIPS | Conference |
Citations | PageRank | References |
10 | 9.26 | 1 |
Authors | ||
3 |
Name | Order | Citations | PageRank |
---|---|---|---|
Ziehe, Andreas | 1 | 617 | 72.50 |
Klaus-Robert Müller | 2 | 12756 | 1615.17 |
Noboru Murata | 3 | 855 | 170.36 |