Applications of Riemannian optimization methods to signal processing
Project/Area Number |
19700229
|
Research Category |
Grant-in-Aid for Young Scientists (B)
|
Allocation Type | Single-year Grants |
Research Field |
Sensitivity informatics/Soft computing
|
Research Institution | National Institute of Advanced Industrial Science and Technology |
Principal Investigator |
NISHIMORI Yasunori National Institute of Advanced Industrial Science and Technology, 脳神経情報研究部門, 研究員 (00357724)
|
Project Period (FY) |
2007 – 2009
|
Project Status |
Completed (Fiscal Year 2009)
|
Budget Amount *help |
¥3,860,000 (Direct Cost: ¥3,200,000、Indirect Cost: ¥660,000)
Fiscal Year 2009: ¥1,170,000 (Direct Cost: ¥900,000、Indirect Cost: ¥270,000)
Fiscal Year 2008: ¥1,690,000 (Direct Cost: ¥1,300,000、Indirect Cost: ¥390,000)
Fiscal Year 2007: ¥1,000,000 (Direct Cost: ¥1,000,000)
|
Keywords | 独立成分分析 / ニューラルネットワーク / 信号処理 / 最適化 / 独立部分空間分析 / 多様体学習 / 非線形状次元縮約 / 最適化理論 / 微分幾何 |
Research Abstract |
The set of fixed number of orthogonal subspaces each of which has fixed dimensionality can be regarded as a curved space, termed the flag manifold. Using Riemannian geometry we proposed new optimization methods on this class of manifolds such as Riemannian gradient descent, conjugate gradient method and hybrid MCMC geodesic method. Based on these Riemannian optimization methods new learning algorithms for independent subspace analysis were proposed and their effectiveness over conventional algorithms was experimentally confirmed.
|
Report
(4 results)
Research Products
(14 results)