Name
Abstract | ||
|---|---|---|
In this paper we propose a non-negative matrix factorization (NMF) model with piecewise-constant activation coefficients. This structure is enforced using a total variation penalty on the rows of the activation matrix. The resulting optimization problem is solved with a majorization-minimization procedure. The proposed algorithm is well suited to analyze data explained by underlying piecewise-constant sequences of states. Its properties are first illustrated using synthetic data. We then use it to solve a video structuring problem that involves both segmentation and clustering tasks. An improvement over a state-of-the-art temporally smoothed NMF algorithm of both clustering and segmentation quality measures is observed. |
Year | DOI | Venue |
|---|---|---|
2014 | 10.1109/ICASSP.2014.6854901 | Acoustics, Speech and Signal Processing |
Keywords | Field | DocType |
data analysis,image segmentation,matrix decomposition,minimisation,pattern clustering,video signal processing,NMF model,activation matrix,clustering task,data analysis,majorization-minimization,piecewise constant nonnegative matrix factorization,piecewise-constant activation coefficients,piecewise-constant state sequences,resulting optimization problem,segmentation task,total variation penalty,video structuring problem,Non-negative matrix factorization,temporal smoothing,total variation | Mathematical optimization,Pattern recognition,Matrix (mathematics),Computer science,Segmentation,Matrix decomposition,Synthetic data,Artificial intelligence,Non-negative matrix factorization,Cluster analysis,Optimization problem,Piecewise | Conference |
ISSN | Citations | PageRank |
1520-6149 | 1 | 0.37 |
References | Authors | |
5 | 4 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Nicolas Seichepine | 1 | 2 | 0.77 |
| Slim Essid | 2 | 212 | 32.00 |
| Cédric Févotte | 3 | 2380 | 149.37 |
| O. Cappe | 4 | 2112 | 207.95 |