Probabilistic Latent Variable Models as Nonnegative Factorizations

Figure 4

An example of a higher dimensional shift-invariant positive data decomposition. The original input is shown at the top left, the approximation by the model in (12) is shown in the top middle, and the extracted kernels and impulses are shown in the lower plots.

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

(i)