Similarity Learning and Generalization with Limited Data: A Reservoir Computing Approach

Figure 12

Reservoir activity for the dual reservoir architecture. (a), (b), (c), (d), and (e) show the differential reservoir activity of nodes over timesteps for input relationships noise, rotated, zoomed, blurred, and different respectively. (f) shows the output weight matrix() for 50 reservoir nodes. (g) shows activity of a random node for all output labels over timesteps. : 1000; ; sparsity.

(a)

(b)

(c)

(d)

(e)

(f)

(g)