294878.fig.0011
Figure 11: Average error on the O and C output layers when full contexts and partial objects were given as input.