Figure 4From: An incremental approach to automated protein localisationClassification using the modified SFAM. A new input vector x(t) is presented to an SFAM network which performs a separation of two classes in a two-dimensional feature space. The categories belonging to each of the classes are depicted in blue and red, respectively. In principle, x(t) would be unknown to the network, as it does not lie inside any category. However, the consideration of z ˜ min F 2 ( t ) MathType@MTEF@5@5@+=feaagaart1ev2aaatCvAUfKttLearuWrP9MDH5MBPbIqV92AaeXatLxBI9gBaebbnrfifHhDYfgasaacPC6xNi=xH8viVGI8Gi=hEeeu0xXdbba9frFj0xb9qqpG0dXdb9aspeI8k8fiI+fsY=rqGqVepae9pg0db9vqaiVgFr0xfr=xfr=xc9adbaqaaeGaciGaaiaabeqaaeqabiWaaaGcbaGafmOEaONbaGaadaqhaaWcbaGagiyBa0MaeiyAaKMaeiOBa4gabaGaemOrayKaeGOmaidaaOGaeiikaGIaemiDaqNaeiykaKcaaa@36E4@ enables the input vector to be assigned to the blue class if its distance to the next blue category is smaller than the threshold τ.Back to article page