Figure 3- - uploaded by Eric Granger
Content may be subject to copyright.
Handwriting sample from for NIST SD19 (a) Handwritten sample form. (b) Images of extracted digits.

Handwriting sample from for NIST SD19 (a) Handwritten sample form. (b) Images of extracted digits.

Source publication
Article
Full-text available
Training fuzzy ARTMAP neural networks for classification using data from com-plex real-world environments may lead to category proliferation, and yield poor performance. This problem is known to occur whenever the training set contains noisy and overlapping data. Moreover, when the training set contains identical input patterns that belong to diffe...

Contexts in source publication

Context 1
... have been selected to facilitate the observation of fuzzy ARTMAP be- havior on different tractable problems. Of the four sets, two have simple linear decision boundaries with overlapping class distributions, D µ (ξ tot ) and D σ (ξ tot ), and two have com- plex non-linear decision boundaries without overlap, D CIS and D P2 . The total theoretical Fig. 3. Representation of the synthetic data sets used for computer ...
Context 2
... represented in Figure 3(a), this data consists of two classes, each one defined by a multivariate normal distribution in a two dimensional input feature space. It is assumed that data is randomly generated by sources with the same Gaussian noise. ...
Context 3
... degree of overlap is varied from a total probability of error, ξ tot = 1% to ξ tot = 25%, with 2% increments, by adjusting the mean vector µ 2 of class 2. D σ (ξ tot ): As represented in Figure 3(b), this data is identical to D µ (ξ tot ), except that the degree of overlap between classes is varied by adjusting the variance σ 2 2 of both classes. Note that for a same degree of overlap, D σ (ξ tot ) data sets have a larger overlap boundary than D µ (ξ tot ) yet they are not as dense. ...
Context 4
... that for a same degree of overlap, D σ (ξ tot ) data sets have a larger overlap boundary than D µ (ξ tot ) yet they are not as dense. D CIS : As represented in Figure 3(c), the Circle-in-Square problem 6 requires a classifier to identify the points of a square that lie inside a circle, and those that lie outside a cir- cle. The circle's area equals half of the square. ...
Context 5
... consists of one non-linear decision boundary where classes do not overlap. D P2 : As represented in Figure 3(d), each decision region of the D P2 problem is delimited by one or more of the four following polynomial and trigonometric functions: ...