论文信息 - Back and counter propagation aberrations

Back and counter propagation aberrations

Variations on BPN (backward propagation network) and CPN (counter propagation networks) have been presented. The BPN variations are a little quicker and the CPN variations are a little more accurate. The author suggests that these variations should be tested on some other more difficult problems. To this end, a fault-tolerant (in software) handwriting recognizer is being developed using these BPN and CPN variations, along with an adaptive resonance network. Preliminary results indicate that by using the author's equation with the BPN, a 20% decrease in learning time can be obtained, which represents a reduction from 30 to 24 hours for learning the alphabet.<<ETX>>

D. Woods | Donald Woods

[1] Stephen Grossberg,et al. A massively parallel architecture for a self-organizing neural pattern recognition machine , 1988, Comput. Vis. Graph. Image Process..

[2] Stephen Grossberg,et al. Art 2: Self-Organization Of Stable Category Recognition Codes For Analog Input Patterns , 1988, Other Conferences.

[3] R. Lippmann,et al. An introduction to computing with neural nets , 1987, IEEE ASSP Magazine.

[4] P. Werbos,et al. Beyond Regression : "New Tools for Prediction and Analysis in the Behavioral Sciences , 1974 .

[5] Thomas W. Ryan,et al. VARIATIONS ON ADAPTIVE RESONANCE. , 1987 .

[6] Bernardo A. Huberman,et al. AN IMPROVED THREE LAYER, BACK PROPAGATION ALGORITHM , 1987 .

[7] R. Hecht-Nielsen. Counterpropagation networks. , 1987, Applied optics.