순환 연상 메모리를 이용한 훼손된 필기체 숫자 복원 = Restore of Damaged Handwritten Digits using Circulation Associative Memory

다국어 초록 (Multilingual Abstract)

Formerly, various classical filtering method was suggested to reconstruct damaged images, however, it premised the matrix about WSS restricted condition and original reflection, and asked the complexity or calculation.
Due to these problems, image restoration method, that uses the neural network model, recently started to be developed, a method that minimizes the energy of neural network as renewing the value of neuron repeatedly was suggested after it basically matched an error function of the image restoration problem with the energy function of neural network model.
In order to reconstruct damaged images, many researchs are being alone for using an artificial neural network.
In this monographs, using an artificial neural network, another model of neural network, suggests algorithm that can infer the original of this pattern with parts of damaged images and parts of images, and would like to apply to handwritten digits.