<![CDATA[Autoassociative Memory (AM) is one of the artificial neural network methods and is a special of Heteroassociative Memory network. AM saving one or more vectors using matrics by the training process. Vectors that have been stored can be recognized by computing the weight matrix and in the activation process this method can be done very easily because it does not require calculation of the target value like other artificial neural network methods. The target is identical to the input so that if the input is identical to the output then the input is a pattern is a recognized object. AM ability to recognize patterns can be used for User and Password Authentication in an information system, so as to increase security in applications. The application of AM on a website-based application can reduce direct interaction between the database on the system and users who are not authorized, because the authentication process only involves a weight matrix. To apply the AM method, combine user and password then converted into binary to making a vector to using AM processed. In this research, AM ability to recognize user and password is 88% with a total of 100 data and is able to recognize users and passwords by 60% with an entropy value of 0.29, it can be concluded that the AM method can be used in the authentication process but has not optimal because it has not correct 100%.]]>
Copyrights © 2022
