KOMBINASI ALGORITMA JARINGAN SYARAF TIRUAN LEARNING VECTOR QUANTIZATION (LVQ) DAN SELF ORGANIZING KOHONEN PADA KECEPATAN PENGENALAN POLA TANDA TANGAN
Abstract: Signature is a
special form of handwriting that contain special characters and additional
forms are often used as proof of a person's identity verification. Partially
legible signature, but many signatures that can not be read. However, a
signature can be handled as an image so that it can be recognized using pattern
recognition applications in image processing. Because the signature is the
primary mechanism for authentication and authorization in legal transactions,
the need for research on the development of recognition applications and
automatic signature verification and efficiently increases from year to year.
The method is widely used in signature recognition is a method of artificial
neural network. On artificial neural networks are learning and recognition. One
neural network algorithm is Learning Vector Quantization ( LVQ ) and Self
Organizing Kohonen. Processes that occur in the neural network method requires
a relatively long time. It is influenced by the number of data samples are used
as a means of weight training update. The more and the large size of the pattern
being trained, the longer the time it takes the network. LVQ is a method of
training the unsupervised competitive layer will automatically learn to
classify input vectors into certain classes. The classes are generated depends
on the distance between the input vectors. If there are 2 input vectors are
nearly as competitive layer will then classify both the input vectors into the
same class. Kohonen Self Organizing Network is one of the neural network model
which uses learning methods or unguided unsupervised neural network model that
resembles humans. To speed up the computing process in the training and
recognition is then developed an algorithm and a combination of LVQ and Self
Organizing Kohonen by modifying the weight given to obtain ashorter time
in the process of training andrecognition.
Penulis: Emnita Ginting,
Muhammad Zarlis, Zakarias Situmorang
Kode Jurnal: jptinformatikadd140215
