PENERAPAN ELMAN RECURRENT NEURAL NETWORK UNTUK DIAGNOSIS GANGGUAN AUTIS PADA ANAK

• المؤلفون الرئيسيون: , YULI NOORVIANI,ST, , Drs.Widodo Prijodiprodjo, M.Sc., EE.
• التنسيق: Theses and Dissertations NonPeerReviewed
• منشور في: [Yogyakarta] : Universitas Gadjah Mada 2012
• الموضوعات: ETD
• الوصول للمادة أونلاين: https://repository.ugm.ac.id/97414/; http://etd.ugm.ac.id/index.php?mod=penelitian_detail&sub=PenelitianDetail&act=view&typ=html&buku_id=54275
• المؤسسة: Universitas Gadjah Mada

Current developments in information technology has penetrated all sectors of life. Many problems in various sectors that require ease, speed and accuracy can be solved with the help of information technology. In the field of psychology, psychologists to help diagnose developmental disorders in children. This study makes an artificial neural network applications to be able to diagnose autistic disorder in children with Elman Recurrent Neural Network. This application is created as a tool for diagnosing autistic disorder based on physical symptoms suffered by the child. Artificial neural network method used is the method of Elman Recurrent Neural Network, which is an unsupervised learning. This software is created using the Matlab programming language with MySQL database. The symptoms of autistic disorder are used as input for the diagnosis consists of 48 variables. This study uses 75 units of neurons in the hidden layer with the assumption that the number of these errors reached the optimum (minimum error). This configuration produces MSE 2.96 e-21 and the iteration process runtime 510 with 161.06 seconds and the learning rate is 0.025 Testing the neural network is working well, which reached 80.83% testing accuracy. These results show that the network has been recognized by both the pattern that has been drilled, although there are some data that do not fit with the target.