IMPLEMENTASI ALGORITMA PELATIHAN LEVENBERG MARQUARDT DAN BAYES REGULARISASI PADA JARINGAN SYARAF TIRUAN UNTUK PREDIKSI CURAH HUJAN (Studi Kasus : Stasiun pengamatan curah hujan Pangsuma Putussibau, Kalimantan Barat)

• Main Authors: , Yasinta Lisa, , Drs. Retantyo Wardoyo, M.Sc., Ph.D
• Format: Theses and Dissertations NonPeerReviewed
• Published: [Yogyakarta] : Universitas Gadjah Mada 2012
• Subjects: ETD
• Online Access: https://repository.ugm.ac.id/101013/; http://etd.ugm.ac.id/index.php?mod=penelitian_detail&sub=PenelitianDetail&act=view&typ=html&buku_id=57670
• Institution: Universitas Gadjah Mada

Levenberg Marquardt algorithm is used for training feedforward neural networks because of the effectiveness and convergence acceleration. Levenberg Marquardt is nonliear optimization method that used for backpropagation to find adjusted weights. Regularization can improve the performance of the neural network generalization. Regularization technique that often used is the regularization bayes. This study was conducted to determine the performance of Levenberg Marquardt training neural networks method with the addition of regularization for time series data forecasting, compared with Levenberg Marquardt algorithm without regularization. The data used in this study were monthly data of rainfall, average temperature, humidity, air pressure above the observation stations and the average wind speed at the observation station Pangsuma Putussibau West Kalimantan in 2008-2009. Training data was done on a neural network in a single hidden layer with three neurons and one neuron in the output layer. The network architecture was determined by trial and error and shown the best training MSE generated. In this case, the study results showed that Levenberg Marquardt algorithm with regularization was better used.