ABSTRAKSI: Artificial Neural Network (ANN) merupakan salah satu metode yang baik untuk Patern Recognition (pengenalan pola), berdasarkan arsitekturnya Multi Layer Perceptron (MLP) merupakan ANN yang mempunyai layer hidden. Sedangkan Firefly Algorithm (FA) adalah salah satu algoritma optimasi yang terinspirasi dari perilaku kawanan kunang-kunang di alam, dalam implementasinya FA mampu mencari ruang solusi secara efektif.
Dalam tugas akhir ini FA digunakan untuk algoritma learning MLP dalam Patern Recognition. Ada tiga skenario yang akan di uji dalam permasalahan patern recognition yaitu skenario 1 ( 60% data training, 20% data validation, dan 20% data testing), skenario 2 (33,34% data training, 33,34% data validation, dan 33,34% data testing), dan skenario 3 (30% data training, 30% data validation, dan 40% data testing).
Hasil pengujian menunjukan FA bisa menjadi algoritma learning yang baik bagi MLP, dari tiga skenario pengujian didapat bahwa rata-rata hasil akurasi yang didapatkan diatas 90 %.Kata Kunci : Multi layer perceptron (MLP), Firefly algorithm (FA), Learning, Patern RecognitionABSTRACT: Artificial Neural Network (ANN) is one good method for Patern Recognition, based on a Multi-Layer Perceptron architecture (MLP) is an ANN which has a hidden layer. While the Firefly Algorithm (FA) is one of optimization algorithms inspired by the behavior of fireflies in nature, the FA implementation capable of searching the solution space effectively.
In this thesis the FA used for MLP learning algorithm in Patern Recognition. There are three scenarios that will test in Patern Recognition problems namely scenario 1 (60% training data, 20% validation data, and 20% testing data), scenario 2 (33.34% training data, 33.34% validation data, and 33, 34% testing data), and scenario 3 (30% training data, 30% validation data, and 40% testing data).
Test results showed the FA could be a good learning algorithm for MLP, of the three test scenarios in the can that the average accuracy of the results obtained above 90%.Keyword: Multi-layer perceptron (MLP), Firefly Algorithm (FA), Learning, Patern Recognition