Klasifikasi Kain Tenun Sumba menggunakan Jaringan Saraf Tiruan

Abstract

The evaluation results with epoch 100 have quite good classification accuracy. The correct accuracy value of the classification is 60% of the test data. In other words, the results of this classification can be said to be good. Compared with the classification accuracy at epoch 100, which is 20-30% of the test data. The model obtained is good. At epoch 400 this model has a better level of accuracy than epoch 100. At epoch 1000 the increase in recognition accuracy for test data increases by 20% so that the recognition accuracy becomes 55-60%. Based on the results of research using the backpropagation neural network algorithm, there are several different levels of accuracy, the training and validation accuracy values ​​are quite good. The researcher's suggestion is to continue this research so that it can produce a more accurate process. The training process is carried out using several epoch values, namely epoch 200, epoch 400, epoch 600, epoch 800, epoch 1000, epoch. The best accuracy obtained during training was 89.3% and validation was 82%.