IMPROVING ACCURACY OF NEURAL NETWORK ALGORITHM THROUGH NOISE MINIMATION USING BAGGING FOR MODELLING PREDICTION OF EARLY AGE CONCRETE COMPRESSIVE STRENGTH Stefanus Santosa, Suroso, Marchus Budi Utomo, Martono, Mawardi
Politeknik Negeri Semarang
Artificial Neural Network (ANN) is a Machine Learning (ML) algorithm which learn by itself and organize its thinking to solve problems. Although the learning process involves many hidden layers (Deep Learning) this algorithm still has weaknesses when faced with high noise data. Concrete mixture design data has a high enough noise caused by many unidentified / measurable aspects such as planning, design, manufacture of test specimens, maintenance, testing, diversity of physical and chemical properties, mixed formulas, mixed design errors, environmental conditions, and testing process. Information needs about the compressive strength of early age concrete (under 28 days) are often needed while the construction process is still ongoing. ANN has been tried to predict the compressive strength of concrete, but the results are less than optimal. This study aims to improve the ANN prediction model using Bagging to reduce the influence of noise and overfitting. At the beginning this study based on an ANN multi-layer feed-forward that is trained by stochastic gradient descent using back-propagation. The results show that the number of hidden layers and neurons does not guarantee to reduced error. The Deep Learning paradigm, which is characterized by the depth of learning shown by the number of hidden layers, is not always parallel with the best results. Models with 12 hidden layers and 50 neurons still have higher errors. The best model is achieved by 6 hidden layer and 50 neurons with a RMSE value of 7,418 +/- 1,214. The use of different activation functions also does not significantly affect the performance of the model. Model improvements are carried out using Baging. The best model is obtained by 7 hidden layers and 50 neurons with RMSE 6,140 + / - 0.556. This result proves that Bagging is significant to reduce the Deep Learning error rate for predicting the compressive strength of early age concrete. This Machine Learning model can be used as an alternative / substitute for conventional mix designs, which are environmentally friendly, economical, and accurate. Future work with regard to the concrete industry, this model can be applied to create an intelligent Batching and Mixing Plants.
Keywords: Keywords: deep learning, ANN, bagging, concrete compressive strength
Topic: Electrical Engineering and Computer Science