THE DEVELOPMENT OF BABY FOOD NUTRITION PREDICTION MODEL USING NEAR-INFRARED SPECTROSCOPY AND DEEP LEARNING APPROACH

• Main Author: Nur Aulia, Masyithah
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/42860
• Institution: Institut Teknologi Bandung

In Indonesia, malnutrition and overnutrition can still be found especially among children and toddlers who are supposed to get adequate nutrition. The cause of malnutrition is inadequate nutrition intake, both in terms of the quantity of the food and in terms of quality of the food. Therefore, parents especially mothers, need to know whether the food consumed by children has reached the specified nutrition intake recommendation. This final project develops a model to predict macronutrient content in baby food using Deep Learning approach such as Deep-belief network and Convolutional Neural Network. We used NIRS data of instant-porridge from SCiO to build the model. The prediction results from DBN and CNN are compared with conventional method such as Support Vector Regression, Partial Least Square and Linear Regression. In this research, CNN managed to give the best performance with error for carbohydrate, protein and fat 11.70%, 26.14% and 28.72% respectively. Generally, the error generated in most cases is still within the safe limit of excess/deficiency macronutrient for children and toddlers, but the fat prediction error is still high enough so that the excess fat of 28.72% is outside the safe limit. Therefore, in the future, further research needs to be done to improve performance by using more data to develop the model.