Матеріали конференцій
Постійне посилання колекціїhttps://dspace.nuft.edu.ua/handle/123456789/7498
Переглянути
1 результатів
Результати пошуку
Документ Prediction of allergenic composition in food products using natural language processing based on LSTM(2024) Krysanov, Denys; Melnyk, OksanaThe growing global prevalence of food allergies, with symptoms ranging from mild to severe, is a challenge for scientists, healthcare professionals, food manufacturers and legislators. Regulations, including Ukraine’s Law № 2639-VIII and EU Regulation (EU) № 1169/2011, govern allergen labeling. This study presents an innovative application of Long Short-Term Memory (LSTM) neural networks for predicting allergens in food products based on textual descriptions of their compositions. The research leverages a dataset sourced from Kaggle, which includes names and detailed descriptions of food items, and processes it through an LSTM-based neural network architecture. The architecture consists of an embedding layer, an LSTM layer, and dense layers with a Dropout layer included to prevent overfitting. The model is trained using a binary cross-entropy loss function to evaluate its performance, eventually achieving an accuracy of 98.75% with a low loss of 0.0414. These results underscore the potential of LSTM networks in the field of food safety, offering a robust tool for allergen identification that could assist consumers in making safer dietary choices and managing food allergies more effectively. The findings highlight the growing importance of artificial intelligence and machine learning in enhancing public health measures and the management of food allergies.