外文摘要：This study focuses on predicting and optimizing the quality parameters of cookies enriched with dehydrated peach through the application of Support Vector Machine (SVM) and Artificial Neural Network (ANN) models. The purpose of the study is to employ advanced machine learning techniques to understand the intricate relationships between input parameters, such as the presence of dehydrated peach and treatment methods (lyophilization and lyophilization with osmotic pretreatment), and output variables representing various quality aspects of cookies. For each of the 32 outputs, including the parameters of the basic chemical compositions of the cookie samples, selected mineral contents, moisture contents, baking characteristics, color properties, sensorial attributes, and antioxidant properties, separate models were constructed using SVMs and ANNs. Results showcase the efficiency of ANN models in predicting a diverse set of quality parameters with r2 up to 1.000, with SVM models exhibiting slightly higher coefficients of determination for specific variables with r2 reaching 0.981. The sensitivity analysis underscores the pivotal role of dehydrated peach and the positive influence of osmotic pretreatment on specific compositional attributes. Utilizing established Artificial Neural Network models, multi-objective optimization was conducted, revealing optimal formulation and factor values in cookie quality optimization. The optimal quantity of lyophilized peach with osmotic pretreatment for the cookie formulation was identified as 15%.
外文关键词：support vector machine;artificial neural network;modeling data;peach addition;cookie quality
作者：Pezo, Lato;Loncar, Biljana;Knezevic, Violeta;Nicetin, Milica;Filipovic, Jelena;Petkovic, Marko;Filipovic, Vladimir
作者单位：Inst Gen & Phys Chem;Univ Novi Sad;Univ Kragujevac
期刊名称：FOODS
期刊影响因子：0.0
出版年份：2024
出版刊次：13（5）
原文传递申请：江苏省科技资源（工程技术文献）统筹服务平台