现代电子技术2025，Vol.48Issue(10)：173-178,6.DOI:10.16652/j.issn.1004-373x.2025.10.027
一种融合神经与遗传的食物推荐算法
Food recommendation algorithm fusing neural and genetic information
摘要
Abstract
People pay more and more attention to the nutrition and balance of their diet,the demand for food choices is also higher.In allusion to the problems of lack of nutrient balance,lack of diversity and time-consuming formulation of existing recipes,a double population neural network-genetic algorithm(DDNT-GA)algorithm is constructed by fusing neural networks and combining dual-population genetic algorithm NSGA-Ⅱ to generate specific recipes.In this algorithm,the neural network is used to lower the fitness of overly fit individuals to effectively prevent falling into local optima.Individuals with low fitness are removed to form an elite strategy,screen out the most suitable individuals,and improve the efficiency of the model while achieving food nutrition balance.By optimizing the neural network and introducing the regularization Dropout strategy,the training speed is improved.By using the improved NSGA-Ⅱ genetic algorithm and incorporating the dual-population idea,the taboo search algorithm is used in the sub-population to prevent the generation of similar recipes by means of the taboo list,so as to realize the recipe diversification.The experimental results show that,in comparison with deep genetic algorithms(GA-D,BP-GA,NT-GA,JANUS),DDNT-GA algorithm can increase the fitness by 11.3%and shorten the training time.The resulting recipe not only has diverse changes in food combinations,but also improves the efficiency of selecting recipes,and has certain practical value in consumer recipe formulation.关键词
食谱推荐/食物选择/双种群遗传算法/神经网络/多目标优化/禁忌搜索算法Key words
recipe recommendation/food selection/dual-population genetic algorithm/neural network/multi-objective optimization/taboo search algorithm分类
电子信息工程引用本文复制引用
王客程,李然,吴江,范利利,王宁..一种融合神经与遗传的食物推荐算法[J].现代电子技术,2025,48(10):173-178,6.基金项目
辽宁省教育厅科研项目(LJKZ0730) （LJKZ0730）
中国医药教育协会2022重大科学攻关问题和医药技术难题重点课题(2022KTM036) （2022KTM036）
