Key points are not available for this paper at this time.
In the contemporary era, the digital revolution is fundamentally transforming our modes of living, working and thinking by optimizing processes, enabling deeper insights discovery and enhancing decision-making. The realization of this immense potential lies in the ability to extract valuable information from large datasets through machine learning (ML), thereby generating data-driven insights, informed decisions and accurate predictions. By leveraging the powerful modeling capabilities of ML, particularly in handling complex high-dimensional data, the food industry can more accurately predict or identify potential quality issues, safety risks and shifts in consumer trends. In this review, the basic principles of ML in data processing, model training and performance evaluation were introduced, followed by a comprehensive overview of ML applications across various food industry scenarios, including production optimization, origin traceability, adulteration detection, quality control, pathogen or foreign objects identification, preservation techniques, supply chain management, foods innovation and consumption trends. The types of data processing, feature extraction and model algorithms employed in these retrieved studies are systematically categorized and discussed, to assist readers in selecting appropriate algorithms for solving practical problems that may be encountered in food industry. Despite substantial progress in both theoretical foundation and practical applications of ML technique, there are still challenges in terms of data accessibility, model robustness and results interpretability. Addressing these issues is essential for fully realizing the potential benefits that ML offers to food industry. It is expected that the insights presented will contribute to the advancement of ML-based artificial intelligence technologies for smart food industry applications.
Feng et al. (Thu,) studied this question.