Current diets of most nations either do not meet the nutrition recommendations or transgress environmental planetary boundaries or both. Transitioning toward sustainable diets that are nutritionally adequate and low in environmental impact is key in achieving the United Nations' Sustainable Development Goals. However, designing region-specific sustainable diets that are culturally acceptable is a formidable challenge. Recent studies have suggested that optimization algorithms offer a potential solution to the above challenge, but the evidence is mostly based on case studies from high-income nations using widely varying constraints and algorithms. Here, we employ nonlinear optimization modeling with a consistent study design to identify diets for 152 countries that meet four cultural acceptability constraints, five food-related per capita environmental planetary boundaries (carbon emissions, water, land, nitrogen, and phosphorus use), and the daily recommended levels for 29 nutrients. The results show that a considerable departure from current dietary behavior is required for all countries. The required changes in intake amounts of 221 food items are highly country-specific but in general point toward a need to reduce the intake of meat, dairy, rice, and sugar and an increase in fruits, vegetables, pulses, nuts, and other grains. The constraints for fiber, vitamin B12, vitamin E, and saturated fats and the planetary boundaries for carbon emissions and nitrogen application were the most difficult to meet, suggesting the need to pay special attention to them. The analysis demonstrates that nonlinear optimization is a powerful tool to design diets achieving multiple objectives.

The current expansion of theory and research on artificial intelligence in management and organization studies has revitalized the theory and research on decision-making in organizations. In particular, recent advances in deep learning (DL) algorithms promise benefits for decision-making within organizations, such as assisting employees with information processing, thereby augment their analytical capabilities and perhaps help their transition to more creative work. We conceptualize the decision-making process in organizations augmented with DL algorithm outcomes (such as predictions or robust patterns from unstructured data) as deep learning-augmented decision-making (DLADM). We contribute to the understanding and application of DL for decisionmaking in organizations by (a) providing an accessible tutorial on DL algorithms and (b) illustrating DLADM with two case studies drawing on image recognition and sentiment analysis tasks performed on datasets from Zalando, a European e-commerce firm, and Rotten Tomatoes, a review aggregation website for movies, respectively. Finally, promises and challenges of DLADM as well as recommendations for managers in attending to these challenges are also discussed.

No abstract

Matrix factorization techniques have been widely used as a method for collaborative filtering for recommender systems. In recent times, different variants of deep learning algorithms have been explored in this setting to improve the task of making a personalized recommendation with user-item interaction data. The idea that the mapping between the latent user or item factors and the original features is highly nonlinear suggest that classical matrix factorization techniques are no longer sufficient. In this paper, we propose a multilayer nonlinear semi-nonnegative matrix factorization method, with the motivation that user-item interactions can be modeled more accurately using a linear combination of non-linear item features. Firstly, we learn latent factors for representations of users and items from the designed multilayer nonlinear Semi-NMF approach using explicit ratings. Secondly, the architecture built is compared with deep-learning algorithms like Restricted Boltzmann Machine and state-of-the-art Deep Matrix factorization techniques. By using both supervised rate prediction task and unsupervised clustering in latent item space, we demonstrate that our proposed approach achieves better generalization ability in prediction as well as comparable representation ability as deep matrix factorization in the clustering task.

No abstract