In this paper, we introduce Recipe1M, a new large-scale, structured corpus of over 1m cooking recipes and 800k food images. As the largest publicly available collection of recipe data, Recipe1M affords the ability to train high-capacity models on aligned, multi-modal data. Using these data, we train a neural network to find a joint embedding of recipes and images that yields impressive results on an image-recipe retrieval task. Additionally, we demonstrate that regularization via the addition of a high-level classification objective both improves retrieval performance to rival that of humans and enables semantic vector arithmetic. We postulate that these embeddings will provide a basis for further exploration of the Recipe1M dataset and food and cooking in general. Code, data and models are publicly available 1. * contributed equally. 1 http://im2recipe.csail.mit.edu

In this paper, we introduce Recipe1M + , a new large-scale, structured corpus of over one million cooking recipes and 13 million food images. As the largest publicly available collection of recipe data, Recipe1M + affords the ability to train high-capacity models on aligned, multimodal data. Using these data, we train a neural network to learn a joint embedding of recipes and images that yields impressive results on an image-recipe retrieval task. Moreover, we demonstrate that regularization via the addition of a high-level classification objective both improves retrieval performance to rival that of humans and enables semantic vector arithmetic. We postulate that these embeddings will provide a basis for further exploration of the Recipe1M + dataset and food and cooking in general. Code, data and models are publicly available.

No abstract

Software learnability, a central aspect of usability, can be improved by including a tutorial. Existing research suggests that interactive, gamified tutorials are more effective than passive ones in increasing user engagement and learning. There is disagreement, however, as to whether the tutor should present a reduced or full interface of the tutored program. To determine if either approach offers an advantage, we have created an interactive tutor for the Git revision control system that uses the user's own installation of the program for maximal realism. We intend to compare this tutor to one that uses a simpler, but less accurate simulation of Git. In a pilot study, participants, despite having challenges with secondary programs used by the real Git, remained engaged and were able to discover and learn commands not available in the simplified interface. These findings suggest that designing for authenticity in interactive tutorials may be beneficial.