Amazon.com

About: Amazon.com is a company organization based out in Seattle, Washington, United States. It is known for research contribution in the topics: Service (business) & Service provider. The organization has 13363 authors who have published 17317 publications receiving 266589 citations.

Mapping 123 million neonatal, infant and child deaths between 2000 and 2017

Abstract: Since 2000, many countries have achieved considerable success in improving child survival, but localized progress remains unclear. To inform efforts towards United Nations Sustainable Development Goal 3.2—to end preventable child deaths by 2030—we need consistently estimated data at the subnational level regarding child mortality rates and trends. Here we quantified, for the period 2000–2017, the subnational variation in mortality rates and number of deaths of neonates, infants and children under 5 years of age within 99 low- and middle-income countries using a geostatistical survival model. We estimated that 32% of children under 5 in these countries lived in districts that had attained rates of 25 or fewer child deaths per 1,000 live births by 2017, and that 58% of child deaths between 2000 and 2017 in these countries could have been averted in the absence of geographical inequality. This study enables the identification of high-mortality clusters, patterns of progress and geographical inequalities to inform appropriate investments and implementations that will help to improve the health of all populations.

Transfer Learning in Deep Reinforcement Learning: A Survey

Abstract: Reinforcement Learning (RL) is a key technique to address sequential decision-making problems and is crucial to realize advanced artificial intelligence. Recent years have witnessed remarkable progress in RL by virtue of the fast development of deep neural networks. Along with the promising prospects of RL in numerous domains, such as robotics and game-playing, transfer learning has arisen as an important technique to tackle various challenges faced by RL, by transferring knowledge from external expertise to accelerate the learning process. In this survey, we systematically investigate the recent progress of transfer learning approaches in the context of deep reinforcement learning. Specifically, we provide a framework for categorizing the state-of-the-art transfer learning approaches, under which we analyze their goals, methodologies, compatible RL backbones, and practical applications. We also draw connections between transfer learning and other relevant topics from the RL perspective and explore their potential challenges as well as open questions that await future research progress.

Recommendation system with multiple integrated recommenders

Abstract: A recommendations system is provided in various embodiments for selecting items to recommend to a user. The system includes a recommendation engine with a plurality of recommenders, and each recommender identifies a different type of reason for recommending items. In one embodiment, each recommender retrieves item preference data and generates candidate recommendations responsive to a subset of that data. The recommenders also score the candidate recommendations. In certain embodiments, a normalization engine normalizes the scores of the candidate recommendations provided by each recommender. A candidate selector selects at least a portion of the candidate recommendations based on the normalized scores to provide as recommendations to the user. The candidate selector also outputs the recommendations with associated reasons for recommending the items.

Few-Shot Learning With Embedded Class Models and Shot-Free Meta Training

Abstract: We propose a method for learning embeddings for few-shot learning that is suitable for use with any number of shots (shot-free). Rather than fixing the class prototypes to be the Euclidean average of sample embeddings, we allow them to live in a higher-dimensional space (embedded class models) and learn the prototypes along with the model parameters. The class representation function is defined implicitly, which allows us to deal with a variable number of shots per class with a simple constant-size architecture. The class embedding encompasses metric learning, that facilitates adding new classes without crowding the class representation space. Despite being general and not tuned to the benchmark, our approach achieves state-of-the-art performance on the standard few-shot benchmark datasets.