Abstract
The increasing amount of scholarly literature and the diversity of dissemination channels are challenging several fields and research communities. A continuous interplay between researchers and citizen scientists creates a vast set of possibilities to integrate hybrid, crowd-machine interaction features into crowd science projects for improving knowledge acquisition from large volumes of scientific data. This paper presents SciCrowd, an experimental crowd-powered system under development “from the ground up” to support data-driven research. The system combines automatic data indexing and crowd-based processing of data for detecting topic evolution by fostering a knowledge base of concepts, methods, and results categorized according to the particular needs of each field. We describe the prototype and discuss its main implications as a mixed-initiative approach for leveraging the analysis of academic literature. © Springer Nature Switzerland AG 2018.

Abstract

Abstract

Abstract
Complex optimization scheduling problems frequently arise in the manufacturing and transport industries, where the goal is to find a schedule that minimizes the total amount of time (or cost) required to complete all the tasks. Since it is a critical factor in many industries, it has been, historically, a target of the scientific community. Mathematically, these problems are modelled with Job Shop scheduling approaches. Benchmark results to solve them are achieved with evolutionary algorithms. However, they still present some limitations, mostly related to execution times and the difficulty to generalize to other problems. Deep Reinforcement Learning is poised to revolutionise the field of artificial intelligence. Chosen as one of the MIT breakthrough technologies, recent developments suggest that it is a technology of unlimited potential which shall play a crucial role in achieving artificial general intelligence. This paper puts forward a state-of-the-art review on Job Shop Scheduling, Evolutionary Algorithms and Deep Reinforcement Learning. It also proposes a novel architecture capable of solving Job Shop Scheduling optimization problems using Deep Reinforcement Learning. © 2020, Springer Nature Switzerland AG.

Abstract

Abstract
Visual coherence between virtual and real objects is a major issue in creating convincing augmented reality (AR) applications. To achieve this seamless integration, actual light conditions must be determined in real time to ensure that virtual objects are correctly illuminated and cast consistent shadows. In this paper, we propose a novel method to estimate daylight illumination and use this information in outdoor AR applications to render virtual objects with coherent shadows. The illumination parameters are acquired in real time from context-aware live sensor data. The method works under unprepared natural conditions. We also present a novel and rapid implementation of a state-of-the-art skylight model, from which the illumination parameters are derived. The Sun's position is calculated based on the user location and time of day, with the relative rotational differences estimated from a gyroscope, compass and accelerometer. The results illustrated that our method can generate visually credible AR scenes with consistent shadows rendered from recovered illumination.