Search

Human pose estimation is viewed as a crucial step for understanding human behaviour. Although significant progress has been made in this area in recent years, most studies have focused on feature-level information fusion, while decision-level information fusion has rarely been explored. Compared with feature-level information, decision-level information contains more semantic and interpretable information and can help improve the performance of pose estimation in occluded and crowded scenes. In this paper, we focus on the fusion of decision-level information. We propose a View Fusion module for aggregating decision-level information from different stages to generate a more comprehensive estimation. An Auxiliary Task module is introduced to bridge the gap between the feature extractor and the View Fusion module and to provide prior information about the form of the decision-level information. Considering that the precision of predictions from different stages varies, we use different strategies to guide the learning process. Experiments show that our models outperform previous methods and achieve competitive results on the CrowdPose test set. Further experiments indicate that our method is flexible and can improve the performance of various backbones. [ABSTRACT FROM AUTHOR]

Despite the increasing use of the self-regulated learning process in the smart learning environment, understanding the concepts from a theoretical perspective and empirical evidence are limited. This study used a systematic review to explore models, design tools, support approaches, and empirical research on the self-regulated learning process in the smart learning environment. This review revealed that there is an increasing body of literature from 2012 to 2020. The analysis shows that self-regulated learning is a critical factor influencing a smart learning environment's learning process. The self-regulated learning components, including motivation, cognitive, metacognitive, self-efficiency, and metacognitive components, are most cited in the literature. Furthermore, self-regulated strategies such as goal setting, helping-seeking, time management, and self-evaluation have been founded to be frequently supported in the literature. Besides, limited theoretical models are designed to support the self-regulated learning process in a smart learning environment. Furthermore, most evaluations of the self-regulated learning process in smart learning environment are quantitative methods with limited mixed methods. The design tools such as visualization, learning agent, social comparison, and recommendation are frequently used to motivate students' learning engagement and performance. Finally, the paper presents our conclusion and future directions supporting the self-regulated learning process in the smart learning environment. [ABSTRACT FROM AUTHOR]

Extractive approaches have been the mainstream paradigm for identifying overlapping entity–relation extraction. However, limited by their inherently methodological flaws, which hardly deal with three issues: hierarchical dependent entity–relations, implicit entity–relations, and entity normalization. Recent advances have proposed an effective solution based on generative language models, which cast entity–relation extraction as a sequence-to-sequence text generation task. Inspired by the observation that humans learn by getting to the bottom of things, we propose a novel framework, namely GenRE, Generative multi-turn question answering with contrastive learning for entity–relation extraction. Specifically, a template-based question prompt generation first is designed to answer in different turns. We then formulate entity–relation extraction as a generative question answering task based on the general language model instead of span-based machine reading comprehension. Meanwhile, the contrastive learning strategy in fine-tuning is introduced to add negative samples to mitigate the exposure bias inherent in generative models. Our extensive experiments demonstrate that GenRE performs competitively on two public datasets and a custom dataset, highlighting its superiority in entity normalization and implicit entity–relation extraction. (The code is available at https://github.com/lovelyllwang/GenRE). [ABSTRACT FROM AUTHOR]

To navigate our complex social world, it is crucial to deploy multiple learning strategies, such as learning from directly experiencing action outcomes or from observing other people's behavior. Despite the prevalence of experiential and observational learning in humans and other social animals, it remains unclear how people favor one strategy over the other depending on the environment, and how individuals vary in their strategy use. Here, we describe an arbitration mechanism in which the prediction errors associated with each learning strategy influence their weight over behavior. We designed an online behavioral task to test our computational model, and found that while a substantial proportion of participants relied on the proposed arbitration mechanism, there was some meaningful heterogeneity in how people solved this task. Four other groups were identified: those who used a fixed mixture between the two strategies, those who relied on a single strategy and non-learners with irrelevant strategies. Furthermore, groups were found to differ on key behavioral signatures, and on transdiagnostic symptom dimensions, in particular autism traits and anxiety. Together, these results demonstrate how large heterogeneous datasets and computational methods can be leveraged to better characterize individual differences. People learn by observing others and from personal experience, but which strategy they favor varies. Here, the authors show that while some people dynamically arbitrate and switch to the strategy that is most reliable, others preferentially use one or a fixed mixture of strategies. [ABSTRACT FROM AUTHOR]

The study of social learning examines how individuals learn from others by means of observation, imitation, or compliance with advice. However, it still remains largely unknown whether social learning processes have a distinct contribution to behavior, independent from non-social trial-and-error learning that often occurs simultaneously. 153 participants completed a reinforcement learning task, where they were asked to make choices to gain rewards. Advice from an artificial teacher was presented in 60% of the trials, allowing us to compare choice behavior with and without advice. Results showed a strong and reliable tendency to follow advice (test-retest reliability ~0.73). Computational modeling suggested a unique contribution of three distinct learning strategies: (a) individual learning (i.e., learning the value of actions, independent of advice), (b) informed advice-taking (i.e., learning the value of following advice), and (c) non-informed advice-taking (i.e., a constant bias to follow advice regardless of outcome history). Comparing artificial and empirical data provided specific behavioral regression signatures to both informed and non-informed advice taking processes. We discuss the theoretical implications of integrating internal and external information during the learning process. [ABSTRACT FROM AUTHOR]

Fostering students' competence in applying interdisciplinary knowledge to solve problems has been recognized as an important and challenging issue globally. This is why STEM (Science, Technology, Engineering, Mathematics) education has been emphasized at all levels in schools. Meanwhile, the use of robotics has played an important role in STEM learning design. The purpose of this study was to fill a gap in the current review of research on Robotics-based STEM (R-STEM) education by systematically reviewing existing research in this area. This systematic review examined the role of robotics and research trends in STEM education. A total of 39 articles published between 2012 and 2021 were analyzed. The review indicated that R-STEM education studies were mostly conducted in the United States and mainly in K-12 schools. Learner and teacher perceptions were the most popular research focus in these studies which applied robots. LEGO was the most used tool to accomplish the learning objectives. In terms of application, Technology (programming) was the predominant robotics-based STEM discipline in the R-STEM studies. Moreover, project-based learning (PBL) was the most frequently employed learning strategy in robotics-related STEM research. In addition, STEM learning and transferable skills were the most popular educational goals when applying robotics. Based on the findings, several implications and recommendations to researchers and practitioners are proposed. [ABSTRACT FROM AUTHOR]

The goal of ITS is to support learning content, activities, and resources, adapted to the specific needs of the individual learner and influenced by learner's motivation. One of the major challenges to the mainstream adoption of adaptive learning is the complexity and time involved in guiding the learning process. To tackle these problems, this research proposes an intentional model that adopts Map formalism to support personalized learning guidance by considering learner's motivation. For that, the intentional model adopts the ARCS model of motivation. The proposed model couples the learner's intention with the learning strategies and provides a multitude of paths between learner intentions. Based on the learner's motivation and the corresponding learning mode or the individual learning style, the ITS can adaptively support the learner to achieve his/her intention through the selected strategy. The first results show that this study's provide implications for both theoretical research and the practical development of ITSs that will help education professionals make full use of ITSs. In the initial tests, the process model has met what was expected, however, more studies with experiments must be carried out. [ABSTRACT FROM AUTHOR]