Generating an image from a given text description has two goals: visual realism and semantic consistency. Although significant progress has been made in generating high-quality and visually realistic images using generative adversarial networks, guaranteeing semantic consistency between the text description and visual content remains very challenging. In this paper, we address this problem by proposing a novel global-local attentive and semantic-preserving text-to-image-to-text framework called MirrorGAN. MirrorGAN exploits the idea of learning textto-image generation by redescription and consists of three modules: a semantic text embedding module (STEM), a global-local collaborative attentive module for cascaded image generation (GLAM), and a semantic text regeneration and alignment module (STREAM). STEM generates word-and sentence-level embeddings. GLAM has a cascaded architecture for generating target images from coarse to fine scales, leveraging both local word attention and global sentence attention to progressively enhance the diversity and semantic consistency of the generated images. STREAM seeks to regenerate the text description from the generated image, which semantically aligns with the given text description. Thorough experiments on two public benchmark datasets demonstrate the superiority of Mirror-GAN over other representative state-of-the-art methods.

Math anxiety (MA) has been suggested to decrease the math performance of students. However, it remains unclear what factors moderate this relationship. The aim of this research was to explore the link between MA and math performance. Studies that explored the math anxiety-performance link, conducted from 2000 to 2019 (84 samples, N = 8680), were identified and statistically integrated with a meta-analysis method. The results indicated a robust negative math anxiety-performance link. Furthermore, regarding the analysis of moderator variables, this negative link was stronger in the studies that involved Asian students, but this link was the weakest in the studies that involved European students. Moreover, this negative link was stronger in the studies within a senior high school group, whereas it was the weakest in the studies within an elementary group. Finally, this negative link was strongest among studies that used a custom test and studies that assessed problem-solving skills. Potential explanations and implications for research and practice are discussed.

No abstract

Purpose Despite manager’s investments in facilitating knowledge sharing, such as hiring employees with lots of knowledge, knowledge hiding remains prevalent in organizations. It may stem from that less attention has been paid to the relationship between perceived overqualification and knowledge hiding. Drawing on emotion theory, this study aims to build a mediation framework to examine effects of perceived overqualification on knowledge hiding via negative emotion state and moderating role of team positive affective tone. Design/methodology/approach The paper uses a two-wave survey study among 398 knowledge workers from 106 teams in knowledge-intensive industries and tests the hypotheses by performing a series of hierarchical linear modeling analyzes. Findings The results show that a negative emotion state mediates the U-shaped relationship between employees’ perceived overqualification and knowledge hiding behavior. Team positive affective tone moderates the U-shaped relationship between negative emotions and employees’ knowledge hiding behavior. Originality/value This study extends current knowledge management literature by introducing perceived overqualification as an individual predictor of employees’ knowledge hiding behavior and revealing the both light and dark sides of perceived overqualification on knowledge hiding, as well as its intervening mechanism. The research findings help practitioners to curb such counterproductive behaviors.

No abstract