Advances in Large Language Models (LLMs) have inspired a surge of research exploring their expansion into the visual domain. While recent models exhibit promise in generating abstract captions for images and conducting natural conversations, their performance on textrich images leaves room for improvement. In this paper, we propose the Contrastive Reading Model (Cream), a novel neural architecture designed to enhance the language-image understanding capability of LLMs by capturing intricate details typically overlooked by existing methods. Cream integrates vision and auxiliary encoders, complemented by a contrastive feature alignment technique, resulting in a more effective understanding of textual information within document images. Our approach, thus, seeks to bridge the gap between vision and language understanding, paving the way for more sophisticated Document Intelligence Assistants. Rigorous evaluations across diverse tasks, such as visual question answering on document images, demonstrate the efficacy of Cream as a state-of-the-art model in the field of visual document understanding. We provide our codebase and newly-generated datasets at https://github.com/naver-ai/cream.

No abstract

This study analyzed a path through which factors influencing death anxiety in the community-dwelling elderly, assuming personal organismic factors and microsystemic factors based on the ecology model purported by Belsky (1980). Methods: This study was performed with 189 elderly people. Data were collected through a self-report questionnaire. The data were analyzed using the SPSS and AMOS programs. Results: The factors influencing death anxiety in the elderly were depression, family support, social network, and familism value, and the explanatory power of these variables was 22%. Death anxiety increased with higher depression, higher familism value, larger social network, and lower family support. Spiritual well-being and elderly discrimination experience had indirect effects on death anxiety, and these effects were mediated by depression. Conclusion: Depression, family support, social network, and familism value were found to influence death anxiety in the elderly, and the strongest effect came from depression. To reduce death anxiety in the elderly, it is important to improve their relationship with their family and friends. Moreover, support should be provided by establishing local systems, and intervention should be provided to alleviate depression.

Large, distributed collections of miniaturized, wireless electronic devices may form the basis of future systems for environmental monitoring, population surveillance, disease management and other applications that demand coverage over expansive spatial scales. In this paper, we show that wind-dispersed seeds can serve as the bio-inspiration for unusual aerial schemes to distribute components for such networks via controlled, unpowered flight across natural environments or city settings. Techniques in mechanically guided assembly of three-dimensional (3D) mesostructures provide access to miniature, 3D fliers optimized for such purposes, in processes that align with the most sophisticated production techniques for electronic, optoelectronic, microfluidic and microelectromechanical technologies. We demonstrate a range of 3D macro-, meso- and microscale fliers produced in this manner, including those that incorporate active electronic payloads. Analytical, computational and experimental studies of the aerodynamics of high-performance structures of this type establish a set of fundamental considerations in bio-inspired design, with a focus on 3D fliers that exhibit controlled rotational kinematics and low terminal velocities. Battery-free, wireless devices for atmospheric measurements provide simple examples of a wide spectrum of applications of these unusual concepts.

No abstract