Plastic persists long after it is discarded. Along many coastlines, heat and abrasion fuse it with sand, rock, and organic matter into plastiglomerates, hybrid formations that register human activity at geological timescales. PlaSphere is an immersive installation. It combines a spatialized multichannel sound field with Pepper’s Ghost “holograms” of AI-generated coastal organisms, displayed alongside specimens collected on the island the recordings come from. Field recordings from Haklim-do, South Korea are segmented into short grains, organized by timbral similarity, and fused in the frequency domain. Photographs of collected specimens are used to fine-tune a LoRA that generates organisms modeled on local species. Across sound and image, the work follows the same operations (fragmentation, fusion, and accumulation), treating composite matter as both material and method. The contribution is methodological: the technical operations that produce the work are the same operations that produce its subject, and the resulting speculative organisms make a future ecology shaped by industrial debris perceptible at the scale of a single body.

Question and Answering(QA) in low-resource languages remains a significant challenge due to the scarcity of high-quality training data.To address this, we propose a robust framework for lowresource QA.Our framework enhances performance through the integration of pivot-based translation, sub-question decomposition, and semantic consistency verification.Our proposed approach utilizes pivoting by translating questions into a high-resource language, and then translating the answer back into the original language.To improve the handling of complex queries, we introduce sub-question decomposition, which breaks down the original question into simpler sub-units for independent QA.Also, we incorporate a reverse QA mechanism that generates a new question from the predicted answer and measures its semantic similarity to the original question, thereby validating answer consistency.Evaluated on the TyDi QA benchmark, the proposed framework achieves a 19.21 chrF score and 0.67 BERTScore, corresponding to at least a 12% improvement in metrics over direct generation baselines.

Accurate instrument pose estimation is a crucial step towards the future of robotic surgery, enabling applications such as autonomous surgical task execution. Vision-based methods for surgical instrument pose estimation provide a practical approach to tool tracking, but they often require markers to be attached to the instruments. Recently, more research has focused on the development of markerless methods based on deep learning. However, acquiring realistic surgical data, with ground truth (GT) instrument poses, required for deep learning training, is challenging. To address the issues in surgical instrument pose estimation, we introduce the Surgical Robot Instrument Pose Estimation (SurgRIPE) challenge, hosted at the 26th International Conference on Medical Image Computing and Computer-Assisted Intervention (MICCAI) in 2023. The objectives of this challenge are: (1) to provide the surgical vision community with realistic surgical video data paired with ground truth instrument poses, and (2) to establish a benchmark for evaluating markerless pose estimation methods. The challenge led to the development of several novel algorithms that showcased improved accuracy and robustness over existing methods. The performance evaluation study on the SurgRIPE dataset highlights the potential of these advanced algorithms to be integrated into robotic surgery systems, paving the way for more precise and autonomous surgical procedures. The SurgRIPE challenge has successfully established a new benchmark for the field, encouraging further research and development in surgical robot instrument pose estimation.

A comprehensive review of 3-D/4-D-printed reconfigurable metasurfaces (RMSs) is presented in this article. A metasurface (MS) demonstrates exceptional abilities for electromagnetic (EM) wave molding beyond that offered by conventional planar interfaces, and RMS provides MS with more diverse EM wave-control capabilities. RMSs are categorized by the type of external stimulus used for reconfiguration, such as electrical RMS, fluidic RMS, mechanical RMS, and thermal RMS. To implement these RMSs, it is important to understand the design and fabrication requirements as well as the EM characteristic of each RMS, including its advantages and disadvantages. In particular, except for electrical RMS, RMSs require complex 3-D structures or special materials that are difficult to implement with conventional subtractive manufacturing methods such as printed-circuit-board manufacturing. Recently, advanced 3-D/4-D printing technology has achieved high fabrication freedom and meets the design and fabrication requirements of each type of RMS. In this article, we introduce representative RMSs with the development of 3-D/4-D printing technology and materials. Furthermore, current issues of RMSs based on 3-D/4-D printing technology and future directions are described.