RESUMEN
Conversion of plastic wastes to valuable carbon resources without using noble metal catalysts or external hydrogen remains a challenging task. Here we report a layered self-pillared zeolite that enables the conversion of polyethylene to gasoline with a remarkable selectivity of 99% and yields of >80% in 4 h at 240 °C. The liquid product is primarily composed of branched alkanes (selectivity of 72%), affording a high research octane number of 88.0 that is comparable to commercial gasoline (86.6). In situ inelastic neutron scattering, small-angle neutron scattering, solid-state nuclear magnetic resonance, X-ray absorption spectroscopy and isotope-labelling experiments reveal that the activation of polyethylene is promoted by the open framework tri-coordinated Al sites of the zeolite, followed by ß-scission and isomerization on Brönsted acids sites, accompanied by hydride transfer over open framework tri-coordinated Al sites through a self-supplied hydrogen pathway to yield selectivity to branched alkanes. This study shows the potential of layered zeolite materials in enabling the upcycling of plastic wastes.
RESUMEN
State-of-the-art methods for semantic segmentation are based on deep neural networks trained on large-scale labeled datasets. Acquiring such datasets would incur large annotation costs, especially for dense pixel-level prediction tasks like semantic segmentation. We consider region-based active learning as a strategy to reduce annotation costs while maintaining high performance. In this setting, batches of informative image regions instead of entire images are selected for labeling. Importantly, we propose that enforcing local spatial diversity is beneficial for active learning in this case, and to incorporate spatial diversity along with the traditional active selection criterion, e.g., data sample uncertainty, in a unified optimization framework for region-based active learning. We apply this framework to the Cityscapes and PASCAL VOC datasets and demonstrate that the inclusion of spatial diversity effectively improves the performance of uncertainty-based and feature diversity-based active learning methods. Our framework achieves 95% performance of fully supervised methods with only 5 - 9% of the labeled pixels, outperforming all state-of-the-art region-based active learning methods for semantic segmentation.
RESUMEN
Transfer functions play a key role in volume rendering of medical data, but transfer function manipulation is unintuitive and can be time-consuming; achieving an optimal visualization of patient anatomy or pathology is difficult. To overcome this problem, we present a system for automatic transfer function design based on visibility distribution and projective color mapping. Instead of assigning opacity directly based on voxel intensity and gradient magnitude, the opacity transfer function is automatically derived by matching the observed visibility distribution to a target visibility distribution. An automatic color assignment scheme based on projective mapping is proposed to assign colors that allow for the visual discrimination of different structures, while also reflecting the degree of similarity between them. When our method was tested on several medical volumetric datasets, the key structures within the volume were clearly visualized with minimal user intervention.
