Humans can naturally and effectively find salient regions in complex scenes. Motivated by this observation, attention mechanisms were introduced into computer vision with the aim of imitating this ...aspect of the human visual system. Such an attention mechanism can be regarded as a dynamic weight adjustment process based on features of the input image. Attention mechanisms have achieved great success in many visual tasks, including image classification, object detection, semantic segmentation, video understanding, image generation, 3D vision, multimodal tasks, and self-supervised learning. In this survey, we provide a comprehensive review of various attention mechanisms in computer vision and categorize them according to approach, such as channel attention, spatial attention, temporal attention, and branch attention; a related repository
https://github.com/MenghaoGuo/Awesome-Vision-Attentions
is dedicated to collecting related work. We also suggest future directions for attention mechanism research.
The substorm current wedge (SCW), the characteristic current system of Earth's substorms, has been suggested to be a collective effect of many “wedgelets,” mesoscale currents carried by magnetotail ...flux tubes of strong magnetic fields called dipolarizing flux bundles (DFBs). Each wedgelet contains an asymmetric pair of field‐aligned currents (FACs) so the net FAC of many wedgelets can equal an SCW's FAC content. It is unclear, however, why a wedgelet's FAC is asymmetric. To explore the reason, we investigate how earthward‐traveling DFBs interact with ambient plasma because this interaction leads to their FACs. The interaction is manifested as the pressure and magnetic field distributions around DFBs, which we examine statistically using THEMIS data. The statistical distributions are consistent with an interplay between the DFB‐caused mesoscale perturbations and the global magnetotail configuration and favor the rise of wedgelets' asymmetric FACs. This result reveals the importance of cross‐scale coupling in SCW formation.
Plain Language Summary
The electric current system of one of the most important energy release events in Earth's near‐space provides valuable information on how such events proceed. Previous studies showed that the current system consists of small elements with unexpected but necessary asymmetries. To understand how such asymmetries are formed, we examine the space plasma environment related to them using spacecraft data. The plasma environment suggests that the asymmetries arise from an interaction between small and global‐scale phenomena, indicating the importance of cross‐scale coupling in understanding the geospace.
Key Points
We examine the statistical distribution of pressure and magnetic field strength around dipolarizing flux bundles (DFBs) using THEMIS data
Around a DFB in the dusk (dawn) sector of the magnetotail, the quantities are larger on its morning (evening) side
This asymmetric distribution suggests an interplay between meso and global scales and favors asymmetric wedgelets as previously found
Salient object segmentation, edge detection, and skeleton extraction are three contrasting low-level pixel-wise vision problems, where existing works mostly focused on designing tailored methods for ...each individual task. However, it is inconvenient and inefficient to store a pre-trained model for each task and perform multiple different tasks in sequence. There are methods that solve specific related tasks jointly but require datasets with different types of annotations supported at the same time. In this paper, we first show some similarities shared by these tasks and then demonstrate how they can be leveraged for developing a unified framework that can be trained end-to-end. In particular, we introduce a selective integration module that allows each task to dynamically choose features at different levels from the shared backbone based on its own characteristics. Furthermore, we design a task-adaptive attention module, aiming at intelligently allocating information for different tasks according to the image content priors. To evaluate the performance of our proposed network on these tasks, we conduct exhaustive experiments on multiple representative datasets. We will show that though these tasks are naturally quite different, our network can work well on all of them and even perform better than current single-purpose state-of-the-art methods. In addition, we also conduct adequate ablation analyses that provide a full understanding of the design principles of the proposed framework. To facilitate future research, source code will be released.
We solve the problem of salient object detection by investigating how to expand the role of pooling in convolutional neural networks. Based on the U-shape architecture, we first build a global ...guidance module (GGM) upon the bottom-up pathway, aiming at providing layers at different feature levels the location information of potential salient objects. We further design a feature aggregation module (FAM) to make the coarse-level semantic information well fused with the fine-level features from the top-down path- way. By adding FAMs after the fusion operations in the top-down pathway, coarse-level features from the GGM can be seamlessly merged with features at various scales. These two pooling-based modules allow the high-level semantic features to be progressively refined, yielding detail enriched saliency maps. Experiment results show that our proposed approach can more accurately locate the salient objects with sharpened details and hence substantially improve the performance compared to the previous state-of-the-arts. Our approach is fast as well and can run at a speed of more than 30 FPS when processing a 300×400 image. Code can be found at http://mmcheng.net/poolnet/.
Abstract
Background
In gut microbiome studies, the cultured gut microbial resource plays essential roles, such as helping to unravel gut microbial functions and host-microbe interactions. Although ...several major studies have been performed to elucidate the cultured human gut microbiota, up to 70% of the Unified Human Gastrointestinal Genome species have not been cultured to date. Large-scale gut microbial isolation and identification as well as availability to the public are imperative for gut microbial studies and further characterizing human gut microbial functions.
Results
In this study, we constructed a human Gut Microbial Biobank (hGMB; homepage:
hgmb.nmdc.cn
) through the cultivation of 10,558 isolates from 31 sample mixtures of 239 fresh fecal samples from healthy Chinese volunteers, and deposited 1170 strains representing 400 different species in culture collections of the International Depository Authority for long-term preservation and public access worldwide. Following the rules of the International Code of Nomenclature of Prokaryotes, 102 new species were characterized and denominated, while 28 new genera and 3 new families were proposed. hGMB represented over 80% of the common and dominant human gut microbial genera and species characterized from global human gut 16S rRNA gene amplicon data (
n
= 11,647) and cultured 24 “most-wanted” and “medium priority” taxa proposed by the Human Microbiome Project. We in total sequenced 115 genomes representing 102 novel taxa and 13 previously known species. Further in silico analysis revealed that the newly sequenced hGMB genomes represented 22 previously uncultured species in the Unified Human Gastrointestinal Genome (UHGG) and contributed 24 representatives of potentially “dark taxa” that had not been discovered by UHGG. The nonredundant gene catalogs generated from the hGMB genomes covered over 50% of the functionally known genes (KEGG orthologs) in the largest global human gut gene catalogs and approximately 10% of the “most wanted” functionally unknown proteins in the FUnkFams database.
Conclusions
A publicly accessible human Gut Microbial Biobank (hGMB) was established that contained 1170 strains and represents 400 human gut microbial species. hGMB expands the gut microbial resources and genomic repository by adding 102 novel species, 28 new genera, 3 new families, and 115 new genomes of human gut microbes.
We demonstrated the metabolic benefits of Parabacteroides distasonis (PD) on decreasing weight gain, hyperglycemia, and hepatic steatosis in ob/ob and high-fat diet (HFD)-fed mice. Treatment with ...live P. distasonis (LPD) dramatically altered the bile acid profile with elevated lithocholic acid (LCA) and ursodeoxycholic acid (UDCA) and increased the level of succinate in the gut. In vitro cultivation of PD demonstrated its capacity to transform bile acids and production of succinate. Succinate supplementation in the diet decreased hyperglycemia in ob/ob mice via the activation of intestinal gluconeogenesis (IGN). Gavage with a mixture of LCA and UDCA reduced hyperlipidemia by activating the FXR pathway and repairing gut barrier integrity. Co-treatment with succinate and LCA/UDCA mirrored the benefits of LPD. The binding target of succinate was identified as fructose-1,6-bisphosphatase, the rate-limiting enzyme in IGN. The succinate and secondary bile acids produced by P. distasonis played key roles in the modulation of host metabolism.
Display omitted
•Parabacteroides distasonis alleviates obesity and obesity-related dysfunctions in mice.•P. distasonis generates succinate and secondary bile acids in the gut.•P. distasonis activates intestinal gluconeogenesis (IGN) and FXR pathways in the gut.•Succinate is a ligand of fructose-1,6-bisphosphatase, the rate-limiting enzyme in IGN.
Wang et al. report the metabolic benefits of gut commensal Parabacteroides distasonis via secondary bile acid-activated FXR signaling and succinate-activated intestinal gluconeogenesis (IGN). Succinate binds fructose-1,6-bisphosphatase, the rate-limiting enzyme in IGN.
Over the past decade, wood‐derived materials have attracted enormous interest for both fundamental research and practical applications in various functional devices. In addition to being renewable, ...environmentally benign, naturally abundant, and biodegradable, wood‐derived materials have several unique advantages, including hierarchically porous structures, excellent mechanical flexibility and integrity, and tunable multifunctionality, making them ideally suited for efficient energy storage and conversion. In this article, the latest advances in the development of wood‐derived materials are discussed for electrochemical energy storage systems and devices (e.g., supercapacitors and rechargeable batteries), highlighting their micro/nanostructures, strategies for tailoring the structures and morphologies, as well as their impact on electrochemical performance (energy and power density and long‐term durability). Furthermore, the scientific and technical challenges, together with new directions of future research in this exciting field, are also outlined for electrochemical energy storage applications.
Recent progress in the development of wood and its derivative materials for advanced energy storage devices, including supercapacitors and rechargeable batteries, is reviewed, in particular for their micro/nanostructure, structure engineering strategies, and electrochemical performance.
An unprecedented synthesis of N‐heteroaromatics from biaryl aldehydes and NH3 through reagent‐free C−H/N−H cross‐coupling has been developed. The electrosynthesis uses NH3 as an inexpensive and ...atom‐economic nitrogen donor, requires no oxidizing agents, and allows efficient and regioselective access to a wide range of phenanthridines and structurally related polycyclic N‐heteroaromatic products.
It's electricfying: An unprecedented synthesis of N‐heteroaromatics from biaryl aldehydes and NH3 through reagent‐free C−H/N‐H cross‐coupling has been developed. The electrosynthesis uses NH3 as an inexpensive and atom‐economic nitrogen donor, requires no oxidizing agents, and allows efficient and regioselective access to a wide range of phenanthridines and structurally related polycyclic N‐heteroaromatic products. HFIP=1,1,1,3,3,3‐hexafluoro‐2‐propanol
Perovskite light‐emitting didoes (PeLEDs) have shown considerable potential in solution‐processable display applications. However, the performance of blue PeLEDs in terms of efficiency and stability ...hinders their practicality on account of severe trap‐mediated nonradiative recombination losses and halide phase segregation. To ameliorate these issues, mixed‐halide sky‐blue perovskite materials are strategically modulated through crystal defect passivation with a trifurcate isocyanate oligomer, which leads to the synergistical suppression of charge trap density and halide ion migration. The proposed approach enables the performance improvement for sky‐blue PeLEDs, exhibiting a peak external quantum efficiency of 14.82% and spectrally stable emission at 487 nm. In addition, prolonged operational lifetime and enhanced capability of moisture resistance are achieved simultaneously, approaching a half‐lifetime of ≈2900 s at an initial brightness of 178 cd m–2.
A trifurcate isocyanate oligomer is proposed for modulating mixed‐halide sky‐blue perovskite materials by passivating crystal defects, suppressing halide ion migration, and resisting moisture simultaneously. Sky‐blue perovskite light‐emitting diodes achieve a high external quantum efficiency of 14.82% and distinctly improved operational stability with a half‐lifetime of ≈3000 s.
The U-shape structure has shown its advantage in salient object detection for efficiently combining multi-scale features. However, most existing U-shape-based methods focused on improving the ...bottom-up and top-down pathways while ignoring the connections between them. This paper shows that we can achieve the cross-scale information interaction by centralizing these connections, hence obtaining semantically stronger and positionally more precise features. To inspire the newly proposed strategy's potential, we further design a relative global calibration module that can simultaneously process multi-scale inputs without spatial interpolation. Our approach can aggregate features more effectively while introducing only a few additional parameters. Our approach can cooperate with various existing U-shape-based salient object detection methods by substituting the connections between the bottom-up and top-down pathways. Experimental results demonstrate that our proposed approach performs favorably against the previous state-of-the-arts on five widely used benchmarks with less computational complexity. The source code will be publicly available.