Search

The superradiance instability of rotating black holes leads to the formation of an ultralight boson cloud with distinctive observational signatures, making black holes an effective probe of ultralight dark matter. However, around black holes in a binary system, the superradiance effect of such clouds can be terminated by tidal perturbations from the companion, leading to cloud depletion. In this study, we perform the first analysis of the impact of this termination effect on superradiant black hole binaries which are realistically modelled after their statistics in our Galaxy. Working with a dataset of approximately $10^7$ black hole binaries simulated using the Stellar EVolution for N-body (SEVN) population synthesis code, we identify the superradiant candidates and those that manage to survive the termination effect. We then calculate the cloud survival rate for various boson masses and black hole spin models. Our findings reveal that the $l=m=1$ cloud modes are generally stable against termination, whereas the $l=m=2$ modes can be significantly affected, with survival rates dropping below $10\%$ for boson masses below approximately $0.5\times 10^{-12}$ eV. In addition, our analysis indicates that clouds that overcome termination typically exhibit a higher superradiant growth rate and therefore a higher detectability., Comment: 13 pages, 10 figures

Playing outdoors in nature with peers has been attributed most importance for children's healthy development but is increasingly marginalized because of the attractiveness of screen-based play. Careful merging of digital technology into outdoor play environments rich on nature elements could potentially help bridge digital play with more traditional play activities outdoors. A systematic comparison was made of outdoor play in more or less green settings, with and without digital installations or traditional play equipment. The separate and combined role of digital artefacts, play equipment and natural elements, were investigated, with particular focus on the effects of merging digital materials into nature. A group of children aged 6-8 were involved in a field study in a three-week period playing in a traditional playground, a forest and in a forest with digitally enhanced play artefacts. Children ' s play behavior was evaluated using a behavioral tracking method, a questionnaire and a contextual interview with the children, and a physical activity measure, in combination with inventories including maps to document the design, and the ecological and physical status of the settings. The study documents differences in children's play behavior across the three settings. It differs most between the digital forest setting and the forest setting regarding the play categories imaginative play, physical play and rule play and the digital forest setting stands out when it comes to expressive play. It is discussed how particular attributes in the physical environment influence the overall play flow and the interactive effects of natural material and digital material. Ecologically, the forest and the forest with digitally enhanced artefacts were more diverse than the traditional playground, but the natural material present was important for play in all settings., QC 20240318

Robo-advisors, also known as robo-advisory services, significantly reshape customer service in financial advisory industries, transforming retail investor markets by substituting human financial advisory experts with artificial intelligence empowered services. However, existing literature remains scattered across disciplines, with theories on financial customer service predominantly focused on Internet banking, neglecting artificial intelligence empowered interactions. Thus, service providers need a framework for implementing robo-advisors in frontline service and researchers require an advanced agenda to stimulate future research. Through a systematic, interdisciplinary literature review based on Belanche et al.’s service robot framework, this article contextualizes service robot theories into financial advisory services, synthesizing knowledge on artificial intelligence empowered customer service. We contribute to literature on service robots by contextualizing, refining, and extending the original framework by Belanche et al. and by developing a research agenda with critical perspectives. Moreover, the study yields practical and theoretical insights into artificial intelligence empowered financial advisory services., QC 20240116

Shi-Yuan Fan,1,* Zi-Chi Zhao,1,* Xing-Lv Liu,1 Ying-Gang Peng,1 Hui-Min Zhu,1 Shi-Fan Yan,2 Yan-Juan Liu,2 Qin Xie,2 Yu Jiang,2 Sai-Zhen Zeng1 1The First Affiliated Hospital of Hunan Normal University (Hunan Provincial People’s Hospital), Changsha, Hunan, 410005, People’s Republic of China; 2Department of Emergency, Institute of Emergency Medicine, Hunan Provincial People’s Hospital (The First Affiliated Hospital of Hunan Normal University), Changsha, Hunan, 410005, People’s Republic of China*These authors contributed equally to this workCorrespondence: Sai-Zhen Zeng, The First Affiliated Hospital of Hunan Normal University (Hunan Provincial People’s Hospital), 61 Jie-Fang West Road, Fu-Rong District, Changsha, Hunan, 410005, People’s Republic of China, Email 55445321@qq.com Yu Jiang, Department of Emergency, Institute of Emergency Medicine, Hunan Provincial People’s Hospital (The First Affiliated Hospital of Hunan Normal University), 61 Jie-Fang West Road, Fu-Rong District, Changsha, 410005, People’s Republic of China, Email jiangyu@hunnu.edu.cnBackground: Globally, the subsequent complications that accompany sepsis result in remarkable morbidity and mortality rates. The lung is among the vulnerable organs that incur the sepsis-linked inflammatory storm and frequently culminates into ARDS/ALI. The metformin-prescribed anti-diabetic drug has been revealed with anti-inflammatory effects in sepsis, but the underlying mechanisms remain unclear. This study aimed to ascertain metformin’s effects and functions in a young mouse model of sepsis-induced ALI.Methods: Mice were randomly divided into 4 groups: sham, sham+ Met, CLP, and CLP+ Met. CLP was established as the sepsis-induced ALI model accompanied by intraperitoneal metformin treatment. At day 7, the survival state of mice was noted, including survival rate, weight, and M-CASS. Lung histological pathology and injury scores were determined by hematoxylin–eosin staining. The

Nitrate accumulation in the soil profile in the intensive agricultural region has been widely concerned in the world. However, the changes in nitrate accumulation characteristics caused by climate change, such as extremely high precipitation, are not well quantified, particularly for the regions with thick unsaturated zones. Here, we resampled the soil profiles taken in normal year (2020) after extreme precipitation year (2021) (>800 cm) in three regions in the southern Loess Plateau (LP) with three different water managements including rainfed orchards (n = 10), well-irrigated orchards (n = 4) and canal-irrigated orchards (n = 8). The accumulation amounts, peak depths, and accumulation depths of nitrate soil profiles of the different regions of two years were compared. The results showed that average nitrate accumulation in normal year at the rainfed region (800-cm depth), well-irrigated region (800-cm depth) and canal-irrigated region (1400-cm depth) were 5995 kg N ha−1, 9765 kg N ha−1, and 19,608 kg N ha−1, respectively. Compared with 2020, extreme precipitation in 2021 led to 56–91% reductions (2060–3702 kg N ha−1) in nitrate accumulation in 0–200 cm soil layer, and average nitrate leaching into the aquifer was >1390 kg N ha−1 in the canal-irrigated region. Average migration depths of nitrate peak in rainfed, well-irrigated and canal-irrigated regions were 92 cm, 115 cm, and 188 cm, respectively; as for nitrate accumulation depths, they were 10 cm, 80 cm and 108 cm, respectively. Vertically, the dried soil layer and paleosol layer (high clay content) in the canal-irrigated region significantly hindered nitrate deep migration caused by the extreme precipitation. The result highlights that extreme precipitation significantly accelerated nitrate leaching in the deep soil profiles, and future vulnerability and risk assessment studies must account for the impacts of extreme precipitation on nitrate leaching.

Histoplasmosis is a systemic mycosis caused by the dimorphic fungus in the genus Histoplasma. Histoplasmosis is overlooked in China. This study aims to provide an epidemiological and clinical update on histoplasmosis in China by literature review. We reviewed cases of histoplasmosis reported in recent 11 years and described a case of histoplasmosis-triggered hemophagocytic lymphohistiocytosis (HLH) in an immunocompetent patient. A total of 225 cases of histoplasmosis diagnosed in China between 2012 and 2022 were involved in this study, compared with 300 cases reviewed from 1990 to 2011, an increasing number of cases of histoplasmosis have been diagnosed in the last 11 years. The majority of cases of histoplasmosis were autochthonous cases, mainly from provinces Sichuan (56/225, 24.9%), Hunan (50/225, 22.2%), Guangdong (31/225, 13.8%), and Yunnan (24/225, 10.7%). Higher incidence (52.5%, 53/99) of histoplasmosis occurred in immunocompetent patients which is similar to those from the previous 21 years, and the prevalence of the disease did not vary highly over time. Of note, the number of histoplasmosis cases is increasing, and the geographic distribution is shifting southwards over time. Improved awareness is critically important for informing clinical practice in China.

A superradiant cloud of ultralight bosons near a rotating black hole provides a smoking gun for particle physics in the infrared. However, tidal perturbations from a nearby binary companion can destabilise the boson cloud and even terminate superradiance. In this work, we consider the backreaction of superradiance termination to the dynamics of general binary orbits parametrised by their semi-latus rectum, eccentricity and inclination angle. Our analysis focuses on Extreme Mass Ratio Inspiral (EMRI) systems and employs the period-average approximation to derive evolution equations of these binary parameters in the Newtonian limit. We find that the binary evolution history can be significantly modulated by the backreaction towards large circular equatorial orbits with reduced termination rate. This process can generically happen even away from the resonance bands. Our work therefore serves as a first step towards probing ultralight bosons through the statistics of EMRI binary parameters in the future., Comment: 13 pages, 5 figures

AI-empowered financial advisory services (or robo-advisory services) emerge as one of the channels to attract customers to invest in financial products that integrate the environment, social, and governance (ESG) criteria to their investment objectives. This reasoning is often based on the assumption that AI is more transparent in fees and unbiased than conventional human advisors, while at the same time, lowering the entry bar for young and low-budget customers. In automated services with little or no human intervention, users’ experience and ability to use this type of service play a critical role in supporting consumer investment decisions. However, the impact of robo-advisor user experience towards customers’ preference for sustainable investment choices has not been addressed by previous studies. In the long run, can robo-advisory services significantly support or promote a more sustainable investment portfolio to customers? We provide initial insights on these questions based on a mixed-method user test, including a pre-test survey, observations of robo-advisor usage, and a post-test retrospective interview. The preliminary results show that this AI-empowered system and its service have not fulfilled the expectations to support customers’ sustainable investment decision-making due to the lack of comprehensible information and transparency. Firstly, we explain the relationships between customers’ intention to select a sustainable portfolio with their features and attitudes towards AI. The observation and interview data reveal that a standardized definition and criteria of sustainable investment and portfolios are needed for customers to establish a fundamental understanding of this service. Also, customers demand more explanations in the final recommendation automatically generated by AI., QC 20231207

We offer theory and evidence regarding the impact of content ideology (i.e., emotionally charged beliefs expressed in sentiments) on opinion polarization (i.e., conflicting attitudes about an event) on social media. Specifically, we consider the moderating role of functional affordances and symbolic expressions to draw inferences about opinion polarization. From a sentiment analysis of 3600 posts and a survey of 468 Weibo users, we find that content ideology is positively related to social media opinion polarization. The effect of content ideology is greater when users receive stronger symbolic expressions. Further, our results show an insignificant moderating relationship between functional affordances and this effect. The findings suggest that it is critical to consider content ideology and symbolic expressions when assessing the relationship between published content and polarized opinions on social media.

AI-empowered and algorithm-driven automated financial advisory systems, also known as Robo-advisors, have been rapidly implemented by service providers and customers in financial service markets. Yet, few empirical studies investigate customers’ experience interacting with fully functional Robo-advisors in real-life scenarios. Also, it is still unknown how the design of the automated system can affect customers’ perception and adoption of this new technology. To mitigate these gaps, 24 participants with different levels of experience and understanding of financial investment were asked to use a Robo-advisor from a retail bank and perform the tasks. By conducting observations and retrospective post-test interviews, we find that participants do not fully perceive the social aspects supposed to be provided by Robo-advisors. The overarching problems are, among others, a lack of transparency and incomprehensible information. This results in distrust of the results generated by this system, which negatively affects customers’ adoption of the investment advice provided by the Robo-advisor. The potential of interactive data visualization is also detected. This work contributes to the understanding of customers regarding their perception and adoption based on their use of a functional Robo-advisor and proposes design takeaways for transparent and comprehensible automated advisory systems in financial service contexts., QC 20230504

AI-empowered financial advisory services, also known as robo-advisors, present both innovations and challenges as they replace human financial advisors, reshape customer service, and attract customers with different characteristics than their predecessors. Therefore, it is more important than ever for financial service providers to understand customers’ perception and experience of using robo-advisors on the service front line. However, our systematic literature review indicates that research on robo-advisors is scattered across different disciplines with a narrow focus on sharing knowledge across fields. Moreover, empirical studies on robo-advisors have primarily focused on customer acceptance and intentional behaviors, often based on data collected through surveys. Also, these studies have paid less attention to the role of robo-advisor design and the context in which customers interact with fully functional robo-advisors in real-life situations. To address these gaps, this thesis aims to synthesize interdisciplinary knowledge and identify gaps in the research on robo-advisors. It also aims to explore customers’ experience of using and interacting with robo-advisors and how the experience affects their perception and adoption of the service. By conducting a systematic literature review and a qualitative user study, this thesis finds that customers’ perceptions of robo-advisors often do not meet their expectations. Nontransparency and incomprehensible information about the system’s decision-making are significant barriers to customers’ adoption of robo-advisors. This thesis contributes to a deeper understanding of customers in AI-empowered financial advisory services by using theories and approaches across different disciplines. It also provides practical implications for practitioners in the robo-advisory service industry., AI-förstärkta finansiella robotrådgivare innebär både innovationer och utmaningar eftersom de ersätter mänskliga finansiella rådgivare, omformar kundservice och lockar till sig andra typer av kunder än deras föregångare. Därför är det viktigare än någonsin för tjänsteleverantörer i det finansiella systemet att förstå kunders erfarenheter och upplevelser av robotrådgivning i servicefronten. Dock indikerar avhandlingens systematiska litteraturgenomgång att tidigare forskning om robotrådgivare är spridd över olika discipliner med begränsat fokus på att dela kunskap mellan disciplinerna. Dessutom har de empiriska studierna rörande robotrådgivare primärt fokuserat på kundacceptans och intentionsbeteenden – ofta baserat på data insamlad via enkäter. Tidigare studier har också ägnat mindre uppmärksamhet åt robotrådgivares design och den kontext där kunder interagerar med robotrådgivare i verkliga situationer. Baserat på dessa brister i tidigare studier syntetiserar avhandlingen tvärvetenskaplig kunskap och identifierar luckor i forskning om robotrådgivare. Avhandlingen utforskar också kunders erfarenhet av att använda och interagera med robotrådgivare och hur dessa erfarenheter påverkar deras upplevelser av och användande av tjänsten. Genom en systematisk litteraturgenomgång och en kvalitativ intervjubaserad användarstudie visar avhandlingens studier att kunders upplevelse av robotrådgivare inte motsvarar deras förväntningar. Bristande transparens och svårbegriplig information om systemets beslutsfattande är betydande hinder för kundernas antagande av sådana tjänster när de interagerar med robotrådgivare. Avhandlingen bidrar till en djupare förståelse för kundupplevelser av AI-förstärkta finansiella robotrådgive genom att integrera teori och metod från olika discipliner. Det ger också praktiska implikationer för finanssektorns företag med verksamhet inom robotrådgivning., QC 20230508

We offer theory and evidence regarding the impact of content ideology (i.e., emotionally charged beliefs expressed in sentiments) on opinion polarization (i.e., conflicting attitudes about an event) on social media. Specifically, we consider the moderating role of functional affordances and symbolic expressions to draw inferences about opinion polarization. From a sentiment analysis of 3600 posts and a survey of 468 Weibo users, we find that content ideology is positively related to social media opinion polarization. The effect of content ideology is greater when users receive stronger symbolic expressions. Further, our results show an insignificant moderating relationship between functional affordances and this effect. The findings suggest that it is critical to consider content ideology and symbolic expressions when assessing the relationship between published content and polarized opinions on social media.

Han-Ze Du,1,* Kang Chen,1,* Li-Yuan Zhang,1,* Yue-Lun Zhang,2 Dan-Ning Wu,1 Xiao-Yuan Guo,1 Lian Duan,1 Lin-Jie Wang,1 Hong-Bo Yang,1 Shi Chen,1 Hui Pan,1 Hui-Juan Zhu1 1Key Laboratory of Endocrinology of National Health Commission, Department of Endocrinology, State Key Laboratory of Complex Severe and Rare Diseases, Peking Union Medical College Hospital, Chinese Academy of Medical Science & Peking Union Medical College, Beijing, People’s Republic of China; 2Medical Research Center, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences & Peking Union Medical College, Beijing, People’s Republic of China*These authors contributed equally to this workCorrespondence: Hui Pan; Hui-Juan Zhu, Key Laboratory of Endocrinology of National Health Commission, Department of Endocrinology, State Key Laboratory of Complex Severe and Rare Diseases, Peking Union Medical College Hospital, Chinese Academy of Medical Science & Peking Union Medical College, No. 1 Shuaifuyuan, Wangfujing Street, Dongcheng District, Beijing, 100730, People’s Republic of China, Email panhui20111111@163.com; shengxin2004@163.comContext: The high prevalence of hypothalamic obesity (HO) and dyslipidemia in individuals with craniopharyngioma (CP) following surgery is a cause for increasing concern. However, few studies have explored the lipid profile in pediatric CP patients, with inconsistent findings. In addition, the role of recombinant human growth hormone (rhGH) replacement remains unclear in these patients.Objective: To compare the blood lipid profile among post-operative craniopharyngioma children and adolescents with that among healthy controls and to reveal the effects of rhGH replacement.Methods: Data of 79 post-operative craniopharyngioma children and adolescents in our center were retrospectively collected. Sixty patients underwent rhGH replacement during the follow-ups. We selected 36 patients who received rhGH replacement the

Xue-Qing Yu,1,2,* Jia-Qi Di,2,* Wei Zhang,3 Geng-Shu Wei,4 Zhan-Ping Ma,5 Lei Wu,6 Xue-Feng Yu,7 Hui-Zhi Zhu,8 Miao Zhou,9 Cui-Ling Feng,10 Ji-Hong Feng,11 Ping Fan,12 Jian-Sheng Li,1,2 Jian-Ya Yang1 1Department of Respiratory Disease, the First Affiliated Hospital of Henan University of Chinese Medicine, Zhengzhou, Henan Province, 450000, People’s Republic of China; 2Co-Construction Collaborative Innovation Center for Chinese Medicine and Respiratory Diseases by Henan & Education Ministry of P.R. China, Henan University of Chinese Medicine, Zhengzhou, Henan, 450046, People’s Republic of China; 3Department of Respiratory Disease, Shanghai Shuguang Hospital, Shanghai University of Chinese Medicine, Shanghai, 200000, People’s Republic of China; 4Department of Respiratory Disease, the Affiliated Hospital of Shaanxi University of Chinese Medicine, Xianyang, Shaanxi Province, 712000, People’s Republic of China; 5Department of Respiratory Disease, Shaanxi Province Hospital of Traditional Chinese Medicine, Xi’an, Shaanxi Province, 710000, People’s Republic of China; 6Department of Respiratory Disease, Hebei Province Hospital of Traditional Chinese Medicine, Shijiazhuang, Hebei, 050000, People’s Republic of China; 7Department of Respiratory Disease, the Second Affiliated Hospital of Liaoning University of Traditional Chinese Medicine, Shenyang, Liaoning Province, 110000, People’s Republic of China; 8Department of Respiratory Disease, the First Affiliated Hospital of Anhui University of Chinese Medicine, Hefei, Anhui Province, 230000, People’s Republic of China; 9Department of Respiratory Disease, the Third Affiliated Hospital of Henan University of Chinese Medicine, Zhengzhou, Henan Province, 450000, People’s Republic of China; 10Department of Traditional Chinese Medicine, People’s Hospital Affiliated to Peking University, Beijing, 100000, People’s Republic of China; 11Department of Respiratory Disease, the Second Affiliated Hospi

Hui Zhu,1,2 Ke Yang,3 Huan Yao,1 Xueying Chen,1 Shujin Yan,1,2 Yiman He,1,2 Yang Cao,2 Jie Luo,1 Dong Wang1 1Department of Ultrasound, The First Affiliated Hospital of Chongqing Medical University, Chongqing, People’s Republic of China; 2Chongqing Key Laboratory of Ultrasound Molecular Imaging, Chongqing Medical University, Chongqing, People’s Republic of China; 3Pediatric Research Institute, Children’s Hospital of Chongqing Medical University, Chongqing, People’s Republic of ChinaCorrespondence: Jie Luo; Dong Wang, Department of Ultrasound, The First Affiliated Hospital of Chongqing Medical University, 1 Youyi Road, Yuzhong District, Chongqing, 400042, People’s Republic of China, Email luojie1983@cqmu.edu.cn; wangdong4429@cqmu.edu.cnBackground: Breast cancer has become the most common cancer in women. Compare with other subtypes of breast cancer, triple-negative breast cancer (TNBC) is more likely to relapse and metastasize. Highly effective therapeutic strategies are desperately needed to be explored. In this study, a multifunctional nanoplatform is expected to mediate chemo-photothermal therapy, which can combine immunogenic cell death with checkpoint blockade to combat TNBC and distant metastasis.Methods: Poly (lactic acid-glycolic acid)-Poly (ethylene glycol) (PLGA-PEG) nanoparticles (NPs), a type of polymeric NPs, loaded with IR780, a near-infrared (NIR) dye, and doxorubicin (DOX) as the chemotherapeutic drug, were assembled by an improved double emulsification method (designated as IDNPs). The characterization, intracellular uptake, biosafety, photoacoustic (PA) imaging performance, and biodistribution of IDNPs were studied. Chemo-photothermal therapeutic effect and immunogenic cell death (ICD) were evaluated both in vitro and in vivo. The potency of chemo-photothermal therapy-triggered ICD in combination with anti-PD-1 immune checkpoint blockade (ICB) immunotherapy in eliciting immune response and treating distant tumors was further investigated.Re

Herein is reported the structural characterization and scalable preparation of the elusive iron–phosphido complex FpP(t Bu)(F) (2-F, Fp = (Fe(h5 -C5H5)(CO)2)) and its precursor FpP(t Bu)(Cl) (2-Cl) in 51% and 71% yields, respectively. These phosphide complexes are proposed to be relevant to an organoiron catalytic cycle for phosphinidene transfer to electron-deficient alkenes. Examination of their properties led to the discovery of a more efficient catalytic system involving the simple, commercially available organoiron catalyst Fp2. This improved catalysis also enabled the preparation of new phosphiranes with high yields ( t BuPCH2CHR; R = CO2Me, 41%; R = CN, 83%; R = 4-biphenyl, 73%; R = SO2Ph, 71%; R = POPh2, 70%; R = 4-pyridyl, 82%; R = 2-pyridyl, 67%; R = PPh3 +, 64%) and good diastereoselectivity, demonstrating the feasibility of the phosphinidene group-transfer strategy in synthetic chemistry. Experimental and theoretical studies suggest that the original catalysis involves 2-X as the nucleophile, while for the new Fp2-catalyzed reaction they implicate a diiron–phosphido complex Fp2(Pt Bu), 4, as the nucleophile which attacks the electron-deficient olefin in the key first P–C bond-forming step. In both systems, the initial nucleophilic attack may be accompanied by favorable five-membered ring formation involving a carbonyl ligand, a (reversible) pathway competitive with formation of the three-membered ring found in the phosphirane product. A novel radical mechanism is suggested for the new Fp2-catalyzed system.

Livestock overgrazing has led to worldwide grassland degradation, posing a significant threat to plant and soil microbial diversity. However, little is known about whether livestock overgrazing influences plant and soil microbial diversity linkages. We examined relationships between plant and soil microbial beta diversity in eight pairs of ungrazed and overgrazed sites across temperate steppes in northern China. Our results revealed a positive correlation between plant and microbial beta diversity across ungrazed grasslands, and overgrazing did not change this relationship. However, different mechanisms underlay the correlations between plant and microbial beta diversity in ungrazed and overgrazed grasslands. In ungrazed grasslands, plant and microbial diversity associations were maintained mainly due to their similar responses to the shared environmental factors. While in overgrazed grasslands, the maintenance of plant and microbial diversity associations was primarily due to their functional associations. Furthermore, the positive links between plant species and microbial taxa increased in overgrazed grasslands, indicating that more soil microbial taxa form close associations with plant species in overgrazed grasslands. Our work provides new insights regarding the mechanisms of plant and microbial communities that associate under different ecological contexts, ultimately suggesting that the functional associations of plant and microbial communities are tighter as grazing intensifies in grasslands.

The Segment Anything Model (SAM) has shown impressive performance when applied to natural image segmentation. However, it struggles with geographical images like aerial and satellite imagery, especially when segmenting mobility infrastructure including roads, sidewalks, and crosswalks. This inferior performance stems from the narrow features of these objects, their textures blending into the surroundings, and interference from objects like trees, buildings, vehicles, and pedestrians - all of which can disorient the model to produce inaccurate segmentation maps. To address these challenges, we propose Geographical SAM (GeoSAM), a novel SAM-based framework that implements a fine-tuning strategy using the dense visual prompt from zero-shot learning, and the sparse visual prompt from a pre-trained CNN segmentation model. The proposed GeoSAM outperforms existing approaches for geographical image segmentation, specifically by 26%, 7%, and 17% for road infrastructure, pedestrian infrastructure, and on average, respectively, representing a momentous leap in leveraging foundation models to segment mobility infrastructure including both road and pedestrian infrastructure in geographical images. The source code can be found on this GitHub repository: https://github.com/rafiibnsultan/GeoSAM/tree/main.

Galactic supernova remnants (SNRs) and their environments provide the nearest laboratories to study SN feedback. We performed molecular observations toward SNR W49B -- the most luminous Galactic SNR in the X-ray band, aiming to explore signs of multiple feedback channels of SNRs on nearby molecular clouds (MCs). We found very broad HCO+ lines with widths of dv = 48--75 km/s in the SNR southwest, providing strong evidence that W49B is perturbing MCs at a systemic velocity of $V_{LSR}=61$--65 km/s, and placing W49B at a distance of $7.9\pm 0.6$ kpc. We observed unusually high intensity ratios of HCO+ J=1-0/CO J=1-0 not only at shocked regions ($1.1\pm 0.4$ and $0.70\pm 0.16$), but also in quiescent clouds over 1 pc away from the SNR's eastern boundary (> 0.2). By comparing with the magnetohydrodynamics (MHD) shock models, we interpret that the high ratio in the broad-line regions can result from a cosmic-ray (CR) induced chemistry in shocked MCs, where the CR ionization rate is enhanced to around 10--100 times of the Galactic level. The high HCO+/CO ratio outside the SNR is probably caused by the radiation precursor, while the luminous X-ray emission of W49B can explain a few properties in this region. The above results provide observational evidence that SNR can strongly influence the molecular chemistry in and outside the shock boundary via their shocks, CRs, and radiation. We propose that the HCO+/CO ratio is a potentially useful tool to probe SNR's multi-channel influence on MCs., Comment: 19 pages, 11 figures. Submitted to ApJ. Comments welcome

Solving variational image segmentation problems with hidden physics is often expensive and requires different algorithms and manually tunes model parameter. The deep learning methods based on the U-Net structure have obtained outstanding performances in many different medical image segmentation tasks, but designing such networks requires a lot of parameters and training data, not always available for practical problems. In this paper, inspired by traditional multi-phase convexity Mumford-Shah variational model and full approximation scheme (FAS) solving the nonlinear systems, we propose a novel variational-model-informed network (denoted as FAS-Unet) that exploits the model and algorithm priors to extract the multi-scale features. The proposed model-informed network integrates image data and mathematical models, and implements them through learning a few convolution kernels. Based on the variational theory and FAS algorithm, we first design a feature extraction sub-network (FAS-Solution module) to solve the model-driven nonlinear systems, where a skip-connection is employed to fuse the multi-scale features. Secondly, we further design a convolution block to fuse the extracted features from the previous stage, resulting in the final segmentation possibility. Experimental results on three different medical image segmentation tasks show that the proposed FAS-Unet is very competitive with other state-of-the-art methods in qualitative, quantitative and model complexity evaluations. Moreover, it may also be possible to train specialized network architectures that automatically satisfy some of the mathematical and physical laws in other image problems for better accuracy, faster training and improved generalization.The code is available at \url{https://github.com/zhuhui100/FASUNet}., Comment: 18 pages

Automated social behaviour analysis of mice has become an increasingly popular research area in behavioural neuroscience. Recently, pose information (i.e., locations of keypoints or skeleton) has been used to interpret social behaviours of mice. Nevertheless, effective encoding and decoding of social interaction information underlying the keypoints of mice has been rarely investigated in the existing methods. In particular, it is challenging to model complex social interactions between mice due to highly deformable body shapes and ambiguous movement patterns. To deal with the interaction modelling problem, we here propose a Cross-Skeleton Interaction Graph Aggregation Network (CS-IGANet) to learn abundant dynamics of freely interacting mice, where a Cross-Skeleton Node-level Interaction module (CS-NLI) is used to model multi-level interactions (i.e., intra-, inter- and cross-skeleton interactions). Furthermore, we design a novel Interaction-Aware Transformer (IAT) to dynamically learn the graph-level representation of social behaviours and update the node-level representation, guided by our proposed interaction-aware self-attention mechanism. Finally, to enhance the representation ability of our model, an auxiliary self-supervised learning task is proposed for measuring the similarity between cross-skeleton nodes. Experimental results on the standard CRMI13-Skeleton and our PDMB-Skeleton datasets show that our proposed model outperforms several other state-of-the-art approaches.

The youngest Galactic supernova remnant G1.9+0.3 shows a discrete feature between its radio and X-ray morphologies. The observed radio morphology features a single maximum in the north, while the X-ray observation shows two opposite 'ears' on the east and west sides. Using 3D magneto hydrodynamical simulations, we investigate the formation of the discrete feature of the remnant. We have tested different parameters for better simulation and reproduced similar discrete features under an environment with density gradient and an environment with clump, which provides a possible explanation of the observation., Comment: 7 pages, 3 figures, accepted for publication in ApJ

The study on the dynamic evolution of young supernova remnants (SNRs) is an important way to understand the density structure of the progenitor's circumstellar medium. We have reported the acceleration or deceleration, proper motion and brightness changes of 260 compact radio features in the second youngest known SNR Cas A at 5\,GHz based on the VLA data of five epochs from 1987 to 2004. The radio expansion center locates at $\alpha(1950)=23^{\rm h}21^{\rm m}9^{\rm s}_{\cdot}7 \pm 0^{\rm s}_{\cdot}29, \delta(1950)=+58^{\circ}32^{\prime}25^{\prime\prime}_{\cdot}2 \pm 2^{\prime\prime}_{\cdot}2$. Three-quarters of the compact knots are decelerating, this suggests that there are significant density fluctuations in the stellar winds of the remnant's progenitor. We have verified that the acceleration or deceleration of compact knots are not related with the distribution of brightness. The brightening, fading, disappearing or new appearing of compact radio features in Cas A suggests that the magnetic field in the remnant is changing rapidly., Comment: 20 pages,9 figures, ApJ accepted

We find several missing one-loop-order contributions in previous considerations about secondary gravitational waves induced at nonlinear order in cosmological perturbations. We consider a consistent perturbative expansion to third-order in cosmological perturbations, including higher-order interactions and iterative solutions ignored in the previous literature. Tensor fluctuations induced by the source with two scalar and one tensor perturbations are correlated with the first-order tensor fluctuation and thus give a one-loop-order correction to the tensor power spectrum. The missing loop correction is \textit{scale-invariant} and \textit{negative} in the superhorion region, which secondarily reduces the initial primordial tensor power spectrum prior to the horizon re-entry. Such an IR behavior is very different from the auto-spectrum of second-order induced tensor modes discussed in the previous literature and can be important for the actual gravitational wave measurements. For a sharp peak of scalar fluctuations with $A_\zeta=10^{-2}$ at $k_*=10^{5}h/{\rm Mpc}$ motivated by the LIGO/Virgo events, we show that the tensor power spectrum at the cosmic microwave background scale reduces by at most 35%. Hence, the polarization B-mode might not be seen because of the reduction of the original tensor spectrum due to the secondary effect of primordial black hole formation., Comment: 13 pages, 6 figures

The spectral observations and analysis for the W80 Region are presented by using the data of Medium-Resolution Spectroscopic Survey of Nebulae (MRS-N) with the Large Sky Area Multi-Object Fiber Spectroscopy Telescope (LAMOST). A total of 2982 high-quality nebular spectra have been obtained in the 20 square degree field of view (FoV) which covers the W80 complex, and the largest sample of spectral data have been established for the first time. The relative intensities, radial velocities (RVs), and Full Widths at Half Maximum (FWHMs) are measured with the high spectral resolution of LAMOST MRS, for H$\alpha$ $\lambda$ 6563 \AA, [\ion{N}{ii}] $\lambda$$\lambda$ 6548 \AA, 6584 \AA \ , and [\ion{S}{ii}] $\lambda$$\lambda$ 6716 \AA, 6731 \AA \ emission lines. In the field of view of whole W80 Region, the strongest line emissions are found to be consistent with the bright nebulae, NGC 7000, IC 5070, and LBN 391, and weak line emissions also truly exist in the Middle Region, where no bright nebulae are detected by the wide-band optical observations. The large-scale spectral observations to the W80 Region reveal the systematic spatial variations of RVs and FWHMs, and several unique structural features. A 'curved feature' to the east of the NGC 7000, and a 'jet feature' to the west of the LBN 391 are detected to be showing with larger radial velocities. A 'wider FWHM region' is identified in the eastern part of the NGC 7000. The variations of [\ion{S}{ii}] / H${\alpha}$ ratios display a gradient from southwest to northeast in the NGC 7000 region, and manifest a ring shape around the 'W80 bubble' ionized by an O-type star in the L935. Further spectral and multi-band observations are guaranteed to investigate in detail the structural features., Comment: 18 pages, 14 figures, accepted for publication on RAA

Monitoring range queries over moving objects is essential to extensive location-based services. The challenge faced with these location-based services is having to process numerous concurrent range queries over a large volume of moving objects. However, the existing range query processing algorithms are almost centralized based on one single machine, which are hard to address the challenge due to the limited memory and computing resources. To address this issue, we propose a distributed search solution for processing concurrent range queries over moving objects in this work. Firstly, a Distributed Dynamic Index (DDI) that consists of a global grid index and local dynamic M-ary tree indexes was proposed to maintain the moving objects and support the search algorithm. Next, a Distributed Range Query Algorithm (DRQA) was designed based on DDI, which introduces an incremental search strategy to monitor the range queries as objects evolve; during the process, it further designs a computation sharing paradigm for processing multiple concurrent queries by making full use of their common computation to decrease the search cost. Finally, three object datasets with different distributions were simulated on a New York road network and three baseline methods were introduced to more sufficiently evaluate the performance of our proposal. Compared with state-of-the-art method, the initial query cost of the DRQA algorithm reduces by $22.7\%$ and the incremental query cost drops by 15.2%, which certifies the superiority of our method over existing approaches.

In reality, supply chain members need to determine the best suitable payment scheme when transacting with their business partners. We investigate the optimization of payment schemes with quality investment and bank credit in a decentralized supply chain consisting of one manufacturer and one retailer, both of which are capital-constrained and in need of short-term financing from the bank for their operations. First, we provide a function to describe the market demand by analyzing the impacts of price and quality on the demand. Then, we develop the profit functions of advance payment scheme (APS), normal payment scheme (NPS), and delay payment scheme (DPS) for the capital-constrained manufacturer and retailer. Furthermore, we construct the joint pricing and quality level decision models under APS, NPS, and DPS to determine the optimal profits of the supply chain, manufacturer, and retailer, respectively. On this basis, we conduct theoretical and numerical analyses on the selection of the optimal payment scheme. Our research results in manufacturer Stackelberg game indicate that, the optimal payment scheme from the perspective of supply chain, manufacturer or retailer is not uniquely determined, but generally tends to be DPS. For different perspectives, the optimal payment schemes may be different, and can be affected by the parameters. Especially, for increasing bank rate for the loan of the retailer, the supply chain, manufacturer and retailer generally tend to select DPS. In addition, for increasing bank rate for the loan of the manufacturer, the supply chain and retailer tend to select DPS, but the manufacturer tends to select APS.

Increasing plant diversity can increase ecosystem functioning, stability, and services in both natural and managed grasslands, but the effects of herbivore diversity, and especially of livestock diversity, remain underexplored. Given that managed grazing is the most extensive land use worldwide, and that land managers can readily change livestock diversity, we experimentally tested how livestock diversification (sheep, cattle, or both) influenced multidiversity (the diversity of plants, insects, soil microbes, and nematodes) and ecosystem multifunctionality (including plant biomass production, plant leaf N and P, above-ground insect abundance, nutrient cycling, soil C stocks, water regulation, and plant–microbe symbiosis) in the world’s largest remaining grassland. We also considered the potential dependence of ecosystem multifunctionality on multidiversity. We found that livestock diversification substantially increased ecosystem multifunctionality by increasing multidiversity. The link between multidiversity and ecosystem multifunctionality was always stronger than the link between single diversity components and functions. Our work provides insights into the importance of multitrophic diversity to maintain multifunctionality in managed ecosystems and suggests that diversifying livestock could promote both multidiversity and ecosystem multifunctionality in an increasingly managed world.

A rotating black hole can be clouded by light bosons via superradiance, and thus acquire an atom-like structure. If such a gravitational atom system is companioned with a pulsar, the pulsar can trigger transitions between energy levels of the gravitational atom, and these transitions can be detected by pulsar timing. We show that in such pulsar-black hole systems, fine and hyperfine structure transitions are more likely to be probed than the Bohr transition. Also, the calculation of these fine and hyperfine structure transitions are under better analytic control. Thus, these fine and hyperfine structure transitions are more ideal probes in the search for gravitational collider signals in pulsar-black hole systems., Comment: 9 pages, 4 figures; v2: references added

We apply the idea of landscape (motivated by string theory) to study the statistical nature of parameters/couplings in the standard model of strong and electroweak interactions. Following the success of this approach on the fermion masses, we discuss the divergent behavior of the probability distributions of other physical parameters/couplings to obtain some insights on the quantities that cannot be measured by current experiments but can be relevant in cosmology, in particular those in the neutrino sector. From a relatively strongly mixed PMNS matrix, we argue that the probability distribution of heavy neutrino mass $P(M)$ does not diverge at $M=0$. This analysis favors the degenerate heavy neutrino scenarios., Comment: 29 pages,7 figures

We present the analysis of Fermi Large Area Telecope (LAT) data of the gamma-ray emission in the vicinity of a radio supernova remnant (SNR), G045.7$-$00.4. To study the origin of the gamma-ray emission, we also make use of the CO survey data of Milky Way Imaging Scroll Painting to study the massive molecular gas complex that surrounds the SNR. The whole size of the GeV emission is significantly larger than that of the radio morphology. Above 3 GeV, the GeV emission is resolved into two sources: one is spatially consistent with the position of the SNR with a size comparable to that of the radio emission, and the other is located outside of the western boundary of the SNR and spatially coincident with the densest region of the surrounding molecular cloud. We suggest that the GeV emission of the western source may arise from cosmic rays (CRs) which have escaped the SNR and illuminated the surrounding molecular cloud. We find that the gamma-ray spectra of the western source can be consistently explained by this scenario with a total energy of $\sim 10^{50}{\rm erg}$ in escaping CRs assuming the escape is isotropic., Comment: 11 pages, 7 figures and 1 table -Accepted for publication in ApJ

The Large sky Area Multi-Object Fiber Spectroscopic Telescope (LAMOST) medium-resolution spectral survey of Galactic Nebulae (MRS-N) has conducted for three years since Sep. 2018 and observed more than 190 thousands nebular spectra and 20 thousands stellar spectra. However, there is not yet a data processing pipeline for nebular data. To significantly improve the accuracy of nebulae classification and their physical parameters, we developed the MRS-N Pipeline. This article presented in detail each data processing step of the MRS-N Pipeline, such as removing cosmic rays, merging single exposure, fitting sky light emission lines, subtracting skylight, wavelength recalibration, measuring nebular parameters, creating catalogs and packing spectra. Finally, a description of the data products, including nebular spectra files and parameter catalogs, is provided., Comment: Submitted to RAA. 12 pages, 7 figures, 1 table

We introduce a method of subtracting geocoronal Halpha emissions from the spectra of LAMOST medium-resolution spectral survey of Galactic nebulae (LAMOST-MRS-N). The flux ratios of the Halpha sky line to the adjacent OH lambda6554 single line do not show a pattern or gradient distribution in a plate. More interestingly, the ratio is well correlated to solar altitude, which is the angle of the sun relative to the Earth's horizon. It is found that the ratio decreases from 0.8 to 0.2 with the decreasing solar altitude from -17 to -73 degree. Based on this relation, which is described by a linear function, we can construct the Halpha sky component and subtract it from the science spectrum. This method has been applied to the LAMOST-MRS-N data, and the contamination level of the Halpha sky to nebula is reduced from 40% to less than 10%. The new generated spectra will significantly improve the accuracy of the classifications and the measurements of physical parameters of Galactic nebulae., Comment: 12 pages, 10 figures, accepted for publication in Research in Astronomy and Astrophysics

Fast-paced fitness training (e.g. aerobics, Bodycombat, step workouts) is one of the most popular activities at fitness clubs worldwide because it is beneficial for physical health and it can enhance participants' motivation and engagement of fitness training. Yet, their fast pace and required coordination make it difficult for some participants, especially for beginners and those with coordination problems, to follow the class. Here we present the design, implementation and qualitative user evaluation of TwinkleBands, a piece of wearable technology that provides real-time support with trainees' movement and coordination learning by providing discriminative visual cues on the extremities. We show that TwinkleBands helps movement imitation and coordination in several ways. Based on this, we discuss key design takeaways for future technology design to support movement teaching and learning in fast-paced activities.

The growing interest in wearables for sports and fitness calls for design knowledge and conceptualizations that can help shape future designs. Towards that end, we present and discuss a design space of wearables for these practices, based on a survey of previous work. Through a thematic analysis of 47 research publications in the domain, we surface core design decisions concerning wearability, technology design, and wearable use in practice. Building on these, we show how the design space takes into account the goals of introducing technology, that design decisions can be either directly designed, or left open for appropriation by end-users; and the social organization of the practice. We characterize prior work based on the design space elements, which yields trends and opportunities for design. Our contributions can help designers think about key design decisions, exploit trends and explore new areas in the domain of wearables for sports and fitness practices.

Ji Li,1,* Wang Jing,2,* Xiaoyang Zhai,2 Wenxiao Jia,2 Hui Zhu,2 Jinming Yu1 1Department of Oncology, Renmin Hospital of Wuhan University, Wuhan, Hubei, 430060, People’s Republic of China; 2Department of Radiation Oncology, Shandong Cancer Hospital and Institute, Shandong First Medical University and Shandong Academy of Medical Sciences, Jinan, 250117, Shandong, People’s Republic of China*These authors contributed equally to this workCorrespondence: Hui Zhu; Jinming Yu Tel +86-531-67626782; +86-531-67626947Email drzhuh@126.com; sdyujinming@163.comPurpose: A new tool based on clinical characteristics and molecular factors (Lung-molGPA) was developed to predict the survival of patients with non-small-cell lung cancer but was has not been validated. This study aims to validate the feasibility of the Lung-molGPA in NSCLC.Patients and Methods: Patients diagnosed NSCLC between Feb 2012 and July 2018 were retrospectively reviewed and scored using the Lung-molGPA tool to compare clinical outcomes. Hazard ratios (HRs) and 95% confidence intervals (CIs) were calculated by Cox regression analyses.Results: A total of 618 patients (524 adenocarcinoma [ADC], 94 non-adenocarcinoma [non-ADC]) were collected. For all patients, the median survival time (MST) was 33.0 months (33.6 and 28 months in the ADC and non-ADC groups, respectively; p = 0.21). In the ADC group, the MST for patients with a Lung-molGPA score of 3.5 to 4 was more than 4 years, while the MST was only 25 months in patients scoring 0– 1, 30.0 months in patients scoring 1.5– 2, and 35.0 months for scores of 2.5– 3 (p = 0.048). For the non-ADC group, the MST for scores 0– 1, 1.5– 2, 2.5– 3, and 3.5– 4 were 12.0, 20.2, 29.0, and 33.0 months, respectively (p = 0.017).Conclusion: Our findings provided evidence validating the Lung-molGPA score as a useful tool to determine treatment strategies and to predict prognosis. The model is stil

Recent observations of VER J2227+608 reveal a broken power $\gamma$-ray spectrum with the spectral index increasing from $\sim 1.8$ in the GeV energy range to $\sim 2.3$ in the TeV range. Such a spectral break can be attributed to radiative energy loss of energetic electrons in the leptonic scenario for the $\gamma$-ray emission, which, in combination with characteristic age of the nearby pulsar, can be used to constrain magnetic field in the emission region. We show that the radio and X-ray observations can also be explained in such a scenario. In the hadronic scenario, the spectral break can be attributed to diffusion of energetic ions in a turbulent medium and detailed spectral measurement can be used to constrain the diffusion coefficient. These two models, however, predict drastically different spectra above 100 TeV, which will be uncovered with future high-resolution observations, such as LHAASO and/or CTA., Comment: 7 pages,2 figures, Accepted at ApJL

Since Sep. 2018, LAMOST started the medium-resolution (R $\sim 7500$) spectral survey (MRS). We proposed the spectral survey of Galactic nebulae, including \hii\ regions, HH objects, supernova remnants, planetary nebulae and the special stars with MRS (LAMOST MRS-N). LAMOST MRS-N covers about 1700 square degrees of the northern Galactic plane within $40^\circ < l < 215^\circ$ and $-5^\circ < b < 5^\circ$. In this 5-year survey, we plan to observe about 500 thousand nebulae spectra. According to the commissioning observations, the nebulae spectra can provide precise radial velocity with uncertainty less than 1 km s$^{-1}$. These high-precision spectral data are of great significance to our understanding of star formation and evolution., Comment: Submitted to RAA. 13 pages, 7 figures, 1 table

H9N2 viruses can cause great economic losses to the domestic poultry industry when co-infected with other influenza viruses or pathogens. . To better understand the molecular characteristics of H9N2 avian influenza viruses (AIVs) and analyze the genetic evolutionary relationship, we isolated three H9N2 subtypes AIVs from nasopharyngeal swab specimens from the three cases reported in Anhui province since 2015, and systematically reviewed the genome-wide data of 21 poultry--isolated H9N2 viruses during 1998–2017. The six internal genes of three human-isolated viruses and recent poultry-isolated viruses (since 2014) in Anhui province presented high gene homologies with HPAI H7N9, even including H10N8 and H5N6. The three human-isolated H9N2 AIVs and poultry-isolated viruses (since 2008) in Anhui province were highly similar, and classified into genotype S. Seven N-linked potential glycosylation sites in the HA protein were detected in the three human-isolated viruses, which also appeared in poultry-isolated H9N2 AIVs. None of the human-isolated H9N2 AIVs had the I368V mutation in PB1 protein, but all the poultry-isolated H9N2 viruses in 2017 carried this mutation. Multidisciplinary, cross-regional and cross-sectoral approaches are warranted to address complex public health challenges and achieve the goal of ‘one health’.

Livestock overgrazing influences both microbial communities and nutrient cycling in terrestrial ecosystems. However, the role of overgrazing in regulating the relationship between soil biodiversity and nitrogen availability remains largely unexplored.We performed long-term grazing exclusion experiments across eight sites along precipitation gradient covering three major types of grassland in northern China to compare the linkage between soil microbial diversity and N availability in overgrazed versus non-grazed conditions.We found a significantly positive association between fungal diversity and soil available N in non-grazed grasslands. However, the positive association was absent in overgrazed environments. Bacterial diversity is not related to soil available N in either non-grazed or overgrazed grasslands. Moreover, in bacterial community, we found a positive link between the relative abundance of Actinobacteria with soil available N in non-grazed, but not overgrazed, grasslands. Instead we found the links between relative abundance of Bacteroidetes and Acidobacteria with soil available N in overgrazed grasslands, but not non-grazed, grasslands.Synthesis. Our work provides evidence that the relationships between microbial diversity and ecosystem functions are context-dependent, and so microbial community diversity is likely not the major driver of soil N mineralization in overgrazed grasslands. Our study suggests that high intensity anthropogenic activities in grasslands restrains the capacity of diverse soil microbial communities to sustain ecosystem function, and more broadly the capacity of entire ecosystems to maintain important ecosystem processes such as plant production. Our study also indicates that the fundamental microbial communities associated with N availability change with differing land management strategies (e.g. livestock grazing).

Accurate radial velocity determinations of optical emission lines (i.e. [NII]${\lambda}{\lambda}$6548,6584, H${\alpha}$, and [SII]${\lambda}{\lambda}$6717,6731) are very important for investigating the kinematics and dynamics properties of nebulae. The second stage survey program of Large sky Area Multi-Object fiber Spectroscopic Telescope (LAMOST) has started a sub-survey of nebulae (MRS-N) which will spectroscopically observe the optical emission lines of a large sample of nebulae near the Galactic plane. Until now, 15 MRS-N plates have been observed from 2017 September to 2019 June. Based on fitting the sky emission lines in the red band spectra of MRS-N, we investigate the precision of wavelength calibration and find there are systematic deviations of radial velocities (RVs) from $\sim$0.2 to 4 km/s for different plates. Especially for the plates obtained in 2018 March, the systematic deviations of RVs can be as large as $\sim$4 km/s, which then go down to $\sim$0.2-0.5 km/s at the end of 2018 and January 2019. A RVs calibration function is proposed for these MRS-N plates, which can simultaneously and successfully calibration the systematic deviations and improve the precision of RVs., Comment: 19 pages, 18 figures, 8 tables. Accepted for publication in Research in Astronomy and Astrophysics

Deep convolutional neural networks (CNN) always depend on wider receptive field (RF) and more complex non-linearity to achieve state-of-the-art performance, while suffering the increased difficult to interpret how relevant patches contribute the final prediction. In this paper, we construct an interpretable AnchorNet equipped with our carefully designed RFs and linearly spatial aggregation to provide patch-wise interpretability of the input media meanwhile localizing multi-scale informative patches only supervised on media-level labels without any extra bounding box annotations. Visualization of localized informative image and text patches show the superior multi-scale localization capability of AnchorNet. We further use localized patches for downstream classification tasks across widely applied networks. Experimental results demonstrate that replacing the original inputs with their patches for classification can get a clear inference acceleration with only tiny performance degradation, which proves that localized patches can indeed retain the most semantics and evidences of the original inputs.

Public intelligent services enabled by machine learning algorithms are vulnerable to model extraction attacks that can steal confidential information of the learning models through public queries. Differential privacy (DP) has been considered a promising technique to mitigate this attack. However, we find that the vulnerability persists when regression models are being protected by current DP solutions. We show that the adversary can launch a query-flooding parameter duplication (QPD) attack to infer the model information by repeated queries. To defend against the QPD attack on logistic and linear regression models, we propose a novel High-Dimensional Gaussian (HDG) mechanism to prevent unauthorized information disclosure without interrupting the intended services. In contrast to prior work, the proposed HDG mechanism will dynamically generate the privacy budget and random noise for different queries and their results to enhance the obfuscation. Besides, for the first time, HDG enables an optimal privacy budget allocation that automatically determines the minimum amount of noise to be added per user-desired privacy level on each dimension. We comprehensively evaluate the performance of HDG using real-world datasets and shows that HDG effectively mitigates the QPD attack while satisfying the privacy requirements. We also prepare to open-source the relevant codes to the community for further research., Comment: it has some mistakes. Since I submitted the paper for the first time, there were many mistakes in the paper. At the same time, I found a serious mistake in the content of the paper, so I thought it was inappropriate to publish it now after careful consideration.

Existing approaches to improve the performances of convolutional neural networks by optimizing the local architectures or deepening the networks tend to increase the size of models significantly. In order to deploy and apply the neural networks to edge devices which are in great demand, reducing the scale of networks are quite crucial. However, It is easy to degrade the performance of image processing by compressing the networks. In this paper, we propose a method which is suitable for edge devices while improving the efficiency and effectiveness of inference. The joint decision of multi-participants, mainly contain multi-layers and multi-networks, can achieve higher classification accuracy (0.26% on CIFAR-10 and 4.49% on CIFAR-100 at most) with similar total number of parameters for classical convolutional neural networks.

Dropout regularization has been widely used in deep learning but performs less effective for convolutional neural networks since the spatially correlated features allow dropped information to still flow through the networks. Some structured forms of dropout have been proposed to address this but prone to result in over or under regularization as features are dropped randomly. In this paper, we propose a targeted regularization method named TargetDrop which incorporates the attention mechanism to drop the discriminative feature units. Specifically, it masks out the target regions of the feature maps corresponding to the target channels. Experimental results compared with the other methods or applied for different networks demonstrate the regularization effect of our method.