The class activation maps are generated from the final convolutional layer of CNN. They can highlight discriminative object regions for the class of interest. These discovered object regions have ...been widely used for weakly-supervised tasks. However, due to the small spatial resolution of the final convolutional layer, such class activation maps often locate coarse regions of the target objects, limiting the performance of weakly-supervised tasks that need pixel-accurate object locations. Thus, we aim to generate more fine-grained object localization information from the class activation maps to locate the target objects more accurately. In this paper, by rethinking the relationships between the feature maps and their corresponding gradients, we propose a simple yet effective method, called LayerCAM. It can produce reliable class activation maps for different layers of CNN. This property enables us to collect object localization information from coarse (rough spatial localization) to fine (precise fine-grained details) levels. We further integrate them into a high-quality class activation map, where the object-related pixels can be better highlighted. To evaluate the quality of the class activation maps produced by LayerCAM, we apply them to weakly-supervised object localization and semantic segmentation. Experiments demonstrate that the class activation maps generated by our method are more effective and reliable than those by the existing attention methods. The code will be made publicly available.
Delving Deep Into Label Smoothing Zhang, Chang-Bin; Jiang, Peng-Tao; Hou, Qibin ...
IEEE transactions on image processing,
2021, Letnik:
30
Journal Article
Recenzirano
Odprti dostop
Label smoothing is an effective regularization tool for deep neural networks (DNNs), which generates soft labels by applying a weighted average between the uniform distribution and the hard label. It ...is often used to reduce the overfitting problem of training DNNs and further improve classification performance. In this paper, we aim to investigate how to generate more reliable soft labels. We present an Online Label Smoothing (OLS) strategy, which generates soft labels based on the statistics of the model prediction for the target category. The proposed OLS constructs a more reasonable probability distribution between the target categories and non-target categories to supervise DNNs. Experiments demonstrate that based on the same classification models, the proposed approach can effectively improve the classification performance on CIFAR-100, ImageNet, and fine-grained datasets. Additionally, the proposed method can significantly improve the robustness of DNN models to noisy labels compared to current label smoothing approaches. The source code is available at our project page: https://mmcheng.net/ols/
Modulating effector immune cells via monoclonal antibodies (mAbs) and facilitating the co-engagement of T cells and tumor cells via chimeric antigen receptor- T cells or bispecific T cell-engaging ...antibodies are two typical cancer immunotherapy approaches. We speculated that immobilizing two types of mAbs against effector cells and tumor cells on a single nanoparticle could integrate the functions of these two approaches, as the engineered formulation (immunomodulating nano-adaptor, imNA) could potentially associate with both cells and bridge them together like an 'adaptor' while maintaining the immunomodulatory properties of the parental mAbs. However, existing mAbs-immobilization strategies mainly rely on a chemical reaction, a process that is rough and difficult to control. Here, we build up a versatile antibody immobilization platform by conjugating anti-IgG (Fc specific) antibody (αFc) onto the nanoparticle surface (αFc-NP), and confirm that αFc-NP could conveniently and efficiently immobilize two types of mAbs through Fc-specific noncovalent interactions to form imNAs. Finally, we validate the superiority of imNAs over the mixture of parental mAbs in T cell-, natural killer cell- and macrophage-mediated antitumor immune responses in multiple murine tumor models.
The recycling of waste masks (melt-blown polypropylene nonwoven fabrics) generated during the COVID-19 outbreak has now become a social issue of high priority. Converting these waste masks into ...electromagnetic wave (EMW) absorbers is one of the most cost-effective and high-value-added methods. In this study, we successfully converted household masks into magnetic metal doping carbon micro-particles with columnar structure to achieve excellent microwave absorption properties through sulfuric acid modification and doping magnetic metal. Specifically, the cobalt-doped carbon columnar micro-particles have a minimum reflection loss (RLmin) of −52.8 dB at 7.0 GHz (d = 3 mm) and an effective absorption bandwidth (EAB) of 4.6 GHz (d = 1.4 mm). The nickel/carbon hybrid materials have an RLmin of −26.2 dB at 8.8 GHz (d = 3 mm) and an EAB of 5.3 GHz (d = 2.4 mm). The multiple reflection or scattering of EMWs, conduction and magnetic losses, dipolar polarization, and interfacial polarization effects can be enhanced in those magnetic metal doping carbon micro-particles with columnar structure, which leads to the excellent microwave absorption performance.
Display omitted
•A simple and efficient carbonization route for recycling waste masks is given.•Magnetic metal doping carbon microparticles with columnar structure are prepared.•Cobalt and nickel doping samples exhibit excellent microwave absorption.•Cobalt and nickel doping samples have strong dipolar polarization.•Cobalt and nickel doping samples show dielectric-magnetic synergistic effect.
With the increasing utilization of electric vehicles (EVs), transportation systems and electrical power systems are becoming increasingly coupled. However, the interaction between these two kinds of ...systems are not well captured, especially from the perspective of transportation systems. This paper studies the reliability of integrated transportation and electrical power system (ITES). A bidirectional EV charging control strategy is first demonstrated to model the interaction between the two systems. Thereafter, a simplified transportation system model is developed, whose high efficiency makes the reliability assessment of the ITES realizable with an acceptable accuracy. Novel transportation system reliability indices are then defined from the view point of EV's driver. Based on the charging control model and the transportation simulation method, a daily periodic quasi sequential reliability assessment method is proposed for the ITES system. Case studies based on RBTS system demonstrate that bidirectional charging controls of EVs will benefit the reliability of power systems, while decreasing the reliability of EVs travelling. Also, the optimal control strategy can be obtained based on the proposed method. Finally, case studies are performed based on a large scale test system to verify the practicability of the proposed method.
The spikelet number per spike (SNS) contributes greatly to grain yield in wheat. Identifying various genes that control wheat SNS is vital for yield improvement. This study used a recombinant inbred ...line population genotyped by the Wheat55K single-nucleotide polymorphism array to identify two major and stably expressed quantitative trait loci (QTLs) for SNS. One of them (QSns.sau-2SY-2D.1) was reported previously, while the other (QSns.sau-2SY-7A) was newly detected and further analyzed in this study. QSns.sau-2SY-7A had a high LOD value ranging from 4.46 to 16.00 and explained 10.21–40.78％ of the phenotypic variances. QSns.sau-2SY-7A was flanked by the markers AX-110518554 and AX-110094527 in a 4.75-cM interval on chromosome arm 7AL. The contributions and interactions of both major QTLs were further analyzed and discussed. The effect of QSns.sau-2SY-7A was successfully validated by developing a tightly linked kompetitive allele specific PCR marker in an F2:3 population and a panel of 101 high-generation breeding wheat lines. Furthermore, several genes including the previously reported WHEAT ORTHOLOG OF APO1 (WAPO1), an ortholog of the rice gene ABERRANT PANICLE ORGANIZATION 1 (APO1) related to SNS, were predicted in the interval of QSns.sau-2SY-7A. In summary, these results revealed the genetic basis of the multi-spikelet genotype of wheat line 20828 and will facilitate subsequent fine mapping and breeding utilization of the major QTLs.
Abstract
Background
Inflammatory bowel disease (IBD) has become a global public health problem. The prevalence of IBD in China increased annually in past two decades.
Methods
This study was to ...translate and validate the rating form of IBD patients' concerns (RFIPC), and to describe disease-related worries and concerns of patients with IBD. The simplified Chinese version of the RFIPC was developed according to translation and back-translation procedure. Patients with IBD were consecutively enrolled from the First Affiliated Hospital of Guangzhou University of Chinese Medicine. The participants were assessed using the RFIPC and the Short Inflammatory Bowel Disease Questionnaire (SIBDQ). Internal consistency, test–retest reliability, measurement error, confirmatory factor analysis (CFA) and correlation of the RFIPC with the SIBDQ were performed to evaluate the psychometric characteristics of the RFIPC.
Results
A total of 116 patients with IBD, 73 with ulcerative colitis (UC) and 43 with Crohn’s disease (CD), were enrolled in this study. Thirty-seven of them recompleted the questionnaires for the second time between 7 and 14 days after the first interview. The results of CFA indicated the original structure of the RFIPC was reasonable. Cronbach's alpha value of the RFIPC were 0.97. The intraclass correlation coefficients of four domains ranged from 0.85 to 0.92. The standard error of measurement was 7.10. The correlation coefficients between total score of the RFIPC and the SIBDQ score ranged from − 0.54 to − 0.70. Median total score of the RFIPC was 39.4 (IQR 24.0–59.3). Patients with severe symptoms reported higher scores of the RFIPC. The uncertain nature of disease, having surgery, having an ostomy bag, developing cancer, feeling out of control, being a burden on others and financial difficulties were highest concerns of patients with IBD. Comparing with patients with UC, patients with CD had more concerns of the ability to have children and being treated as different (
P
< 0.05).
Conclusions
The simplified Chinese version of RFIPC is a valid and reliable tool. It could be used for assessing disease-related worries and concerns of patients with IBD in China. Specific concerns of patients with UC and CD are different, therefore, health workers should consider the specific needs of UC and CD patients.
Celotno besedilo
Dostopno za:
DOBA, IZUM, KILJ, NUK, PILJ, PNG, SAZU, SIK, UILJ, UKNU, UL, UM, UPUK
Phosphorus (P) cycling present in sediments associated with iron (Fe), manganese (Mn) and sulfur (S) geochemical processes may cause secondary pollution in overlying water. Understanding the ...mechanisms of P release from sediments should help to restore water quality. This study used the diffusive gradients in thin film (DGT) technique to investigate the seasonal variation in the lability, remobilization mechanisms, and release characteristics of sediment P in the uncontaminated Xizhi River and the severely contaminated Danshui River, South China. P accumulation in sediments contributed to higher DGT-labile P concentrations in contaminated reaches, and the highest labile P concentrations were generally observed in summer season at each site. The significant positive relationships (p < 0.05) between labile Fe and P confirmed the Fe–P coupling release mechanism in uncontaminated sediments. Stronger relationships between labile Mn and P at contaminated sites indicated that Mn oxides played an important role in P remobilization. However, sulfate reduction associated with microbial activities (crucial genera: Desulfobulbus, Desulfomicrobium and Desulforhabdus) was considered to decouple the Fe & Mn–P cycling relationship, promoting P release at contaminated sites. The effluxes of sediment P were much higher in the Danshui River (mean 0.132 mg cm−2·d−1) than in the Xizhi River (mean 0.038 mg cm−2·d−1). And hot season led to growth in P effluxes that was much greater in contaminated river.
Display omitted
•Fe–P coupling release was the major mechanism leading to remobilization of sediment P.•Reductive dissolution of Mn oxides was responsible for P release in polluted sediment.•Sulfate reduction related to specific microbial activity triggered decoupling Fe/Mn–P.•Upward diffusion fluxes of P and Fe were greater at polluted sites in summer seasons.
•Tin in e-waste was efficiently separated and recycled by a novel two-stage CVT process.•The conversion of sn-SnO(g)-SnO2 during staged oxidation was revealed.•The inhibition mechanism of O2 content ...on SnO disproportionation was revealed.•Tetragonal nano-SnO2 with a crystallinity of 99.9% was obtained from e-waste.
E-waste is a valuable resource for the recovery of secondary metals. However, traditional methods only focused on the extraction of Cu and noble metals (Au, Ag, etc.), and significant tin (Sn) loss occurred during the smelting or the leached. In this paper, a novel chemical vapor transport (CVT) process was proposed to separate and recycle Sn from e-waste to prepare nano-SnO2. The effect of roasting parameters on Sn volatilization and characterization of nano-SnO2 were investigated using thermodynamic analysis, XRD, SEM, TEM, etc. The results indicated that Sn volatilization of 92.8 % was obtained under optimal roasting parameters under CO-CO2-N2 atmosphere. In addition, nano-SnO2 with a crystallinity of 99.9 %, an average grain size of 24.8 nm and a specific surface area of 97.9 m2/g was synthesized successfully.
•Smart EMI shielding materials are focused and reviewed.•Strain-responsive EMI shielding materials are summarized.•Chemical reagent-responsive EMI shielding materials are summarized.•Shape ...memory-responsive EMI shielding materials are summarized.•Phase transition-responsive EMI shielding materials are summarized.
With the rapid advancement of the intelligent era, intelligent electromagnetic interference (EMI) shielding devices are receiving more and more attention due to their advantages in environmental self-adaption response. Accordingly, appropriate EMI shielding materials are crucial to blocking harmful electromagnetic radiation and passing serviceable electromagnetic waves. Smart EMI shielding materials that can dynamically adjust their EMI shielding effectiveness (SE) in response to specific application requirements and environmental changes are extremely advantageous in both military and civil areas. To date, materials with adjustable EMI SE for various responses have been developed. This review pays special attention to smart materials with tunable EMI SE. The design strategies, mechanism and recent progress of smart EMI shielding materials are discussed together with different stimuli responses, including compression strain, tensile strain, chemical reagent, shape memory, phase transition and crossover angle change-induced responses. The review ends up to discuss challenges and perspectives for smart EMI shielding materials.
Display omitted
PDF
