RESUMO
OBJECTIVE: This study aimed to explore the characteristics of carbapenem-resistant Enterobacterales (CRE) patients in the intensive care unit (ICU) in different regions of Henan Province to provide evidence for the targeted prevention and treatment of CRE. METHODS: This was a cross-sectional study. CRE screening was conducted in the ICUs of 78 hospitals in Henan Province, China, on March 10, 2021. The patients were divided into provincial capital hospitals and nonprovincial capital hospitals for comparative analysis. RESULTS: This study involved 1009 patients in total, of whom 241 were CRE-positive patients, 92 were in the provincial capital hospital and 149 were in the nonprovincial capital hospital. Provincial capital hospitals had a higher rate of CRE positivity, and there was a significant difference in the rate of CRE positivity between the two groups. The body temperature; immunosuppressed state; transfer from the ICU to other hospitals; and use of enemas, arterial catheters, carbapenems, or tigecycline at the provincial capital hospital were greater than those at the nonprovincial capital hospital (P < 0.05). However, there was no significant difference in the distribution of carbapenemase strains or enzymes between the two groups. CONCLUSIONS: The detection rate of CRE was significantly greater in provincial capital hospitals than in nonprovincial capital hospitals. The source of the patients, invasive procedures, and use of advanced antibiotics may account for the differences. Carbapenem-resistant Klebsiella pneumoniae (CR-KPN) was the most prevalent strain. Klebsiella pneumoniae carbapenemase (KPC) was the predominant carbapenemase enzyme. The distributions of carbapenemase strains and enzymes were similar in different regions.
Assuntos
Antibacterianos , Temperatura Corporal , Humanos , Estudos Transversais , Antibacterianos/farmacologia , Antibacterianos/uso terapêutico , Cânula , Carbapenêmicos/farmacologia , Klebsiella pneumoniaeRESUMO
LiMn0.5Fe0.5PO4 cathodes have a high energy density but a poor rate and poor cycling performance. To this end, a series of N/S-doped LiMn0.5Fe0.5PO4/C composite cathodes modified with different contents of Li2ZrO3 were prepared by a solvothermal synthesis combined with calcination. The microstructure, chemical composition, and electrochemical properties are analyzed. Li2ZrO3 adsorbed on the LiMn0.5Fe0.5PO4 primary particles' surface in an amorphous state and on spherical particles (5-10 nm). The cycling life and rate performance of the cathodes are improved by the modification of a moderate amount of Li2ZrO3. The LMFP/NS-C/LZO1 shows available capacities of 166.8 and 118.9 mAh·g-1 at 0.1 and 5 C, respectively. The LMFP/NS-C/LZO1 shows no capacity loss after 100 cycles of charging/discharging (1 C), and still has a high capacity retention of 92.0% after 1000 cycles of charging/discharging (5 C). The excellent cycling performance of the LMFP/NS-C/LZO1 can be attributed to the improvement of the cathode microstructure and the electrochemical kinetics and the inhibition of Mn2+ dissolution by the moderate Li2ZrO3 modification.
RESUMO
Selective filtration based on superwetting materials has brought about widespread attention in the field of oil/water separation. In this study, a ZIF-L@8-coated stainless steel mesh (ZIF-L@8-coated SSM) was prepared via in situ growth of two-dimensional leaf-shaped ZIF-L nanosheets on SSM, followed by heterogeneous epitaxial growth of ZIF-8 on a ZIF-L coating. The synthesized ZIF-L@8-coated SSM with a hierarchical micro/nanoscale structure exhibited outstanding switchable wettability between underwater superoleophobicity and underoil superhydrophobicity upon respective prewetting using water and oil without additional external stimuli. It possessed excellent separation performances and stabilities with respect to various types of oil/water mixtures. The switchable wettability mechanism was analyzed and elucidated in detail. The synthesized ZIF-L@8-coated SSM with switchable wettability in this study would have great potential in on-demand oil/water separation.
RESUMO
Heteroatom-doped carbon can significantly improve the electrochemical performance of LiFePO4cathodes, but it is limited by the complex preparation process and expensive dopants. A self-assembled S-doped LiFePO4@N/S-doped C core-shell structured composites were synthesized by a convenient solvothermal method are reported. The structure and the electrochemical performance of the composites were characterized. In the S-doped LiFePO4@N/S-doped C composites, the glucose-derived carbon microspheres were attached by LiFePO4/C particles to form secondary particles in the core-shell structure. The thioacetamide regulated the morphology of LiFePO4/C particles and provided N and S atoms to dope the composites. The S-doped LiFePO4@N/S-doped C composites delivered specific discharge capacities of 157.81 mAh g-1at 0.1 C and 121.26 mAh g-1at 5 C, and capacity retention of 99.88% after 100 charge/discharge cycles. The excellent electrochemical performance of the S-doped LiFePO4@N/S-doped C composites can be attributed to the synergism of thioacetamide and glucose.
RESUMO
OBJECTIVE: To explore the effect of atrial septal defect (ASD) and venoarterial extracorporeal membrane oxygenation (VA-ECMO) in the treatment of ARDS combined with left ventricular dysfunction (LVD) to find a new effective method for treating severe COVID-19 patients. MATERIALS AND METHODS: Five large animal ARDS models of sheep were established by intravenous injection of Lipopolysaccharide. ASD was made under general anesthesia and VA-ECMO was simulated by extracorporeal circulation machine. The oxygenation of peripheral blood, systemic circulation, and cardiac function were observed under conditions of closed and opened ASD, and the significance of ASD shunt in improving cardiopulmonary function was evaluated. RESULTS: With ASD closed, the atrial shunts disappeared, the peripheral artery pressure of oxygen(PaO2): 141.2±21.4mmHg, the oxygenation index (PaO2/FiO2): 353.0±53.5, the mean blood pressure (MAP): 49.3±13.5 mmHg, the heart was full; with ASD opened, the left-to-right shunt was observed, PaO2: 169.3±18.9mmHg, PaO2/FiO2: 423.3±47.3, MAP: 68.2±16.1 mmHg, the range of cardiac motion significantly increased, heart beat was powerful, and systemic circulation significantly improved. Statistical analysis showed that there were significant differences between opened and closed ASD (P < .01). CONCLUSION: ASD plus VA-ECMO is an effective method for the treatment of ARDS combined with LVD, which is the main cause of death in severe COVID-19 patients. However, further clinical validation is needed.
Assuntos
Betacoronavirus , Infecções por Coronavirus/complicações , Oxigenação por Membrana Extracorpórea , Comunicação Interatrial/complicações , Pneumonia Viral/complicações , Síndrome do Desconforto Respiratório/terapia , Disfunção Ventricular Esquerda/complicações , Animais , COVID-19 , Diagnóstico Diferencial , Modelos Animais de Doenças , Pandemias , Síndrome do Desconforto Respiratório/etiologia , SARS-CoV-2 , OvinosRESUMO
Heat shock proteins 70 (Hsp70) constitute a highly conserved protein family of cellular chaperones widely distributed in plants, where they play a fundamental role in response to biotic and abiotic stress. Until now, genome-wide analyses of the Hsp70 gene family have been conducted for some species. However, reports about Hsp70 genes in Nicotiana tabacum are scarce. In this study, we systematically conducted genome-wide identification and expression analysis of the Hsp70 gene family in tobacco, including gene structure, classification, evolutionary relationships, promoters, and transcript levels in response to abiotic stress treatments. In all, 61 Hsp70 members were identified and classified into six groups that were mapped onto 18 chromosomes, where most were distributed on both ends of the chromosome. The conserved structures and motifs of NtHsp70 proteins in the same subfamily were highly consistent. At least 15 pairs of NtHsp70 genes underwent gene duplication by segment and tandem duplications. Most NtHsp70 proteins contained N-terminal hexokinase conserved motifs. Phylogenetic analysis showed that most species expanded according to their own species-specific approach during the evolution of Hsp70s. Tissue-specific expression analysis indicated that all NtHsp70 genes were involved in at least one or more abiotic stress responses, highlighting the wide participation of NtHsp70 genes in environmental adaptation. This is the first genome-wide analysis of Hsp70 in N. tabacum. These results indicate that each NtHsp70 member fulfilled distinct functions in response to various abiotic stresses.
Assuntos
Proteínas de Choque Térmico HSP70/genética , Proteínas de Choque Térmico HSP70/metabolismo , Nicotiana/genética , Evolução Molecular , Duplicação Gênica/genética , Perfilação da Expressão Gênica/métodos , Regulação da Expressão Gênica de Plantas/genética , Estudo de Associação Genômica Ampla , Proteínas de Choque Térmico HSP70/classificação , Chaperonas Moleculares/genética , Família Multigênica , Filogenia , Proteínas de Plantas/genética , Estresse Fisiológico/genética , Nicotiana/metabolismoRESUMO
The availability of continuous and well-defined AFI-type aluminophosphate membranes (AFI membranes) would trigger their applications in innovative materials. A well-designed manipulation strategy is proposed to produce continuous AFI membranes with four different microstructures over porous α-Al2O3 substrates. A double-layer and highly c-oriented AFI membrane of hexagonal prisms is obtained when a thin layer of medium molecular weight (MMW) chitosan is employed as the structure-directing matrix together with aluminum isopropoxide (AIP) as the Al source. It can be transformed to a single-layer and highly c-oriented AFI membrane of hexagonal prisms if the structure-directing matrix is replaced by a thin layer of low molecular weight (LMW) chitosan. When the Al source is changed to pseudo-boehmite, the single-layer AFI membrane is composed of highly ordered spherical agglomerates of small crystals. Furthermore, the membrane will turn to the double-layer AFI membrane of highly-ordered crystal agglomerates if a thin layer of MMW chitosan is used once again, keeping pseudo-boehmite as the Al source. The manipulation methodology established here is rather reliable with a pretty high reproducibility.
Assuntos
Óxido de Alumínio/química , Alumínio/química , Fosfatos/química , Cristalografia , Membranas Artificiais , Porosidade , Espectrofotometria Ultravioleta , Difração de Raios XRESUMO
Poor mental health in doctors appears to be a global phenomenon, but there are limited data on changes in doctors' psychological symptoms over time in mainland China. Through a detailed meta-analysis of cross-sectional studies, our aim was to examine the prevalence of psychological symptoms in Chinese physicians as measured with the Symptom Checklist 90-R (SCL-90-R) and to explore the factors associated with doctors' mental health. A comprehensive search was performed in major English and Chinese databases. Thirty studies involving a total of 6,099 subjects were included in the meta-analysis. The pooled estimates of psychological symptoms including somatization, obsession-compulsion, interpersonal-sensitivity, depression, anxiety, hostility, phobic anxiety, and paranoid ideation among doctors were significantly higher than those in the general population. Only psychoticism was similar in prevalence to Chinese population norms. The prevalence increased with the study year but decreased with physicians' increasing age. Doctors from central and western China experienced more mental health symptoms than those from eastern China. Psychiatrists scored significantly less favorably than other doctors on most subscales of the SCL-90-R. Doctors' mental health may be associated with age discrepancy, quantitative workload, effort-reward ratio, doctor-patient relationships, professional identity, and individual traits. To minimize the risk of poor mental health in doctors, screening and professional intervention services should be provided at early career stages to raise physicians' awareness about the importance of maintaining psychological well-being and to reduce the prevalence of psychological symptoms.
Assuntos
Transtornos Mentais/epidemiologia , Saúde Mental/estatística & dados numéricos , Médicos/psicologia , Médicos/estatística & dados numéricos , Adolescente , Adulto , Fatores Etários , China/epidemiologia , Estudos Transversais , Feminino , Humanos , Masculino , Pessoa de Meia-Idade , Prevalência , Fatores Socioeconômicos , Fatores de Tempo , Adulto JovemRESUMO
Solid polymer electrolytes (SPEs) are the key components of lithium metal batteries to overcome the obstacle of insecurity in conventional liquid electrolytes; however, the trade-off between their ionic conductivity and mechanical properties remains a significant challenge. In this work, two-dimensional ZSM-5 nanosheets as fillers are incorporated into a poly(ethylene oxide) (PEO) matrix and lithium salts to obtain composite polymer electrolytes (CPEs). The improved physicochemical and electrochemical properties of the CPE membranes are characterized in full detail. Stripping/plating measurements in symmetric Li/Li cells and cyclic charge/discharge tests are performed to investigate the cyclability and stability of the CPEs. All-solid-state LiFePO4/Li batteries deliver excellent cycling performance with an initial discharge capacity of 152.3 mAh g-1 and 91.4% capacity retention after 200 cycles at 0.2 C, with a discharge specific capacity of 118.8 mAh g-1 remaining after 350 cycles at 0.5 C. Therefore, CPEs containing ZSM-5 nanosheets are a promising option for all-solid-state lithium-ion batteries.
RESUMO
BACKGROUND: Hypertension is one of the global public health problems. Family physician-contracted service (FPCS) is widely used in the health management of hypertension patients in China. The purpose of this study was to assess the effect of FPCS on hypertension control. METHODS: PubMed, Web of Science, the Cochrane Library, China National Knowledge Network, Chinese Scientific and Technological Journal Database (CQVIP), and Wanfang Database were searched for randomized controlled trials related to family physician-contracted service and hypertension control effect, and meta-analysis was performed on the literature meeting the inclusion criteria. The source of heterogeneity was discovered by meta-regression, and it was further investigated by subgroup analysis. The risk difference (RD) and 95% confidence interval (CI) were utilized as effect values. Evaluations of publication bias and sensitivity analysis were also conducted. RESULTS: A total of 46 studies were included, and the pooled RD suggested that FPCS could effectively improve the control rate by 19% (RD = 0.19; 95%CI: 0.16-0.21; P < 0.001; I2 = 59.3%). The average age (ß = 0.28; P = 0.05) and the intervention mode (ß = 0.36; P < 0.001) were found to be heterogeneous sources by the meta-regression. According to subgroup analysis, the hypertension control rates of the elderly and working-age population in the experimental group were 93.6% and 90.1%, respectively; the control rates of the "family physician" mode (FP), "family physician + patient" mode (FPP) and "family physician + patient + family member" mode (FPPF) in the experimental group were 90.1%, 94.4%, and 92.6%, respectively. The sensitivity analysis revealed steady results, with no discernible publication bias. CONCLUSIONS: The FPCS is beneficial to the control of hypertension. The control effect is influenced by average age and intervention mode. The control effect of hypertension in the elderly is better than that in the working-age population, and FPP and FPPF are more beneficial to the management of hypertension than FP. The quality and continuity of FPCS should receive more focus in the future, patient self-management and family support are also essential for managing hypertension.
Assuntos
Hipertensão , Médicos de Família , Humanos , Idoso , Hipertensão/epidemiologia , Hipertensão/terapia , ChinaRESUMO
Regional wind speed prediction is an important spatiotemporal prediction problem which is crucial for optimizing wind power utilization. Nevertheless, the complex dynamics of wind speed pose a formidable challenge to prediction tasks. The evolving dynamics of wind could be governed by underlying physical principles that can be described by partial differential equations (PDE). This study proposes a novel approach called PDE-assisted network (PaNet) for regional wind speed prediction. In PaNet, a new architecture is devised, incorporating both PDE-based dynamics (PDE dynamics) and unknown dynamics. Specifically, this architecture establishes interactions between the two dynamics, regulated by an inter-dynamics communication unit that controls interactions through attention gates. Additionally, recognizing the significance of the initial state for PDE dynamics, an adaptive frequency-gated unit is introduced to generate a suitable initial state for the PDE dynamics by selecting essential frequency components. To evaluate the predictive performance of PaNet, this study conducts comprehensive experiments on two real-world wind speed datasets. The experimental results indicated that the proposed method is superior to other baseline methods.
Assuntos
Redes Neurais de Computação , VentoRESUMO
OBJECTIVE: This study was designed to investigate the effects of hypertensive disorders of pregnancy (HDP) on the complications in very low birth weight (VLBW) neonates. METHODS: We retrospectively included VLBW neonates (<37 weeks) who were delivered by HDP pregnant women with a body weight of < 1,500 g (HDP group) hospitalized in our hospital between January 2016 and July 2021. Gestational age matched VLBW neonates delivered by pregnant women with a normal blood pressure, with a proportion of 1:1 to the HDP group in number, served as normal control. RESULTS: Then we compared the peripartum data and major complications between HDP group and control. The body weight, prelabor rupture of membrane (PROM), maternal age, cesarean section rate, fetal distress, small for gestational age (SGA), mechanical ventilation, RDS, necrotizing enterocolitis (NEC) (≥2 stage), Apgar score at 1 min, and mortality in HDP group showed statistical differences compared with those of the control (all p < 0.05). To compare the major complications among HDP subgroups, we classified the VLBW neonates of the HDP group into three subgroups including gestational hypertension group (n = 72), pre-eclampsia (PE) group (n = 222), and eclampsia group (n = 14), which showed significant differences in the fetal distress, Apgar score at 1 min, SGA, ventilation, RDS and NEC (≥2 stage) among these subgroups (all p < 0.05). Multivariate regression analysis showed that eclampsia and PE were the independent risk factors for SGA and NEC, respectively. CONCLUSION: HDP was associated with increased incidence of neonatal asphyxia, fatal distress, SGA, mechanical ventilation, RDS, NEC and mortality. Besides, eclampsia and PE were independent risk factors for SGA and NEC.
Assuntos
Eclampsia , Hipertensão Induzida pela Gravidez , Doenças do Recém-Nascido , Pré-Eclâmpsia , Recém-Nascido , Gravidez , Humanos , Feminino , Hipertensão Induzida pela Gravidez/epidemiologia , Estudos Retrospectivos , Sofrimento Fetal , Cesárea , Recém-Nascido de muito Baixo Peso , Pré-Eclâmpsia/epidemiologia , Retardo do Crescimento Fetal , Peso Corporal , Peso ao NascerRESUMO
The development of a thermoplastic, biodegradable composite material to replace conventional polymers derived from petroleum was the main area of concentration. Herein, a method for preparing antibacterial, UV-blocking and degradable CNF/Lignin/PBAT composite films (CLP) using cellulose nanofibrils (CNF), lignin, and Poly (butylene adipate-terephthalate) (PBAT) as raw materials by solution casting method was described. With the adding of PBAT, the thermal stability, thermoplastic, mechanical properties were enhanced by improving the compatibility between components. The maximum tensile strength of CLP could reach 189.72 MPa, which increased 25.5 % compared to CNF/Lignin film. The average initial decomposition temperature could reach 321 °C, which was much higher than that of CNF and lignin. At the same time, its good heat-sealing performance made it suitable for practical use. Meanwhile, the composite films had excellent UV resistance and could block over 95 % of UV light. The antibacterial results indicated that the films had a good inhibitory effect on E. coli and S. aureus, with a maximum inhibitory ring diameter of 5.56 and 6.36 mm. In addition, the composite film also had excellent barrier capability to liquid and gas. The prepared film had potential to produce flexible packing, industrial compositing and biomedical fields.
Assuntos
Antibacterianos , Biomassa , Celulose , Escherichia coli , Lignina , Poliésteres , Staphylococcus aureus , Raios Ultravioleta , Antibacterianos/química , Antibacterianos/farmacologia , Lignina/química , Escherichia coli/efeitos dos fármacos , Celulose/química , Poliésteres/química , Staphylococcus aureus/efeitos dos fármacos , Resistência à Tração , Nanofibras/química , Temperatura , Testes de Sensibilidade MicrobianaRESUMO
BACKGROUND: To construct a nomogram for predicting necrotizing enterocolitis (NEC) in preterm infants. METHODS: A total of 4,724 preterm infants who were admitted into 8 hospitals between April 2019 and September 2020 were initially enrolled this retrospective multicenter cohort study. Finally, 1,092 eligible cases were divided into training set and test set based on a 7:3 ratio. A univariate logistic regression analysis was performed to compare the variables between the two groups. Stepwise backward regression, LASSO regression, and Boruta feature selection were utilized in the multivariate analysis to identify independent risk factors. Then a nomogram model was constructed based on the identified risk factors. RESULTS: Risk factors for NEC included gestational diabetes mellitus, gestational age, small for gestational age, patent ductus arteriosus, septicemia, red blood cell transfusion, intravenous immunoglobulin, severe feeding intolerance, and absence of breastfeeding. The nomogram model developed based on these factors showed well discriminative ability. Calibration and decision curve analysis curves confirmed the good consistency and clinical utility of the model. CONCLUSIONS: We developed a nomogram model with strong discriminative ability, consistency, and clinical utility for predicting NEC. This model could be valuable for the early prediction of preterm infants at risk of developing NEC.
RESUMO
Noble metal (Ru, Pd, and Pt) catalyzed preparation of Ni2P/α-Al2O3 was proved by XRD characterization. As is known from H2-TPR experiments, Pd exhibits superior catalytic activity over Ru and Pt. These noble metal catalysts can be uniformly deposited on nickel phosphate precursors, and generate spillover hydrogen. Such a surface catalytic effect promotes the preparation of Ni2P/α-Al2O3, without avoiding the formation of the minor phase Ni12P5. They can also be homogeneously dispersed in nickel phosphate precursors, and boost bulk diffusion of oxygen during reduction process, this is termed the "bulk catalytic effect". It is feasible to produce pure phase Ni2P by the latter preparation method using 0.05 wt% Pd or 0.1 wt% Ru at 768 K. The bulk catalytic effect may be considered as a meaningful option for developing catalytic materials.
RESUMO
Designing anisotropic lignin-based particles and promoting the high-value utilization of lignin have nowadays drawn much attention from scientists. However, systematic studies addressing the self-assembly mechanisms of anisotropic lignin-based particles are scarce. In this work, an interaction including the electrostatic forces and chelating forces between lignin and Ag+ was regulated via carboxymethylation modification. Subsequently, the aggregation morphology of carboxymethylated lignin in a Ag+ solution was observed via SEM. The result showed that a large number of Ag+ intercalated into the lignin molecules when the grafting degree of the carboxyl groups increased from 0.17 mmol/g to 0.53 mmol/g, which caused the lignin molecules to gradually transform from disordered blocks to ordered layers. Dynamics research indicated that the adsorption process of Ag+ in carboxymethylated lignin conforms to the Pseudo-first-order kinetic model. The saturated adsorption amount of Ag+ in the carboxymethylated lignin reached 1981.7 mg/g when the grafting rate of carboxyl groups increased to 0.53 mmol/g, which then fully intercalated into lignin molecules and formed a layered structure. The thermodynamic parameters showed that the thermal adsorption process conforms to the Langmuir model, which indicates that Ag+ is monolayer-adsorbed and intercalated into lignin molecules. Meanwhile, the ΔH values are more than 0, which suggests that this adsorption process is a endothermic reaction and that a higher temperature is conducive to an adsorption reaction. Therefore, self-assembly of lignin in a Ag+ solution under 70 °C is more conducive to the formation of a nanoflower structure, which is consistent with our experimental result. Finally, pH-responsive Pickering emulsions were successfully prepared using a lignin-based nanoflowers, which demonstrated their potential as a catalytic platform in the interface catalysis field. This work offers a deeper understanding into the formation mechanism of anisotropic lignin-based nanoflowers and hopes to be helpful for designing and preparing anisotropic lignin-based particles.
RESUMO
In wind speed prediction technologies, deep learning-based methods have achieved promising advantages. However, most existing methods focus on learning implicit knowledge in a data-driven manner but neglect some explicit knowledge from the physical theory of meteorological dynamics, failing to make stable and long-term predictions. In this paper, we explore introducing explicit physical knowledge into neural networks and propose Physical Equations Predictive Network (PEPNet) for multi-step wind speed predictions. In PEPNet, a new neural block called the Augmented Neural Barotropic Equations (ANBE) block is designed as its key component, which aims to capture the wind dynamics by combining barotropic primitive equations and deep neural networks. Specifically, the ANBE block adopts a two-branch structure to model wind dynamics, where one branch is physic-based and the other is data-driven-based. The physic-based branch constructs temporal partial derivatives of meteorological elements (including u-component wind, v-component wind, and geopotential height) in a new Neural Barotropic Equations Unit (NBEU). The NBEU is developed based on the barotropic primitive equations mode in numerical weather prediction (NWP). Besides, considering that the barotropic primitive mode is a crude assumption of atmospheric motion, another data-driven-based branch is developed in the ANBE block, which aims at capturing meteorological dynamics beyond barotropic primitive equations. Finally, the PEPNet follows a time-variant structure to enhance the model's capability to capture wind dynamics over time. To evaluate the predictive performance of PEPNet, we have conducted several experiments on two real-world datasets. Experimental results show that the proposed method outperforms the state-of-the-art techniques and achieve optimal performance.
Assuntos
Redes Neurais de Computação , Vento , Movimento (Física)RESUMO
Few-shot learning (FSL) aims to recognize novel classes with few examples. Pre-training based methods effectively tackle the problem by pre-training a feature extractor and then fine-tuning it through the nearest centroid based meta-learning. However, results show that the fine-tuning step makes marginal improvements. In this paper, 1) we figure out the reason, i.e., in the pre-trained feature space, the base classes already form compact clusters while novel classes spread as groups with large variances, which implies that fine-tuning feature extractor is less meaningful; 2) instead of fine-tuning feature extractor, we focus on estimating more representative prototypes. Consequently, we propose a novel prototype completion based meta-learning framework. This framework first introduces primitive knowledge (i.e., class-level part or attribute annotations) and extracts representative features for seen attributes as priors. Second, a part/attribute transfer network is designed to learn to infer the representative features for unseen attributes as supplementary priors. Finally, a prototype completion network is devised to learn to complete prototypes with these priors. Moreover, to avoid the prototype completion error, we further develop a Gaussian based prototype fusion strategy that fuses the mean-based and completed prototypes by exploiting the unlabeled samples. At last, we also develop an economic prototype completion version for FSL, which does not need to collect primitive knowledge, for a fair comparison with existing FSL methods without external knowledge. Extensive experiments show that our method: i) obtains more accurate prototypes; ii) achieves superior performance on both inductive and transductive FSL settings.
RESUMO
The multi-target data association method is studied in order to realize multi-target tracking in infrared fish-eye warning system. The Neural Joint Probabilistic Data Association (NJPDA) algorithm is analyzed. It is found that the NJPDA algorithm only considers the distance information between the measurement and the target in the data association process, and its tracking accuracy needs to be further improved. Therefore, a new method fused with direction information is proposed based on the NJPDA algorithm. The proposed algorithm defines the concept of direction difference, introduces the direction information of target motion, and modifies the likelihood function by Gaussian weighting method, so as to fuse the direction information of target motion into the calculation of data interconnection probability. Experimental results demonstrate that the tracking success rate of the proposed algorithm is nearly 10 % higher than that of JPDA and NJPDA algorithms and its consuming time meets the real-time requirement of the infrared fish-eye warning system.
RESUMO
Few-Shot Molecular Property Prediction (FSMPP) is an improtant task on drug discovery, which aims to learn transferable knowledge from base property prediction tasks with sufficient data for predicting novel properties with few labeled molecules. Its key challenge is how to alleviate the data scarcity issue of novel properties. Pretrained Graph Neural Network (GNN) based FSMPP methods effectively address the challenge by pre-training a GNN from large-scale self-supervised tasks and then finetuning it on base property prediction tasks to perform novel property prediction. However, in this paper, we find that the GNN finetuning step is not always effective, which even degrades the performance of pretrained GNN on some novel properties. This is because these molecule-property relationships among molecules change across different properties, which results in the finetuned GNN overfits to base properties and harms the transferability performance of pretrained GNN on novel properties. To address this issue, in this paper, we propose a novel Adaptive Transfer framework of GNN for FSMPP, called ATGNN, which transfers the knowledge of pretrained and finetuned GNNs in a task-adaptive manner to adapt novel properties. Specifically, we first regard the pretrained and finetuned GNNs as model priors of target-property GNN. Then, a task-adaptive weight prediction network is designed to leverage these priors to predict target GNN weights for novel properties. Finally, we combine our ATGNN framework with existing FSMPP methods for FSMPP. Extensive experiments on four real-world datasets, i.e., Tox21, SIDER, MUV, and ToxCast, show the effectiveness of our ATGNN framework.