Search

The widely employed tiny neural networks (TNNs) in mobile devices are vulnerable to adversarial attacks. However, more advanced research on the robustness of TNNs is highly in demand. This work focuses on improving the robustness of TNNs without sacrificing the model's accuracy. To find the optimal trade-off networks in terms of the adversarial accuracy, clean accuracy, and model size, we present TAM-NAS, a tiny adversarial multi-objective one-shot network architecture search method. First, we build a novel search space comprised of new tiny blocks and channels to establish a balance between the model size and adversarial performance. Then, we demonstrate how the supernet facilitates the acquisition of the optimal subnet under white-box adversarial attacks, provided that the supernet significantly impacts the subnet's performance. Concretely, we investigate a new adversarial training paradigm by evaluating the adversarial transferability, the width of the supernet, and the distinction between training subnets from scratch and fine-tuning. Finally, we undertake statistical analysis for the layer-wise combination of specific blocks and channels on the first non-dominated front, which can be utilized as a design guideline for the design of TNNs. [ABSTRACT FROM AUTHOR]

Objectives: This study aimed to explore the feasibility of low-dose computed tomography (CT)–based mixed reality and its clinical role in type A aortic dissection (TAAD) operations. Methods: Eighty-seven patients diagnosed with TAAD were prospectively enrolled and underwent thoracoabdominal aorta mixed reality. They were randomly divided into a low-dose mixed reality group, a conventional mixed reality group and a conventional thoracoabdominal aorta computed tomography angiography (CTA) group. Three-dimensional modelling, mixed reality and CT reconstruction technology were selected. The radiation dose and image quality were compared using Student's t test. Doctors with different seniorities evaluated the clinical application value of thoracoabdominal aorta mixed reality using a Likert scale. The consistency was assessed using the Cohen kappa coefficient (k). The Pearson chi-square test was used to test the correlation of perioperative index results in TAAD operations. Results: Low-dose CT technology can be effectively applied to thoracoabdominal aorta mixed reality and reduces the radiation dose by approximately 59% and the operation time and auxiliary cardiopulmonary bypass time by approximately 22% and 29%, respectively. The subjective scores of doctors with different seniorities on the clinical application value of thoracoabdominal aorta mixed reality were higher than those of thoracoabdominal aorta CTA (all p > 0.05). Conclusions: Low-dose CT can be effectively used in thoracoabdominal aortic mixed reality to reduce the radiation dose while ensuring quality. Low-dose thoracoabdominal aortic mixed reality has clinical application value and can effectively reduce the operation time and auxiliary cardiopulmonary bypass time in TAAD operations. Key Points: • Low-dose CT technology can ensure the mixed reality quality of the thoracoabdominal aorta with a radiation dose reduction of approximately 59%. • Compared with thoracoabdominal aorta CTA, low-dose thoracoabdominal aorta mixed reality can reduce the operation time and auxiliary cardiopulmonary bypass time by approximately 20% and 29%, respectively, in TAAD operations. • The application value of low-dose thoracoabdominal aorta mixed reality in operation scheme formulation, operation risk assessment, operation navigation and diagnosis and treatment under safe distance was greater than that of thoracoabdominal aorta CTA in TAAD. [ABSTRACT FROM AUTHOR]

Purpose: To investigate the dose–effect relationship between the dose–volume parameters of residual gross tumor volume (GTVres) and clinical prognosis in MRI image-guided adaptive brachytherapy (IGABT) of patients with locally advanced cervical cancer in our center. Materials and method: The clinical data of 93 patients with locally advanced cervical squamous cell cancer who received external beam radiotherapy (EBRT) combined with IGABT ± chemotherapy in our center were retrospectively analyzed. The disease stage, overall treatment time (OTT), chemotherapy, and the dose–volume parameters D90, D98, and D100 of GTVres, the intermediate-risk clinical target volume (CTVIR), and the high-risk clinical target volume (CTVHR) of the patients were statistically analyzed. Kaplan–Meier and uni- and multivariable Cox regression analyses were used to analyze 2‑year overall survival (OS), progression-free survival (PFS), and local control rate (LC). A probit model was employed to assess the dose–effect relationship between the volume and dose–volume parameters of GTVres and 2‑year OS, PFS, and LC. Results: The median follow-up time was 19.6 months and 2‑year OS, PFS, and LC were 79.6%, 68.8%, and 94.6%, respectively. CTVHR D90 was an independent influencing factor for 2‑year PFS (P = 0.041); GTVresBT1 volume was an independent factor for 2‑year OS, PFS, and LC (P < 0.001). The probit model showed that at GTVresBT1 volume < 32.86 cm3, the expected 2‑year LC was > 90%; at GTVres D98 > 129.12 GyEQD2, the expected 2‑year OS was > 90%. Conclusion: Both the volume and dose–volume parameters of GTVres are promising predictors in assessment of IGABT prognosis of cervical cancer. [ABSTRACT FROM AUTHOR]

Clique, as the most compact cohesive component in a graph, has been employed to identify cohesive subgroups of entities, and explore the sensitive information in the online social network, crowdsourcing network, and cyber-physical network, etc. In this study, we focus on the defense of clique-based attack and target at reducing the risk of entities security/privacy issues in clique structure. Since the ultimate resolution for defending the clique-based attack and risk is wrecking the clique with minimum cost, we establish the problem of clique-destroying in the network from a fundamental algorithm aspect. Interestingly, we notice that the clique-destroying problem in the directed graph is still an unsolved problem, and complexity analysis also does not exist. Therefore, we propose an innovative formal clique-destroying problem and proof the NP-complete problem complexity with solid theoretical analysis, then present effective and efficient algorithms for both undirected and directed graph. Furthermore, we show how to extend our algorithm to data privacy protection applications with controllable parameter k, which could adjust the size of a clique we wish to destroy. By comparing our algorithm with the up-to-date anonymization approaches, the real data experiment demonstrates that our resolution could efficaciously defend the clique-based security and privacy attacks. [ABSTRACT FROM AUTHOR]

Postoperative intestinal ileus is common after laparoscopic surgery, the incidence of those after hysterectomy was 9.2%. Anesthesia is one of the independent risk factors of postoperative ileus. Dexmedetomidine has been widely used in perioperative anesthesia and previous reports suggested that intraoperative dexmedetomidine may be associated with the improvement of gastrointestinal function recovery after abdominal surgery. We hypothesized that dexmedetomidine could improve gastrointestinal function recovery after laparoscopic hysteromyomectomy. Participants in elective laparoscopic hysteromyomectomy were enrolled with a single dose of 0.5 μg kg−1 dexmedetomidine or the same volume of placebo intravenously administered for 15 min, followed by continuous pumping of 0.2 μg kg−1 h−1 of corresponding drugs until 30 min before the end of surgery. The primary outcome was the time to first flatus. Secondary outcomes were the time to first oral feeding and the first defecation, the occurrence of flatulence, pain score and postoperative nausea and vomiting until 48 h after the surgery. Eventually, 106 participants (54 in dexmedetomidine group and 52 in placebo group) were included for final analysis. The time to first flatus (SD, 25.83 [4.18] vs 27.67 [3.77], P = 0.019), oral feeding time (SD, 27.29 [4.40] vs 28.92 [3.82], P = 0.044), the time to first defecation (SD, 59.82 [10.49] vs 63.89 [7.71], P = 0.025), abdominal distension (n%, 12 (22.2) vs 21 (40.4), P = 0.044), PONV at 24 h (n%, 10 (18.5) vs 19 (36.5), P = 0.037), NRS 6 h (3.15(0.68) vs 3.46 (0.87), P = 0.043) and NRS 12 h (3.43 (0.88) vs 3.85 (0.85), P = 0.014) of dexmedetomidine group were significantly shorter than those of the placebo group. Intraoperative dexmedetomidine reduced the time to first flatus, first oral feeding, and first defecation. These results suggested that this treatment may be a feasible strategy for improving postoperative gastrointestinal function recovery in patients undergoing laparoscopic hysteromyomectomy. [ABSTRACT FROM AUTHOR]

The design of CRISPR gRNAs requires accurate on-target efficiency predictions, which demand high-quality gRNA activity data and efficient modeling. To advance, we here report on the generation of on-target gRNA activity data for 10,592 SpCas9 gRNAs. Integrating these with complementary published data, we train a deep learning model, CRISPRon, on 23,902 gRNAs. Compared to existing tools, CRISPRon exhibits significantly higher prediction performances on four test datasets not overlapping with training data used for the development of these tools. Furthermore, we present an interactive gRNA design webserver based on the CRISPRon standalone software, both available via https://rth.dk/resources/crispr/. CRISPRon advances CRISPR applications by providing more accurate gRNA efficiency predictions than the existing tools. High-quality gRNA activity data is needed for accurate on-target efficiency predictions. Here the authors generate activity data for over 10,000 gRNA and build a deep learning model CRISPRon for improved performance predictions. [ABSTRACT FROM AUTHOR]

Region-based Convolutional Neural Network (R-CNN) detectors have achieved state-of-the-art results on various challenging benchmarks. Although R-CNN has achieved high detection performance, the research of local information in producing candidates is insufficient. In this paper, we design a Keypoint-based Faster R-CNN (K-Faster) method for object detection. K-Faster incorporates local keypoints in Faster R-CNN to improve the detection performance. In detail, a sparse descriptor, which first detects the points of interest in a given image and then samples a local patch and describes its invariant features, is first employed to produce keypoints. All 2-combinations of the produced keypoints are second selected to generate keypoint anchors, which are helpful for object detection. The heterogeneously distributed anchors are then encoded in feature maps based on their areas and center coordinates. Finally, the keypoint anchors are coupled with the anchors produced by Faster R-CNN, and the coupled anchors are used for Region Proposal Network (RPN) training. Comparison experiments are implemented on PASCAL VOC 07/12 and MS COCO. The experimental results show that our K-Faster approach not only increases the mean Average Precision (mAP) performance but also improves the positioning precision of the detected boxes. [ABSTRACT FROM AUTHOR]

The diagnosis of drowning is one of the most difficult in forensic medicine. Forensic diatomology has been proposed to be useful in solving the diagnosis of drowning and considered to be a reliable indicator of the site of drowning. The Yangtze River and Jialing River are the main rivers in the Chongqing area (China), and a large number of corpses are found in the rivers every year. However, the distribution of diatoms in the rivers was not fully studied. In the presented study, a Microwave Digestion-Vacuum Filtration-Scanning Electron Microscopy (MD-VF-SEM) method was performed to acquire the qualitative and quantitative data of diatoms of water samples collected from 10 different sites of the Yangtze River and Jialing River in Chongqing section during different seasons. Our study not only created the diatomological maps of water bodies in Chongqing section of the Yangtze River and Jialing River for the first time but also identified some seasonal and site-specific diatoms that can be taken as markers of particular sites or seasons of drowning. The results of our study may provide forensic scientists helpful reference in solving the drowning cases. [ABSTRACT FROM AUTHOR]

This study evaluated the Pb(II) sorption capacity of jujube pit biochar (JPB) in aqueous solution, which was derived from jujube pit by pyrolysis and used as a promising and economical adsorbent. More importantly, the utilization of JPB could realize the recycling of agricultural residues. The JPB was characterized using conventional science technologies, including SEM, BET and FT-IR, and the sorption capacity of JPB for lead ions was investigated according to different adsorption parameters, such as the kinetics data, solution pH, isotherms data, coexisting ions of Na+ and K+, desorption and reusability, and solution temperature. The results of kinetics data suggested that the lead ion adsorption process by JPB could be fast to reach equilibrium within 30 min. Additionally, the adsorption capacity of JPB for Pb(II) was calculated to be maximum for 137.1 mg/g at pH 6.0. More importantly, after five cycles of desorption and reuse, the JPB still reached 70% of its original adsorption capacity. All the results suggested that JPB had a broad application prospect for the purification of lead ions in practical. [ABSTRACT FROM AUTHOR]

A fractional Cattaneo model is derived for studying the heat transfer in a finite slab irradiated by a short pulse laser. The analytical solutions for the fractional Cattaneo model, the classical Cattaneo-Vernotte model, and the Fourier model are obtained with finite Fourier and Laplace transforms. The effects of the fractional order parameter and the relaxation time on the temperature fields in the finite slab are investigated. The results show that the larger the fractional order parameter, the slower the thermal wave. Moreover, the higher the relaxation time, the slower the heat flux propagates. By comparing the fractional order Cattaneo model with the classical Cattaneo-Vernotte and Fourier models, it can be found that the heat flux predicted using the fractional Cattaneo model always transports from the high temperature to the low one, which is in accord with the second law of thermodynamics. However, the classical Cattaneo-Vernotte model shows that the unphysical heat flux sometimes transports from the low temperature to the high one. [ABSTRACT FROM AUTHOR]

The complete chloroplast genome sequence of a wild chestnut species, Castanea mollissima Blume, was determined and mapped based on Illumina sequencing data. The complete chloroplast genome is 160,869 bp and comprises a pair of inverted repeat regions of 25,701 bp each, a large single-copy region of 90,497 bp, and a small single-copy region of 18,970 bp. The genome harbors 131 genes, including 86 protein-coding genes, 8 ribosomal RNA genes, and 37 transfer RNA genes. Phylogenetic analysis based on chloroplast genomes indicates that wild C. mollissima is closely related to Castanopsis echinocarpa and dissimilar to cultivated C. mollissima. [ABSTRACT FROM AUTHOR]

Nowadays, location-based services (LBS) are facilitating people in daily life through answering LBS queries. However, privacy issues including locationprivacy and queryprivacy arise at the same time. Existing works for protecting queryprivacy either work on trusted servers or fail to provide sufficient privacy guarantee. This paper combines the concepts of differential privacy and k-anonymity to propose the notion of differentially private k-anonymity (DPkA) for queryprivacy in LBS. We recognize the sufficient and necessary condition for the availability of 0-DPkA and present how to achieve it. For cases where 0-DPkA is not achievable, we propose an algorithm to achieve 휖-DPkA with minimized 휖. Extensive simulations are conducted to validate the proposed mechanisms based on real-life datasets and synthetic data distributions. [ABSTRACT FROM AUTHOR]

A novel shape-stabilized phase change material (PEG/TAMS), fabricated using tannic acid-templated mesoporous silica (TAMS) as a support for polyethylene glycol, was developed for thermal energy storage. The method used to synthesize TAMS was simple, cost effective, environmentally friendly, and free of surfactant. The characterization results indicated that PEG was physically absorbed to TAMS and that TAMS had no influence on the crystal structure of PEG. According to the TGA thermograms, PEG/TAMS has excellent thermal stability and can be applied over a wide temperature range. Additionally, the differential scanning calorimetry results suggested that PEG/TAMS has good thermal properties and that its fusion and solidification enthalpies reached 114.7 J/g and 102.4 J/g, respectively. The results indicated that PEG/TAMS has great potential for practical applications. [ABSTRACT FROM AUTHOR]

In this study, the buckling behaviors of single-walled carbon nanocones (SWCNCs) under bending at finite temperatures are predicted using a proposed multiscale quasi-continuum approach based on the temperature-dependent higher order Cauchy–Born (THCB) rule. The hyper-elastic constitutive model is derived exactly in the context of the higher order gradient theory that incorporates the details of the interatomic interaction. The numerical simulations for the deformation of SWCNCs are implemented using the developed meshless computational framework based on moving least-squares interpolation, which can precisely reproduce the deformation process and buckling patterns of SWCNCs under bending. The underlying correlations of the critical bending angle with respect to the geometry of SWCNCs and temperature are revealed by the numerical results. Furthermore, our simulation captures the transformation from the local to the global buckling process of SWCNCs, accompanied with an average strain energy jump. Our results correspond with previous studies. [ABSTRACT FROM AUTHOR]

The value of form factor k at different drafts is important in predicting full-scale total resistance and speed for different types of ships. In the ITTC community, most organizations predict form factor k using a low-speed model test. However, this method is problematic for ships with bulbous bows and transom. In this article, a Computational Fluid Dynamics (CFD)-based method is introduced to obtain k for different type of ships at different drafts, and a comparison is made between the CFD method and the model test. The results show that the CFD method produces reasonable k values. A grid generating method and turbulence model are briefly discussed in the context of obtaining a consistent k using CFD. [ABSTRACT FROM AUTHOR]

To improve the efficacy of porcine reproductive and respiratory syndrome virus (PRRSV) subunit vaccine, the immunological enhancement effects of four natural adjuvants were compared, including Astragalus extract, Astragalus/Bacillus production, propolis, and Freund's adjuvant. The results showed that titers of IgG and neutralizing antibody against highly pathogenic PRRSV (HP-PRRSV) glycoprotein 5 (GP5) from the groups of Astragalus/Bacillus adjuvant and Freund's adjuvant co-administered with a recombinant HP-PRRSV GP5 subunit were the highest among all the experimental groups. In cellular immunity, as shown in T lymphocyte proliferation level, Astragalus/Bacillus adjuvant and Freund's adjuvant exhibited the most potent immunological enhancement effects. Exceptionally, pigs inoculated with Freund's adjuvant developed severely delayed humoral immune responses within 28 days post inoculation. The adjuvants of Astragalus extract and propolis also exhibited immunological enhancement effects on humoral immune and cell-mediated immune responses, but which were obviously lower than Astragalus/Bacillus adjuvant and Freund's adjuvant. Following challenge with HP-PRRSV, pigs inoculated with Astragalus/Bacillus adjuvant and Freund's adjuvant co-administered with GP5 subunit showed lower viremia, slighter clinical signs, and less pathological lung lesions compared with the groups inoculated using Astragalus extract and propolis. In conclusion, from immune responses and immunological protection against HP-PRRSV, the Astragalus/ Bacillus adjuvant demonstrated the most potent immunological enhancement effect on HP-PRRSV GP5 subunit. Astragalus/ Bacillus adjuvant showed promise as a candidate immune adjuvant, to more efficiently prevent and control PRRS. [ABSTRACT FROM AUTHOR]

Porcine circovirus type 2 (PCV2), the essential causative agent of postweaning multisystemic wasting syndrome (PMWS), can be divided into distinct genotypes. Thirty-two PCV2 isolates obtained made from pigs in Shandong Province between 2005 and 2013. Complete genome sequences were obtained in three replicates for each virus isolate, and the sequences were submitted to the NCBI database. The ORF1-encoded amino acid sequences had 98.4 %-100 % identity among the 32 isolates, and there were no significant differences among the three potential glycosylation positions: aa 23-25 (NPS), aa 256-258 (NQT) and aa 286-288 (NAT). The amino acid sequences of ORF2 had 88 %-100 % identity among the 32 isolates and the potential glycosylation position in the cap protein, aa 143-145 (NYS), had no variation. Phylogenetic analysis revealed that the PCV2b/1C genetic lineage was prevalent in swine populations in Shandong Province. It also suggested that selection pressure has made the PCV2 isolates more genetically distant from current vaccine strains. [ABSTRACT FROM AUTHOR]

Four Newcastle disease virus (NDV) isolates were obtained from 997 fecal and tissue samples were collected in 2011 from seafowl that included seagull, sea duck, and swan from the coastal areas of Guangdong, Jiangsu, and Shandong in China. These isolates (SD1, SD2, GD1, and JS1) were characterized for their pathogenicity according to their mean death time, intracerebral pathogenicity index and intravenous pathogenicity index. Full-length fusion protein genes containing the cleavage site were sequenced, and amino-acid sequences around the cleavage site were deduced. One isolate (SD2) was virulent to poultry as indicated by its mean death time, intracerebral pathogenicity index, and fusion gene cleavage site sequence, which was specific for virulent NDV (R-R-Q-K-R-F). The phylogenic analysis indicated that three of the isolates (SD1, GD1, and JS1) belonged to genotype II and the virulent isolate (SD2) belonged to genotype VIId. These findings suggest that some seafowl NDVs in the coastal areas of China have different virulences and molecular characterizations, and these NDVs have some similarity with vaccine- or poultry-adapted isolates. [ABSTRACT FROM AUTHOR]

Porcine reproductive and respiratory syndrome virus (PRRSV) has become one of the most economically important diseases to the global pork industry. Current vaccination strategies only provide a limited protective efficacy. In this study, a DNA vaccine, pVAX1-α-γ-GP35, co-expressing GP3 and GP5 of PRRSV with interferon α/γ was constructed, and its immediate and long-lasting protection against highly pathogenic PRRSV (HP-PRRSV) challenge were examined in pigs. For immediate protection, the results showed that pVAX1-α-γ-GP35 could provide partially protective efficacy, which was similar to the pVAX1-α-γ (expressing interferon α/γ). For long-lasting protection, pigs inoculated with pVAX1-α-γ-GP35 developed significantly higher PRRSV-specific antibody response, T cell proliferation, IFN-γ, and IL-4, than those vaccinated with pVAX1-GP35 (expressing GP3 and GP5 of PRRSV). Following homologous challenge with HP-PRRSV strain SD-JN, pigs inoculated with pVAX1-α-γ-GP35 showed almost no clinical signs, no lung lesions, and significantly lower viremia, as compared to those in pVAX1-GP35 group. It indicated that pVAX1-α-γ-GP35 could induce enhanced immune responses and provide both immediate and long-lasting protection against HP-PRRSV challenge in pigs. The DNA vaccine pVAX1-α-γ-GP35 might be an attractive candidate vaccine for the prevention and control of HP-PRRSV infections. [ABSTRACT FROM AUTHOR]

Internet of Things (IoT) is rapidly developed and widely deployed in recent years, which makes the sensory data generated by IoT systems explode. The huge amount of sensory data generated by some IoT systems has already exceeded the storage, transmission, and computation capacities of IoT systems. However, the valuable sensory data which is highly related to a query in an IoT system is relatively small. The sensory data which is highly related to a query Q forms the relative kernel dataset of Q. Therefore, retrieving sensory data in the relative kernel dataset of a query instead of the raw sensory data will reduce the heavy burdens of an IoT system in terms of transmission and computation and then reduce the energy consumption of the IoT system. In this paper, we investigate the problem of retrieving relative kernel dataset from big sensory data for continuous queries in IoT systems. Two algorithms, relative kernel dataset retrieving algorithm and piecewise linear fitting-based relative kernel dataset retrieving algorithm, are proposed to retrieve the relative kernel dataset for continuous queries. Beside, algorithms for estimating the answers of continuous queries based on their relative kernel datasets are also proposed. Extensive simulation results are provided to verify the effectiveness and energy efficiency of the proposed algorithms. [ABSTRACT FROM AUTHOR]