Search

Lactate significantly impacts immune cell function in sepsis and septic shock, transcending its traditional view as just a metabolic byproduct. This review summarizes the role of lactate as a biomarker and its influence on immune cell dynamics, emphasizing its critical role in modulating immune responses during sepsis. Mechanistically, key lactate transporters like MCT1, MCT4, and the receptor GPR81 are crucial in mediating these effects. HIF-1α also plays a significant role in lactate-driven immune modulation. Additionally, lactate affects immune cell function through post-translational modifications such as lactylation, acetylation, and phosphorylation, which alter enzyme activities and protein functions. These interactions between lactate and immune cells are central to understanding sepsis-associated immune dysregulation, offering insights that can guide future research and improve therapeutic strategies to enhance patient outcomes. [ABSTRACT FROM AUTHOR]

Co-training is a semi-supervised algorithm that aims to improve prediction effects by exchanging confident instances and pseudo-labels among multiple learners. One central issue is how to mitigate the negative impact of low-quality pseudo-labels during training. In this paper, we propose semi-supervised regression via embedding space mapping and pseudo-label smearing (S2RMS) to ensure that unlabeled data contribute positively to the prediction process. First, a Triplet neural network is trained using pairwise data generated from labeled data. This network maps the training data to the embedding space to better separate dissimilar instances. Second, the embedded data are randomly partitioned into different subsets to train corresponding regression models (a.k.a. regressors). These regressors are integrated into the prediction process. Third, these subsets are augmented using unlabeled data with high similarity to the labeled data and high-confidence pseudo-labels. Here, the similarity and confidence are calculated using the network and the smearing technique, respectively. Experiments are conducted on fourteen datasets, and the results are compared to those of three excellent algorithms. The results show that S2RMS outperforms other co-training and metric semi-supervised regression algorithms. The source code is available at: https://github.com/BetaCatPro/S2RMS. [ABSTRACT FROM AUTHOR]

Lactate significantly impacts immune cell function in sepsis and septic shock, transcending its traditional view as just a metabolic byproduct. This review summarizes the role of lactate as a biomarker and its influence on immune cell dynamics, emphasizing its critical role in modulating immune responses during sepsis. Mechanistically, key lactate transporters like MCT1, MCT4, and the receptor GPR81 are crucial in mediating these effects. HIF-1a also plays a significant role in lactate-driven immune modulation. Additionally, lactate affects immune cell function through post-translational modifications such as lactylation, acetylation, and phosphorylation, which alter enzyme activities and protein functions. These interactions between lactate and immune cells are central to understanding sepsis-associated immune dysregulation, offering insights that can guide future research and improve therapeutic strategies to enhance patient outcomes. [ABSTRACT FROM AUTHOR]

Co-training is a semi-supervised algorithm that aims to improve prediction effects by exchanging confident instances and pseudo-labels among multiple learners. One central issue is how to mitigate the negative impact of low-quality pseudo-labels during training. In this paper, we propose semi-supervised regression via embedding space mapping and pseudo-label smearing (S2RMS) to ensure that unlabeled data contribute positively to the prediction process. First, a Triplet neural network is trained using pairwise data generated from labeled data. This network maps the training data to the embedding space to better separate dissimilar instances. Second, the embedded data are randomly partitioned into different subsets to train corresponding regression models (a.k.a. regressors). These regressors are integrated into the prediction process. Third, these subsets are augmented using unlabeled data with high similarity to the labeled data and high-confidence pseudo-labels. Here, the similarity and confidence are calculated using the network and the smearing technique, respectively. Experiments are conducted on fourteen datasets, and the results are compared to those of three excellent algorithms. The results show that S2RMS outperforms other co-training and metric semi-supervised regression algorithms. The source code is available at: https://github.com/BetaCatPro/S2RMS. [ABSTRACT FROM AUTHOR]

Objectives: Obstructive sleep apnea (OSA) is a common sleep-disordered breathing condition linked to the accelerated onset of mild cognitive impairment (MCI). However, the prevalence of undiagnosed MCI among OSA patients is high and attributable to the complexity and specialized nature of MCI diagnosis. Timely identification and intervention for MCI can potentially prevent or delay the onset of dementia. This study aimed to develop screening models for MCI in OSA patients that will be suitable for healthcare professionals in diverse settings and can be effectively utilized without specialized neurological training. Methods: A prospective observational study was conducted at a specialized sleep medicine center from April 2021 to September 2022. Three hundred and fifty consecutive patients (age: 18--60 years) suspected OSA, underwent the Montreal Cognitive Assessment (MoCA) and polysomnography overnight. Demographic and clinical data, including polysomnographic sleep parameters and additional cognitive function assessments were collected from OSA patients. The data were divided into training (70%) and validation (30%) sets, and predictors of MCI were identified using univariate and multivariate logistic regression analyses. Models were evaluated for predictive accuracy and calibration, with nomograms for application. Results: Two hundred and thirty-three patients with newly diagnosed OSA were enrolled. The proportion of patients with MCI was 38.2%. Three diagnostic models, each with an accompanying nomogram, were developed. Model 1 utilized body mass index (BMI) and years of education as predictors. Model 2 incorporated N1 and the score of backward task of the digital span test (DST_B) into the base of Model 1. Model 3 expanded upon Model 1 by including the total score of digital span test (DST). Each of these models exhibited robust discriminatory power and calibration. The C-statistics for Model 1, 2, and 3 were 0.803 [95% confidence interval (CI): 0.735--0.872], 0.849 (95% CI: 0.788--0.910), and 0.83 (95% CI: 0.763--0.896), respectively. Conclusion: Three straightforward diagnostic models, each requiring only two to four easily accessible parameters, were developed that demonstrated high efficacy. These models offer a convenient diagnostic tool for healthcare professionals in diverse healthcare settings, facilitating timely and necessary further evaluation and intervention for OSA patients at an increased risk of MCI. [ABSTRACT FROM AUTHOR]

Surface wave suppression plays a vital role in enhancing the quality of reflection seismic exploration. Convolutional neural networks (CNNs) can adaptively learn the characteristics of effective signals and surface waves. However, CNN has limited receptive fields and cannot effectively reuse features. When surface waves and effective signals overlap heavily, CNN struggles to preserve effective signals effectively. In this paper, we propose a deformable convolutional wavelet transform network (DCWTN) with residual dense blocks to suppress surface waves. DCWTN contains three types of modules: (1) The deformable convolution module (DCM) is designed to expand the receptive field and enhance seismic events continuity. (2) The wavelet transform enhancement module (WTEM) combines a wavelet transform and a residual dense block to suppress surface waves. It performs multi-scale feature extraction on surface waves according to their time–frequency characteristics to recover detailed information on overlapping parts. (3) The residual dense convolution module (RDCM) is designed for feature enhancement and further refinement of the acquired features. Experimental results show that DCWTN retains more effective signals than four popular methods. [ABSTRACT FROM AUTHOR]

As deep neural networks for visual recognition gain momentum, many studies have modified the loss function to improve the classification performance on long-tailed data. Typical and effective improvement strategies are to assign different weights to different classes or samples, yielding a series of cost-sensitive re-weighting cross-entropy losses. Granted, most of these strategies only focus on the properties of the training data, such as the data distribution and the samples' distinguishability. This paper works these strategies into a weighted cross-entropy loss framework with a simple production form ( WCEL ∏ ), which takes into account different features of different losses. Also, there is this new loss function, misclassification-guided loss (MGL), that generalizes the class-wise difficulty-balanced loss and utilizes the misclassification rate on validation data to update class weights during training. In respect of MGL, a series of weighting functions with different relative preferences are introduced. Both softmax MGL and sigmoid MGL are derived to address the multi-class and multi-label classification problems. Experiments are undertaken on four public datasets, namely MNIST-LT, CIFAR-10-LT, CIFAR-100-LT, ImageNet-LT, and a self-built dataset of 4 main-classes, 44 sub-classes, and altogether 57,944 images, where the results show that on the self-built dataset, the exponential weighting function achieves higher balanced accuracy than the polynomial function does. Ablation studies also show that MGL sees better performance in combination with most of other state-of-the-art loss functions under the WCEL ∏ framework. [ABSTRACT FROM AUTHOR]

The spatial distribution and source analysis of heavy metals in soil are of great significance for the risk assessment and control of heavy metal pollution. Herein, a total of 87 topsoil samples (0-30 cm) were collected in the areas adjacent to a non-ferrous metal slag field of Gejiu City, Yunnan Province. The typical heavy metal concentrations, available phosphorus, available potassium, ammonium nitrogen, nitrate nitrogen, and pH in each soil sample were determined. Based on the above basic data, distribution and origins of heavy metals was analyzed. The results indicated that the mean contents of Pb, As, Cu, Zn, Ni and Cr in the topsoil of the study site were 247.82, 204.36, 104.83, 411.38, 29.73 and 67.59 mg kg-1, respectively. Exception of the Ni element, the contents of the remaining five heavy metals surpassed their respective background levels in study site. Especially, As element (with the highest accumulation effect) is 11.1l times of the background level, and the remaining four elements are 1.04-6.10 times of the background level in Yunnan. Single factor pollution index showed that the pollution degree of Pb, Cu and Zn was light, and the accumulation of As was a great threat to soil quality. Furthermore, the percentages of sites at mild pollution, moderate pollution and heavy pollution levels were 16.09%, 26.09% and 63.22%, respectively. Additionally, The order of potential ecological hazard degree of soil in the survey area was As > Pb > Cu = Ni > Zn = Cr. The spatial dispersion of heavy metals within the study region is impacted by both natural phenomena and human activities. The areas with severe pollution were the mining area in the northwest and planting sites in the east, middle and south of the study area. The main contribution sources of Ni were the mixed sources of natural parent material and industrial activities. The heavy metal element largely contributed by industrial activities and transportation is Pb. The main sources of metals As, Cu and Zn are the integrated sources of agricultural and industrial activities. The main source of Cr is natural source. Therefore, these useful results can guide the heavy metal pollution control and restoration of study site and this research work can also be used to other similar metal slag field and its affected area. [ABSTRACT FROM AUTHOR]

Effective identification of anomalous data from production time series in the oilfield affects future analysis and forecasting. Such time series is often characterized by irregular time intervals due to uneven manual sampling, and missing values caused by incomplete measurements. Therefore, the identification task becomes more challenging. In this paper, an Attention-Embedded Time-Aware Imputation Network (ATIN) with two sub-networks is proposed for this task. First, Time-Aware Imputation LSTM (TI-LSTM) is designed for modeling irregular time intervals and incomplete measurements. It decays the long-term memory component as the producing well conditions may be varied during the water cut stage. Second, Attention-Embedding LSTM (ATEM) is designed to improve the effectiveness of anomaly detection. It focuses on the correlation between the last and historical measurements in a given sequence. Comparison experiments with several state-of-the-art methods, including mTAN, GRU-D, T-LSTM, ATTAIN, and BRITS are conducted. Results show that the proposed ATIN performs better in accuracy, F 1 -score, and area under curve (AUC). [ABSTRACT FROM AUTHOR]

Microbial data exhibit high dimensionality, feature sparseness, and class imbalance. Popular data augmentation strategies typically generate intersecting or overlapping samples with low diversity. In this paper, we propose a two-stage data augmentation method to synthesize high-quality samples. First, we train a feature extractor by minimizing the cross-entropy. Positive training instances are oversampled to achieve class balance. Second, a transformer network is trained for data augmentation to balance diversity and discernibility. A dropout technique is designed to randomly set some feature values as missing. Experiments were carried out on 10 microbial datasets. The results show that: (1) the constructed feature extraction neural network can effectively reduce the dimension of the data, make the data no longer sparse, improve the distinguishability of the samples, and obtain more obvious classification boundaries, and (2) The trained data augmentation transformer network with dropout technique can generate high quality samples, improve the performance of the classifier and reduce the cost of misclassification. [ABSTRACT FROM AUTHOR]

Semi-supervised regression aims to improve the performance of the learner with the help of unlabeled data. Popular approaches select some unlabeled data with high-quality pseudo labels to enrich the training set. In this paper, we propose a new approach with a semi-supervised regressor, a learner, and a respective loss function. First, an off-the-shelf semi-supervised regressor is trained to provide pseudo labels for all unlabeled data. These labels are often reliable enough to guide the learning process. Second, we design a neural network with dropout to train data with Gaussian noise added. In this way, the robustness of our learners is enhanced. Third, we design a weighted sum combining the supervised and unsupervised loss. The weight for pseudo-labels ramp-up over time, indicating more attention to the pseudo-labels. Six state-of-the-art algorithms are employed as the base model of our framework. Results on 15 real-world data sets show that our model has a significant improvement over the respective base regressor on most data sets. [ABSTRACT FROM AUTHOR]

Copyright of Rare Metals is the property of Springer Nature and its content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's express written permission. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)

Recently discovered regular X-ray bursts known as quasi-periodic eruptions have a proposed model that suggests a tidal stripping white dwarf inspiralling into the galaxy's central black hole on an eccentric orbit. According to this model, the interaction of the stripping white dwarf with the central black hole would also emit gravitational wave signals, their detection can help explore the formation mechanism of quasi-periodic eruptions and facilitate multimessenger observations. In this paper, we investigated the horizon distance of TianQin on this type of gravitation wave signal and found it can be set to 200 Mpc. We also find that those stripping white dwarf model sources with central black hole mass within |$10^4 \!-\! 10^{5.5}\, \mathrm{M}_\odot$| are more likely to be detected by TianQin. We assessed the parameter estimation precision of TianQin on those stripping white dwarf model sources. Our result shows that, even in the worst case, TianQin can determine the central black hole mass, the white dwarf mass, the central black hole spin, and the orbital initial eccentricity with a precision of 10−2. In the optimistic case, TianQin can determine the central black hole mass and the white dwarf mass with a precision of 10−7, determine the central black hole spin with a precision of 10−5, and determine the orbital initial eccentricity with a precision of 10−8. Moreover, TianQin can determine the luminosity distance with a precision of 10−1 and determine the sky localization with a precision of 10−2–10  |$\rm deg^2$|⁠. [ABSTRACT FROM AUTHOR]

Deep neural networks (DNNs) are vulnerable to adversarial examples that fool the models with tiny perturbations. Although adversarial attacks have achieved incredible attack success rates in the white-box setting, most existing adversaries often exhibit weak transferability in the black-box setting, especially for models with defense mechanisms. In this work, we reveal the cross-model channel redundancy and channel invariance of DNNs and thus propose two channel-augmented methods to improve the transferability of adversarial examples, namely, the channel transformation (CT) method and the channel-invariant Patch (CIP) method. Specifically, channel transformation shuffles and rewrites channels to enhance cross-model feature redundancy in convolution, and channel-invariant patches distinctly weaken different channels to achieve loss-preserving transformation. We compute the aggregated gradients of the transformed dataset to create adversaries with higher transferability. In addition, the two proposed methods can be naturally combined with each other and with almost all other gradient-based methods to further improve performance. Empirical results on the ImageNet dataset demonstrate that our attack methods exhibit higher transferability and achieve higher attack success rates than state-of-the-art gradient-based attacks. Specifically, our attack improves the average attack success rate from 86.9% to 91.0% on normally trained models and from 44.6% to 68.3% on adversarially trained models. [ABSTRACT FROM AUTHOR]

First-arrival picking is a fundamental and challenging task in seismic data processing. Existing algorithms such as the cross-correlation algorithm (CCT) struggle to achieve satisfactory noise immunity while maintaining picking efficiency. In this paper, we propose the first-arrival picking through mathematical morphology and edge detection (FPME) algorithm with three steps. The converting step transforms the original data into a binary image though a filtering technique, where the binary image represents the global energy distribution of high-energy and low-energy samples. The rendering step delimits the signal zone in the binary image through morphological operations, where the signal zone eliminates the noise in the upper part. The picking step detects first arrivals in the signal zone through the edge detection methods, where an evaluation function is designed to select the optimal detection results as first arrivals. Experiments were performed on four field data sets. The results demonstrated that FPME is more stable and reliable than five classic and popular algorithms. [ABSTRACT FROM AUTHOR]

Objective: To assess the causal effect of type 2 diabetes mellitus (T2DM) on male infertility (MI) and erectile dysfunction (ED) by Mendelian randomization (MR) analysis. Methods: Data for T2DM, MI, and ED were obtained from genome-wide association studies (GWAS) involving 298, 957, 73, 479, and 223, 805 Europeans, respectively. We performed univariate MR analysis using MR Egger, Weighted median (WM) and Inverse variance weighted (IVW) methods to assess causal effects among the three. Through the Genotype Tissue Expression (GTEx) database, single-nucleotide polymorphisms (SNPs) that affect the expression levels of T2DM-related genes were located using expression quantitative trait loci (eQTL). Results: MR analysis showed a significant causal relationship between T2DM and ED (WM, OR: 1.180, 95%CI: 1.010-1.378, P = 0.037; IVW, OR: 1.190, 95%CI: 1.084- 1.300, P < 0.001). There is also a significant causal relationship between T2DM and MI (MR Egger, OR: 0.549, 95%CI: 0.317-0.952, P = 0.037; WM, OR: 0.593, 95%CI: 0.400, P = 0.010; IVW, OR: 0.767, 95%CI: 0.600-0.980, P = 0.034). ED may not cause MI (P > 0.05). We also found that rs6585827 corresponding to the PLEKHA1 gene associated with T2DM is an eQTL variant affecting the expression of this gene. Conclusion: T2DM has a direct causal effect on ED and MI. The level of PLEKHA1 expression suppressed by rs6585827 is potentially associated with a lower risk of T2DM. [ABSTRACT FROM AUTHOR]

Background: As there is still no consensus on the treatment of carotid stent thrombosis (CST), we would like to describe our experience with the revascularization of CST by mechanical thrombectomy. Methods: We retrospectively studied patients who underwent mechanical thrombectomy after CST at Xuzhou Municipal First People's Hospital and Xuzhou Central Hospital between January 2020 and November 2022. The results of the procedures, complications, and clinical and imaging follow-up were recorded. Results: A total of six patients were included in this study. The stenosis grade before stent implantation was ≥85% in all patients, and the stenosis length ranged from 7 to 20 mm. Patients experienced CST within 6 days to 45 months after carotid artery stenting (CAS); the median admission on the National Institutes of Health Stroke Scale (NIHSS) at CST was 12 (range 8-25). Mechanical thrombectomy was successfully performed in all patients. There was no periprocedural death, and the modified Rankin Scale (mRS) at the 3-month follow-up was 0-2. All patients showed recovery from their neurological deficits. Conclusion: The treatment of symptomatic CST with mechanical thrombectomy resulted in satisfactory clinical outcomes. This regimen could be effective and safe, and future prospective and randomized studies are warranted. [ABSTRACT FROM AUTHOR]

In the evolving landscape of manufacturing, the integration of intelligent control theory stands as a pivotal advancement, driving both process optimization and the paradigm of smart manufacturing. This review delves into the multifaceted applications of intelligent control theory, emphasizing its role in equipment, operations, and controls optimization. With a focus on three primary methodologies—fuzzy logic, neural networks, and genetic algorithms—the paper elucidates their biological parallels and their significance in simulation, modeling, and optimization. The transformative potential of smart manufacturing, synonymous with Industry 4.0, is also explored, highlighting its foundation in data, automation, and artificial intelligence. Drawing from a comprehensive analysis of recent literature, the review underscores the growing interest in this domain, as evidenced by the surge in publications and citations over the past decade. The overarching aim is to provide contemporary discourse on the applications and implications of intelligent control theory in the realms of process optimization and smart manufacturing. [ABSTRACT FROM AUTHOR]

Label distribution learning (LDL) explicitly models label ambiguity by assigning a real-valued vector with label description degrees to each sample. Most LDL methods only build models on the same feature (sub)space shared by all labels. However, they ignore that each label has its own specific features, and there are some common features among labels. In this paper, we propose a novel LDL (LDL-LDF) algorithm that aims to exploit both label-dependent and common features. First, label-dependent feature reconstruction utilizes thresholding for relevant sample subset identification, density peaks clustering for representative sample selection, and Euclidean distance for feature value calculation. Second, common feature reconstruction follows a similar approach, however, on the whole dataset. Finally, the prediction neural network is composed of several components that serve each label with label-dependent features, one component that serves all labels with common features, and the fusion component. The effectiveness and competitiveness of our algorithm are verified through various experiments comparing seven algorithms on fourteen real-world datasets. [ABSTRACT FROM AUTHOR]

Long-tailed image recognition is a challenging task in real scenes with large-scale data. Popular strategies, such as loss reweighting and data resampling, aim to reduce the model bias toward head classes. Specifically, different loss reweighting approaches explore various endogenous or exogenous measures. In this paper, we study a new endogenous measure called discriminant quality (DQ) by considering validation accuracy and discriminant uncertainty. DQ takes advantage of continuous information over a period of time. It is more robust than instantaneous information because of the mitigation of measuring instability caused by random perturbations during training. Additionally, the weight of each class is automatically rebalanced based on DQ. Consequently, the class weight supports the design of a dynamic updating strategy for the significance of the DQ difference. Experiments on MNIST-LT, CIFAR-100-LT, ImageNet-LT, and Places-LT demonstrated the superiority of DQ over state-of-the-art ones in terms of prediction accuracy. [ABSTRACT FROM AUTHOR]

Due to complex geological conditions and instrument inaccuracy, raw seismic data are often characterized by low resolution. Deep learning is an emerging technique for seismic resolution improvement; however, its performance is often limited by the small amount of labeled data available. In this paper, we design a deep feature mining network (DFMN) to deal with this issue. DFMN has three components: shallow feature extraction block (SFB), deep feature mining block (DFB), and enhanced reconstruction block (ERB). First, the SFB component uses multi-scale kernels to learn rich information from low-resolution data. The convolutions incorporate the benefits of different kernel sizes, which are effective for shallow feature extraction. Second, the DFB component employs a dual-branch network architecture for deep feature mining. The dual-branch network learns more complementary features than a single-branch network, thus alleviating the requirement for large amounts of training data. Third, the ERB component combines the shuffled image and the interpolated image during reconstruction. Interpolated images, incorporating prior knowledge, can provide more contextual information in our model. The results show that DFMN is superior to a traditional upscaling algorithm and other deep learning methods in terms of (1) perceptual effects: more complete structural information, such as texture details; (2) quantitative evaluation indices: higher PSNR and SSIM; and (3) generalization ability: better performance on other data. [ABSTRACT FROM AUTHOR]

Background: Our previous studies found that high-intensity focused ultrasound (HIFU) stimulated tumor-specific T cells in a mouse H22 tumor model, and adoptive transfer of the T cells from HIFU-treated mice could subsequently elicit stronger inhibition on the growth and progression of the implanted tumors. The aim of this study was to investigate the mechanism of T cells from focused ultrasound ablation in HIFU-mediated immunomodulation. Methods: Sixty H22 tumor-bearing mice were treated by either HIFU or sham- HIFU, and 30 naïve syngeneic mice served as controls. All mice were euthanized on day 14 after HIFU and splenic T cell suspensions were obtained in each group. Using an adoptive cell transfer model, a total of 1 × 106 T cells from HIFU treatedmice were intravenously injected into each syngeneic H22 tumor-bearing mouse twice on day 3 and 4, followed by the sacrifice for immunological assessments at 14 days after the adoptive transfer. Results: T cells from HIFU-treated mice could significantly enhance the cytotoxicity of CTLs (p < 0.001), with a significant increase of TNF-a (p < 0.001) and IFN-g secretion (p < 0.001). Compared to control and sham-HIFU groups, the number of Fas ligand+ and perforin+ tumor-infiltrating lymphocytes (TILs) and apoptotic H22 tumor cells were significantly higher (p < 0.001) in the HIFU group. There were linear correlations between apoptotic tumor cells and Fas ligand+ TILs (r = 0.9145, p < 0.001) and perforin+ TILs (r = 0.9619, p < 0.001). Conclusion: T cells from HIFU-treated mice can subsequently mediate cellular antitumor immunity, which may play an important role in the HIFU-based immunomodulation. [ABSTRACT FROM AUTHOR]

The aim of this work is to analyze the clinicopathological characteristics and prognostic factors of patients with nuclear pedigree of esophageal cancer. The clinicopathological data and follow-up information of 3,260 patients from different nuclear pedigree of esophageal cancer who underwent radical resection of esophageal cancer were collected, and the clinicopathological characteristics and prognostic factors of the patients were analyzed. The male to female ratio of 3,260 patients with esophageal cancer was 1.7:1. The diagnosis age was ranged from 32 to 85 (60.2 ± 8.1) years old. About 53.8% of the patients were = 60 years old; About 88.8% of the patients came from the high incidence area of esophageal cancer; About 82.5% of the tumors were located in the middle and lower segments of esophagus; Poor, moderate and well differentiation accounted for 26.6%, 61.9% and 11.5% respectively; The surgical margin accounted for 94.3%; 47.6% of the tumors were shorter than 4 cm in length; Clinicopathological TNM stage (0+I) accounted for 15.2%, and stage II, III and IV accounted for 54.5%, 29.9% and 0.4%, respectively. Cox analysis showed that male, diagnosed age = 60 years, tumor located in neck and upper esophageal segments, poor differentiation, tumor length = 4 cm, and advanced TNM were independent risk factors for the prognosis of patients in nuclear pedigree with esophageal cancer. Gender, diagnosis age, tumor location, degree of differentiation, tumor length and TNM stage are the influencing factors for the prognosis of patients with nuclear pedigree of esophageal cancer, which will provide important data for the future study of esophageal cancer family aggregation. [ABSTRACT FROM AUTHOR]

With the rapid development of industry, oily wastewater has caused serious environmental problems. Highly efficient oil–water separation based on superhydrophobic materials holds promise; however, superhydrophobic surfaces often lose their superhydrophobicity when exposed to hot water. Consequently, oil–water separation under high temperature is rarely explored. Herein, a heat resistant superhydrophobic porous glass membrane is fabricated by the vapor phase growth of a network layer of silicone nanofilaments using economical ethyltrichlorosilane as the only surface modifier. The superhydrophobic porous glass membrane possesses hot water resistance and can be used for highly efficient separation of water from oil under high temperatures, regardless of the water/oil ratio, the composition of the oil, and the temperature. Taking advantage of the superior performance in oil–water separation at high temperatures, we successfully adopted such a superhydrophobic membrane for removing water from condensation reactions such as amidation and esterification reactions. It was demonstrated that both the reaction rate and the conversion can be improved significantly. The idea to use superhydrophobic oil–water separation materials for condensation reactions paves the way for industrial synthesis of amides and esters. [ABSTRACT FROM AUTHOR]

Co-training is a popular semi-supervised learning method. The learners exchange pseudo-labels obtained from different views to reduce the accumulation of errors. One of the key issues is how to ensure the quality of pseudo-labels. However, the pseudo-labels obtained during the co-training process may be inaccurate. In this paper, we propose a safe co-training (SaCo) algorithm for regression with two new characteristics. First, the safe labeling technique obtains pseudo-labels that are certified by both views to ensure their reliability. It differs from popular techniques of using two views to assign pseudo-labels to each other. Second, the label dynamic adjustment strategy updates the previous pseudo-labels to keep them up-to-date. These pseudo-labels are predicted using the augmented training data. Experiments are conducted on twelve datasets commonly used for regression testing. Results show that SaCo is superior to other co-training style regression algorithms and state-of-the-art semi-supervised regression algorithms. [ABSTRACT FROM AUTHOR]

Phytophthora nicotianae causes substantial economic losses in most countries where tobacco is produced. At present, the control of P. nicotianae mainly depends on chemical methods, with considerable environmental and health issues. We investigated the effects of ethanol extracts from Scutellaria baicalensis Georgi (SBG) and Magnolia officinalis (MO). On mycelial growth, sporangium formation, and zoospore release of P. nicotianae. Both extracts inhibited the growth of P. nicotianae, with mycelial growth inhibition rates of 88.92% and 93.92%, respectively, at 40 mg/mL, and EC50 values of 5.39 and 5.74 mg/mL, respectively. The underlying mechanisms were the inhibition of sporangium formation, the reduction of zoospore number, and the destruction of the mycelium structure. At an SBG extract concentration of 16.17 mg/mL, the inhibition rates for sporangia and zoospores were 98.66% and 99.39%, respectively. At an MO extract concentration of 2.87 mg/mL, the production of sporangia and zoospores was completely inhibited. The hyphae treated with the two plant extracts showed different degrees of deformation and damage. Hyphae treated with SBG extract showed adhesion and local swelling, whereas treatment with MO extract resulted in broken hyphae. Mixture of the extracts resulted in a good synergistic effect. [ABSTRACT FROM AUTHOR]

Seismic analysis and interpretation are sensitive to data resolution. High-density acquisition, deconvolution, and inverse $Q$ filtering are traditional methods for resolution improvement. These methods have drawbacks such as strict assumptions and complex parameter solving processes. In this letter, we propose a channel attention U-Net and physical convolution combination (CAUC) algorithm to overcome these limitations. First, we design a channel attention U-Net (CAU) to establish a sophisticated nonlinear relationship between low-frequency data and coarse reflection coefficients. Specifically, we use a channel attention block after each downsampling convolution block to extract important local features of seismic data. Second, we convolve the coarse reflection coefficients with the wavelets to obtain high-resolution data. This physical convolution operation has a solid theoretical foundation. Therefore, our algorithm benefits not only from the learning ability of the neural network but also from the explainability of the physical operation. Both synthetic and field data are employed to verify the validity of the new algorithm. The traditional fast iterative shrinkage-thresholding algorithm (FISTA) and deep convolutional neural network (CNN) are used for comparison. The results show the superiority of the proposed method, especially in enhancing details such as thin layers and continuity. [ABSTRACT FROM AUTHOR]

Deep learning has made substantial progress in remote sensing image segmentation tasks. It usually requires a large number of high-quality annotation maps (i.e., clean labels), which are labor-intensive. In this letter, we propose to use abundant electronic maps (i.e., noisy labels) to supplement a small number of clean labels to solve the problem of massive label production. In addition, we propose a multistage noise supervised framework to prevent noise from deteriorating the performance of deep model. Noise-supervised deep learning framework (NSDI) consists of clean training, weight initialization, and hybrid training stages. In the clean training stage, we train a segmentation model using a small amount of clean labels and remote sensing images to compute the confusion probability matrix. In the weight initialization stage, we use the confusion probability matrix and the prediction probability to calculate the label reliability of the electronic map. In the hybrid training stage, an adaptive weighted loss function based on cross-entropy is used to dynamically update the label reliability. Then we train the model further using electronic maps with the support of the adaptive weighted loss function and label reliability. Experiments were undertaken on 2656 images of 512 × 512 pixels. Ablation studies show that NSDI improves the model robustness and the segmentation quality. [ABSTRACT FROM AUTHOR]

Multi-label learning with missing labels is a challenging task, especially in text classification applications. Existing approaches considering label correlations are effective in recovering missing labels. However, they are often unstable because severely imbalanced positive and negative labels are treated in the same way. In this paper, we propose the self-dependence multi-label learning with a double k label recovery algorithm to address this issue. First, two label count matrices are constructed from the original label matrix from the perspective of positive and negative labels independently. This is done through statistics on the k nearest neighbors according to the input features. Second, positive and negative label matrices are decomposed and recovered using matrix factorization, namely double k (k nearest neighbors and k latent semantics). Third, new features are generated according to the recovered matrices by label concept. Fourth, we exploit the independence of the labels to guide the training process. Extensive experiments and analyses on multiple benchmark data sets illustrate the effectiveness of the proposed method. In addition, with the increase of missing labels, the stability of our algorithm becomes significantly better than the state-of-the-art ones. [ABSTRACT FROM AUTHOR]

Path planning is an important task for robot motion, unmanned aerial vehicle obstacle avoidance, and smart cars. Sampling-based algorithms have achieved significant success in this task. The rapidly random-exploring tree (RRT) is one of the representative algorithms. However, it is exceedingly difficult to strike a balance between path quality and search efficiency. In this paper, we propose an enhanced RRT* with clustering and presearching (ERCP) algorithm to handle this issue. In the clustering phase, we construct an 8-degree undirected graph according to the neighborhood and obstacles. Then, we adopt the Markov clustering technique, which is appropriate for spatial data. The clustering process is efficient since the coarse-grained map considers only points at integer locations. In the presearching phase, we utilize the clustering results to abstract the intact state space as an undirected graph. We employ Dijkstra's algorithm to perform a presearch on the graph to determine the effective sampling area. In the path planning phase, we use RRT* to extend the space-filling tree within the effective sampling regions. Experiments were conducted on ten maps with different levels of obstacle complexity. The results reveal that ERCP can achieve a more convincing balance between path quality and search efficiency than the three state-of-the-art algorithms with little sacrifice. [ABSTRACT FROM AUTHOR]

Microbial data used for the diagnosis of human diseases are highly class imbalanced, that is, there are many more negative instances than positive instances. Popular data augmentation methods are ineffective in handling microbial data because of data sparsity, that is, most feature values are 0. In this paper, we propose a matrix factorization-based approach using three techniques. First, we decompose the matrix into object and feature subspaces using L1 loss to address sparsity. For data augmentation, we use positive vectors in the object subspace to generate a synthetic vector. We multiply it by the feature subspace to obtain a synthetic sample that we label as positive. Then, we control the number of new samples according to the cost factor. This is more reasonable than KMeansSMOTE and AMSCO, which resolve class imbalance. Finally, we filter the synthetic vectors by distance to increase the diversity of the data. Traditional oversampling techniques only increase samples in the original space, whereas our method uses subspaces for augmentation. This better captures the essence of the data and enriches the diversity of samples. We conducted experiments on 10 microbial datasets. The results demonstrated that 1) the new technique generated samples that were more diverse than those generated using the selected resampling approaches, such as AMSCO, and 2) the new method correctly identified more positive samples and minimized the cost of the classification results. Moreover, our method achieved excellent classification results in combination with existing classification algorithms, which demonstrated the effectiveness and robustness of the new technique. [ABSTRACT FROM AUTHOR]

Modeling and predicting player behavior is of the utmost importance in game development and matchmaking. A variety of methods have been proposed to build artificial intelligence (AI), human-like players. However, these human-like players have a limited ability to imitate the behavior of individual players. In this paper, we propose a player behavior imitation method using imitation learning under the framework of meta-learning. A generic behavior model of game players was learned from historical records using adversarial imitation learning. Then, we personalized the policy by imitating the behavior of each individual player. Convolutional neural networks were used to construct the feature extractor of game board states. The experiments were conducted using the Reversi game, and 18,000 game records of different players were used to train the generic behavior model. The behavior of each new player was learned using only hundreds of records. The results demonstrate that our method can be utilized to imitate individual behavior in terms of action similarity well. [ABSTRACT FROM AUTHOR]

Random noise suppression is an important technique to improve the efficiency and accuracy of seismic data processing. Physical denoising methods such as f - x deconvolution and K-SVD have been widely adopted by the industry, while popular learning-based methods such as neural networks have emerged as good alternatives. In this paper, we propose a multi-scale residual dense network (MSRDN) for random noise suppression of seismic raw data. First, the network consists of a shallow feature extraction module, multiple residual blocks and multiple up-sampling modules. They are used for feature extraction, noise learning and size restoration. Second, each residual block is composed of multiple dense blocks. They are designed to alleviate network degradation. Third, dense blocks are tightly connected by multi-scale convolutional layers. They can enhance the regularization effect of the network. The experimental results show that MSRDN is more accurate and stable than previous algorithms. [ABSTRACT FROM AUTHOR]

The data used for microbial-based disease diagnosis are characterized by small sample sizes, imbalanced categories, high dimensionality, and strong sparsity. They pose challenges to machine learning algorithms that aim to achieve good classification performance. In this paper, we propose a two-stage data augmentation method to enhance training data quality. The first stage is feature transformation. We design a KPCA-based method to map microbial data to a low-rank feature space, resulting in cleaner and more efficient data representation. This processing step addresses high dimensionality and strong sparsity in microbial data. The second stage is data augmentation. New synthetic data are obtained by augmenting the positive samples through the GAN. The misclassification cost is used to control the ratio of positive/negative samples in new data. The combination of the augmented data with the original data constitutes a cost-sensitive dataset, which can increase sample diversity while addressing the imbalance problem. This is more reasonable than traditional sampling methods that resolve the class imbalance. We compare the new method with four popular data augmentation algorithms on 12 imbalanced datasets. The experimental results demonstrate that (1) the samples augmented by the proposed algorithm are more diverse than those generated using compared resampling methods, such as SMOTE_ENN, and (2) the proposed algorithm not only achieves the lowest total misclassification cost but also outperforms other methods in terms of F 2 and G-mean metrics. [ABSTRACT FROM AUTHOR]

Spike architecture is an indicative trait of grain yield in common wheat (Triticum aestivum). A segregating population was generated for mapping genes contributing to spike morphometric traits by crossing the two common wheat cultivars 'CItr 17600' with branching spikes and 'Yangmai 18' with normal spikes. A major quantitative trait locus for spike length was mapped to the Q5A region of chromosome 5A. Yangmai 18 carried a Q5Ab allele for short spikes, which harbored one SNP in the last intron, and a 1-bp InDel in the 720-bp fragment from the start codon, compared to Q5Aa in Chinese Spring. CItr 17600 harbored a q5Ab allele for long spikes, which has a 6-bp deletion compared to the reported q5Aa allele that was involved in the binding site of microRNA 172 (miR172). This 6-bp deletion in immediately upstream of this binding site was involved in changes of four amino acids. The natural q5A allele appeared to be rare in common wheat but frequent in tetraploid T. turgidum accessions with branching spikes. The CRISPR/Cas9 technology was used to edit the upstream region involving in the miR172 binding site in Yangmai 18 and identified two independent editing events, one with a 1-bp insertion in Q5A and the other with a 2-bp deletion in Q5D, resulting in several shapes of spikes in the transgenic progeny. In addition to the effects of natural q5A allele and the edited Q5A genes, this study indicated the regeneratability and transformability of Yangmai 18 as an elite cultivar. Altogether, this study provides insight into future modification and engineering of spike architecture in common wheat. [ABSTRACT FROM AUTHOR]

Object recognition and tracking is a challenge for underwater vehicles. Traditional algorithm requires a clear feature definition, which suffers from uncertainty as the variation of occlusion, illumination, season and viewpoints. A deep learning approach requires a large amount of training data, which suffers from the computation. The proposed method is to avoid the above drawbacks. The Siamese Region Proposal Network tracking algorithm using two weights sharing is applied to track the target in motion. The key point to overcome is the one-shot detection task when the object is unidentified. Various complex and uncertain environment scenarios are applied to evaluate the proposed system via the deep learning model's predictions metrics (accuracy, precision, recall, P-R curve, F1 score). The tracking rate based on Siamese Region Proposal Network Algorithm is up to 180 FPS. [ABSTRACT FROM AUTHOR]

Cantharidin and its analogs are potent serine/threonine protein phosphatase inhibitors with rare reports on inhibiting expression of Hepatitis B virus (HBV) DNA. Five imide cantharidin derivatives were synthesized, and their properties were studied. The inhibition efficiency was determined, and preliminary studies on the structure-activity relationships were conducted. The results indicated that the compounds characterized by low Log P values and the cantharidin moieties containing nitrogen exhibited low cytotoxicity and helped inhibit the expression of HBV DNA. Compound 2 and the non-cytotoxic compound 3 significantly inhibited the expression of HBV DNA. The IC50 values recorded for both compounds were 2.0±1.2 and 19.2±2.0 nM, respectively. [ABSTRACT FROM AUTHOR]

Oil pollution results from daily activities and a variety of industries have caused not only severe environmental problems but also wastage of valuable petrochemical resources. Separation based on superwettable materials holds promise; however, practical applications of a single type of superwettable materials were often limited due to their ability in treatment of complicated oil‐water systems. Herein, a Gemini‐type separator was created through the cooperation of two kinds of superwettable sand particles with opposite wettability, i. e. one is superhydrophobic whereas the other is superhydrophilic. Cooperatively by the two types of superwettable sand, consecutive separation and purification of both water and oil phases from complicated oil‐water systems (e. g. water mixed with a lighter or denser oil, water emulsified in oil, oil emulsified in water, and/or a combination of them in one batch) could be achieved with high flux and superior efficiency just in one single operation unit. [ABSTRACT FROM AUTHOR]

The government employs innovation subsidies as a key incentive strategy to promote companies to innovate more technically. This study analyses how innovation subsidies influences the quality of corporate innovation. We create an innovation quality index for pharmaceutical corporations using categorizing data from patent applications submitted by pharmaceutical companies. Using data from 180 listed Chinese pharmaceutical companies between 2010 and 2020, this study proposes a panel regression model to assess the influence of government innovation subsidies on innovation quality, as well as the moderating effect of CEOs' academic capital. How well innovations are subsidized is also affected by the heterogeneity of property rights. Innovation subsidy has a greater and more positive impact on non-SOEs. This article demonstrates that CEOs with academic credentials and executives with ties to the pharmaceutical industry have a variety of moderate effects. The research offers novel suggestions for enhancing business creativity and the innovation subsidy programme. [ABSTRACT FROM AUTHOR]

Conversational recommendation is ubiquitous in e-commerce, while three-way recommendation provides friendly choices for service providers and users. However, their combination has not been studied yet. In this paper, we introduce the three-way conversational recommendation problem and design the hybrid conversational recommendation (HTCR) algorithm to address it. First, a new recommendation problem is defined by considering the man–machine interaction as well as the misclassification and promotion costs. The optimization objective of the problem is to minimize the total cost. Second, a popularity-based technique is designed for user cold-start recommendation, where the user maturity is responsible for deciding when HTCR turns to the second technique. Third, an incremental matrix factorization technique is designed for regular recommendation. It is efficient since only a few rounds of training are needed for newly acquired user feedback. Experiments were carried out on four well-known datasets, including Jester, MovieLens 100K, MovieLens 1M and Yelp. Results demonstrated that our algorithm outperformed state-of-the-art ones in terms of average cost. [ABSTRACT FROM AUTHOR]

Learning how to navigate autonomously in an unknown indoor environment without colliding with static and dynamic obstacles is important for mobile robots. The conventional mobile robot navigation system does not have the ability to learn autonomously. Unlike conventional approaches, this paper proposes an end-to-end approach that uses deep reinforcement learning for autonomous mobile robot navigation in an unknown environment. Two types of deep Q-learning agents, such as deep Q-network and double deep Q-network agents are proposed to enable the mobile robot to autonomously learn about collision avoidance and navigation capabilities in an unknown environment. For autonomous mobile robot navigation in an unknown environment, the process of detecting the target object is first carried out using a deep neural network model, and then the process of navigation to the target object is followed using the deep Q-network or double deep Q-network algorithm. The simulation results show that the mobile robot can autonomously navigate, recognize, and reach the target object location in an unknown environment without colliding with static and dynamic obstacles. Similar results are obtained in real-world experiments, but only with static obstacles. The DDQN agent outperforms the DQN agent in reaching the target object location in the test simulation by 5.06%. [ABSTRACT FROM AUTHOR]