Search

As climate change intensifies, extreme weather events like floods are occurring with increasing frequency. While data-driven deep learning methods are effective for extracting flood disaster information, their efficiency is constrained by the scarcity of postdisaster samples, the high cost of annotations, and the models’ strong dependence on both the quantity and quality of data. This study introduces an interactive semantic segmentation model based on multisource UAV flood images, incorporating four types of prompts. By embedding expert knowledge into the prompt design, the model reduces annotation costs and enhances generalization capabilities. First, a prompt encoder is developed to map different types of prompt information into a three-channel space using convolutional techniques, thereby reducing sample labeling costs. Moreover, an image encoder that integrates Mamba and convolution is developed to effectively extract global spatial and channel features from flood images while minimizing computational load. Finally, a spatial and channel attention module with residual connections is introduced to enable multiscale fusion and filtering of prompt information and image features across both spatial and channel dimensions, improving the utilization of prompt information. To validate the model's performance, we conduct experiments using UAV flood imagery collected from diverse regions, backgrounds, and angles. The results demonstrate that, under consistent prompt conditions, our model extracts flood areas more efficiently, reducing misclassification and omission errors. Compared with the next best benchmark model, the intersection over union for the flood category improves by at least 3.75%.

Designed to enhance the privacy of user transactions, mixed coin technology has created disruptions to the address clustering rules typically used for cryptocurrency regulation.Consequently, hackers have begun utilizing mixed coin technology as a tool for money laundering and fund evasion, which has raised concerns among financial regulators regarding the detection of mixed coin transactions.Currently, most detection methods for mixed coin transactions rely on data analysis and statistics, lacking a comprehensive understanding of the internal workings of these transactions.This lack of knowledge may undermine the credibility and effectiveness of detection methods due to the absence of reliable verification data.CoinJoin, a decentralized mixed coin concept, represents one approach, and commercial implementations like Wasabi have gained popularity.Drawing from the characteristics of CoinJoin and its restriction on the size of anonymous transaction sets and mixed coin amounts, a general detection method for CoinJoin mixed coin transactions was devised.Such transactions typically involved multiple inputs and outputs, with more output items than UTXOs in the input, and a high occurrence of duplicate values among the output amounts.A basic detection method for Wasabi was developed by combining the generic detection method for CoinJoin with specific features of Wasabi, as identified in related studies, to complete the detection process.A trusted validation dataset was acquired from the Wasabi platform service interface, and this dataset was analyzed to achieve two objectives.First, the alignment of rule parameters in the Wasabi base detection method was accomplished.Second, a new metric was proposed, measuring the ratio of the highest frequency of duplicate values in the output amount of transactions to the number of UTXOs in the input.This metric assessed the level of user participation in mixed coin transactions, providing a measure of user freedom.Using these two approaches, significant progress is made in the detection of mixed coin transactions.The experiments show that the recall rate of Wasabi’s basic detection method is 94.2% and the accuracy rate is 67.2%.After the analytical feedback from the credible validation dataset, the recall rate of the improved detection method reaches 100% and the accuracy rate is above 99%.The total market size of the entire CoinJoin type of mixed coin transactions was evaluated and predicted based on a common test methodology.It is concluded that the number of CoinJoin mixed coin transactions in today’s mixed coin market represents 1.9 per 1 000 of all Bitcoin transactions and 3.7 per 1 000 of the transaction value at most.

Abstract Background Dental pulp stem cells (DPSCs) are adult stem cells with multi-directional differentiation potential derived from ectoderm. Vitro experiments have shown that adding cytokines can help DPSCs to be transformed from multipotent stem cells to osteoblasts. TGF-β has been proved to have an effect on the proliferation and mineralization of bone tissue, but its effect on the osteogenesis and proliferation of dental pulp stem cells is still uncertain. We aim to determine the effect of TGF-β on the osteogenesis and proliferation of dental pulp stem cells. Methods We have identified studies from the Cochrane Central Register of Controlled Trials, PubMed, Embase, and China national knowledge infrastructure (CNKI) for studies interested in TGF-β and proliferation and differentiation of dental pulp stem cells in the following indicators: A490 (an index for evaluating cell proliferation), bone sialoprotein (BSP), Col plasmid-1 (Col-1), osteocalcin (OCN), runt-related transcription factor 2 (Runx-2); and the number of mineralized nodules. Any language restrictions were rejected. Furthermore, we drew a forest plot for each outcome. We conducted a sensitivity analysis, data analysis, heterogeneity, and publication bias test. We evaluate the quality of each study under the guidance of Cochrane's tool for quality assessment. Results The pooled data showed that TGF-β could promote the proliferation and ossification of dental pulp stem cells. All the included results support this conclusion except for the number of mineralized nodules: TGF-β increases the A490 index (SMD 3.11, 95% CI [0.54–5.69]), promotes the production of BSP (SMD 3.11, 95% CI [0.81–6.77]), promotes the expression of Col-1 (SMD 4.71, 95% CI [1.25–8.16]) and Runx-2 (SMD 3.37, 95% CI [− 0.63 to 7.36]), increases the content of OCN (SMD 4.32, 95% CI [1.20–7.44]) in dental pulp, and has no significant effect on the number of mineralized nodules (SMD 3.87, 95% CI [− 1.76 to 9.51]) in dental pulp stem cells. Conclusions TGF-β promotes the proliferation and osteogenesis of dental pulp stem cells.

[This corrects the article DOI: 10.1371/journal.pone.0263482.].

The combination of Unmanned Aerial Vehicle (UAV) technologies and computer vision makes UAV applications more and more popular. Computer vision tasks based on deep learning usually require a large amount of task-related data to train algorithms for specific tasks. Since the commonly used datasets are not designed for specific scenarios, in order to give UAVs stronger computer vision capabilities, large enough aerial image datasets are needed to be collected to meet the training requirements. In this paper, we take low-altitude aerial image object detection as an example to propose a framework to demonstrate how to construct datasets for specific tasks. Firstly, we introduce the existing low-altitude aerial images datasets and analyze the characteristics of low-altitude aerial images. On this basis, we put forward some suggestions on data collection of low-altitude aerial images. Then, we recommend several commonly used image annotation tools and crowdsourcing platforms for data annotation to generate labeled data for model training. In addition, in order to make up the shortage of data, we introduce data augmentation techniques, including traditional data augmentation and data augmentation based on oversampling and generative adversarial networks.

Background: Due to high invasiveness and heterogeneity, the morbidity and mortality of intrahepatic cholangiocarcinoma (ICC) remain unsatisfied. Recently, the exploration of genomic variants has decoded the underlying mechanisms of initiation and progression for multiple tumors, while has not been fully investigated in ICC.Methods: We comprehensively analyzed 899 clinical and somatic mutation data of ICC patients from three large-scale cohorts. Based on the mutation landscape, we identified the common high-frequency mutation genes (FMGs). Subsequently, the clinical features, prognosis, tumor mutation burden (TMB), and pharmacological landscape from patients with different mutation carriers were further analyzed.Results: We found TP53 and KRAS were the common FMGs in the three cohorts. Kaplan–Meier survival curves and univariate and multivariate analysis displayed that TP53 and KRAS mutations were associated with poor prognosis. Considering the co-mutation phenomenon of TP53 and KRAS, we stratified patients into “Double-WT,” “Single-Hit,” and “Double-Hit” phenotypes by mutation status. Patients with the three phenotypes showed significant differences in the mutation landscape. Additionally, compared with “Double-WT” and “Single-Hit” phenotypes, patients with “Double-Hit” presented a dismal prognosis and significantly high TMB. Through chemotherapy sensitivity analysis, we identified a total of 30 sensitive drugs for ICC patients, of which 22 were drugs sensitive to “Double-WT,” 7 were drugs sensitive to “Double-Hit,” and only one was a drug sensitive to “Single-Hit.”Conclusion: Our study defined a novel mutation classification based on the common FMGs, which may contribute to the individualized treatment and management of ICC patients.

This study aims to analyze the balance and coordination during green and sustainable development and estimates the evolution track of the green development system from 2010-2019. The projection pursuit model (PPM), as well as the system gray prediction model GM (1, N) and other measurement methods, were used to construct an evaluation system accounting for green fortune, growth, and benefits to analyze the temporal evolution, developmental trends, and influencing factors of the ecological engineering construction area in Baoshan, Yunnan Province. The results revealed a relatively good green development system, with an upward trend, an average growth rate of 18.3%, and a synergistic coupling effect among the three subsystems. Regional green development has achieved remarkable progress, but challenges and pressure among the three subsystems remain. The prediction analysis also showed that the green development index in Baoshan will continue to increase over the next two years, suggesting that the regional green development system is effective. Green development is primarily affected by environmental pollution, industrial structure, urbanization, population, market, and other factors from the three subsystems of ecology, economy, and social benefits. This study provides practical reference values for prompting regional ecological advancement and green development, along with regional support for the post-2020 Global Biodiversity Framework and the achievement of the United Nations Sustainable Development Goals.

The existing methods of generating adversarial texts usually change the original meanings of texts significantly and even generate the unreadable texts. These less readable adversarial texts can misclassify the machine classifier successfully, but they cannot deceive the human observers very well. In this paper, we propose a novel method that generates readable adversarial texts with some perturbations that can also confuse human observers successfully. Based on the continuous bag-of-words (CBOW) model, the proposed method looks for the appropriate perturbations to generate the adversarial texts through controlling the perturbation direction vectors. Meanwhile, we apply adversarial training to regularize the classification model and extend it to semi-supervised tasks with virtual adversarial training. Experiments are conducted to show that the generated adversaries are interpretable and confused to humans and the virtual adversarial training effectively improves the robustness of the model.

Due to varietal differences, diminutive size, and similar morphological characters, it is difficult to classify and identify Liriomyza spp., a genus comprised of economically-important, highly-polyphagous insect pests. In this study, we reconfirmed the morphological characteristics of three closely-related invasive leafminers, L. trifolii, L. sativae, and L. huidobrensis. Morphological results showed that characteristics imparted by the male genitalia were the most reliable morphological features for identification. The colors exhibited by vertical setae were variable among species, and the ratio of the length of the ultimate section of vein CuA1 divided by penultimate section also varied within species. Although the patterns of abdominal tergites were diverse among Liriomyza spp., L. trifolii exhibited a unique pattern with a yellow patch at the 5th black visible tergite; this pattern can be profiled as a prominent characteristic for morphological identification. In order to identify the three Liriomyza spp. quickly and accurately, we developed an improved molecular identification method using multiplex PCR based on the gene encoding mitochondrial cytochrome oxidase I (COI); this method enabled direct identification based on the size of amplified products. The results of this study provide a valuable reference for the identification of Liriomyza spp., which will ultimately improve our ability to control individual species.

Nowadays, deep learning has been increasingly applied in real-world scenarios involving the collection and analysis of sensitive data, which often causes privacy leakage. Differential privacy is widely recognized in the majority of traditional scenarios for its rigorous mathematical guarantee. However, it is uncertain to work effectively in the deep learning model. In this paper, we introduce the privacy attacks facing the deep learning model and present them from three aspects: membership inference, training data extraction, and model extracting. Then we recall some basic theory about differential privacy and its extended concepts in deep learning scenarios. Second, in order to analyze the existing works that combine differential privacy and deep learning, we classify them by the layers differential privacy mechanism deployed, such as input layer, hidden layer, and output layer, and discuss their advantages and disadvantages. Finally, we point out several key issues to be solved and provide a broader outlook of this research direction.

Differentially private data publication has recently received considerable attention. However, it faces some challenges in differentially private high-dimensional data publication, such as the complex attribute relationships, the high computationalcomplexity and data sparsity. Therefore, we propose PrivMN, a novel method to publish high-dimensional data with differential privacy guarantee. We first use the Markov model to represent the mutual relationships between attributes to solve the problem that the direction of relationship between variables cannot be determined in practical application. We then take advantage of approximate inference to calculate the joint distribution of high-dimensional data under differential privacy to figure out the computational and spatial complexity of accurate reasoning. Extensive experiments on real datasets demonstrate that our solution makes the published high-dimensional synthetic datasets more efficient under the guaranteeof differential privacy.

With the rapid development of mobile telecommunication, voice call spam has become a growing problem in China. Many mobile phone users have become the victim of spam calls and suffered heavy financial loss. Discovering of call spammers can benefit mobile network operators as well as users. Nowadays, the popular method for the task of mining call spammers has been performed by different applications on smartphones. These applications combine manual and automatic methods to detect spammers. Although the results of these client-based solutions are quite satisfying, it is extremely unfortunate that many people still use feature phones, which can not be equipped with third party applications. In this paper, we propose a server-based solution and take a call log file as an example, to analyze the characteristics of mobile call patterns. A time-based graph model and a simple and effective call log rank (CLRank) algorithm with ranking and classification were proposed to find potential call spammers. Compared with existing methods, our model just uses link information, and thus protects user privacy to the maximum extent. Experimental results show that our proposed model can find spammers from call logs automatically, dynamically, and effectively (with 84.5~91.8% of accuracy) without any manual interventions.

This study was conducted to synthesize a series of nanosized BiOI-TiO2 catalysts to photodegrade Bisphenol A solution. The BiOI-TiO2 nanoparticles were synthesized in the reverse microemulsions, consisting of cyclohexane, Triton X-100, n-hexanol, and aqueous salt solutions. The synthesized particles were characterized by X-ray diffraction (XRD), Brunauer-Emmett-Teller (BET) surface analyzer, Fourier transform-infrared spectroscopy (FT-IR), ultraviolet-visible light (UV-Vis) absorption spectra and transmission electron microscope (TEM). The photodegradation of Bisphenol A (BPA) in aqueous suspension under visible light irradiation was investigated to explore the feasibility of using the photocatalytic method to treat BPA wastewater. The effects of different molar ratios of BiOI to TiO2 on the photocatalytic activity were discussed. The experimental results revealed that the photocatalytic effect of the BiOI-TiO2 particles was superior to the commercial P25 TiO2. The BPA degradation could be approached by a pseudo-first-order rate expression. The observed reaction rate constant (kobs) was related to nanoparticles dosage and initial solution pH.

Praseodymium doped Bi2Sn2O7 (BSO), as a visible-light responsive photocatalyst, was prepared by a hydrothermal method with different dopant contents. The as-prepared photocatalysts were investigated by X-ray diffraction (XRD), scanning electron microscope (SEM), transmission electron microscope (TEM), N2 adsorption-desorption isotherm, X-ray photoelectron spectroscopy analysis (XPS), and UV-Vis diffuse reflectance spectroscopy (DRS). The photocatalytic activity of prepared catalysts was evaluated by the degradation of Rhodamine Bextra (RhB) and 2,4-dichlorophenol (2,4-DCP) in aqueous solution under visible light irradiation. It was found that Pr doping inhibited the growth of crystalline size and the as-prepared materials were small in size (10–20 nm). In our experiments, Pr-doped BSO samples exhibited enhanced visible-light photocatalytic activity compared to the undoped BSO, and the optimal dopant amount of Pr was 1.0 mol% for the best photocatalytic activity. On the basis of the calculated PL spectra, the mechanism of enhanced photocatalytic activity has been discussed.

Pyrochlore-type Bi2Sn2O7 (BSO) nanoparticles have been prepared by a hydrothermal method assisted with a cationic surfactant cetyltrimethylammonium bromide (CTAB). These BSO products were characterized by powder X-ray diffraction (XRD), infrared spectroscopy (IR), scanning electron microscopy (SEM), transmission electron microscopy (TEM), Brunauer-Emmett-Teller (BET), and UV-visible diffuse reflectance spectroscopy (DRS). The results indicated that CTAB alters the surface parameters and the morphology and enhances the photoinduced charge separation rate of BSO. The photocatalytic degradation test using rhodamine B as a model pollutant showed that the photocatalytic activity of the BSO assisted with CTAB was two times that of the reference BSO. Close investigation revealed that the size, the band gap, the structure, and the existence of impurity level played an important role in the photocatalytic activities.