Image processing networks have gained great success in many fields,and thus the issue of copyright protection for image processing networks hasbecome a focus of attention. Model watermarking techniques are widely usedin model copyright protection, but there are two challenges: (1) designinguniversal trigger sample watermarking for different network models is stilla challenge;(2) existing methods of copyright protection based on trigger swatermarking are difficult to resist forgery attacks. In this work, we propose adual model watermarking framework for copyright protection in image processingnetworks. The trigger sample watermark is embedded in the trainingprocess of the model, which can effectively verify the model copyright. And wedesign a common method for generating trigger sample watermarks based ongenerative adversarial networks, adaptively generating trigger sample watermarksaccording to different models. The spatial watermark is embedded intothe model output. When an attacker steals model copyright using a forgedtrigger sample watermark, which can be correctly extracted to distinguishbetween the piratical and the protected model. The experiments show that theproposed framework has good performance in different image segmentationnetworks of UNET, UNET++, and FCN (fully convolutional network), andeffectively resists forgery attacks.
Tumor hypoxia is one of the hostile tumor micro-environment characteristics occurring in solid tumors.This feature is closely related to tumor progression and can negatively impair the effec-tiveness of cancer therapeutics.Recently,various strategies have been developed that enable effi-cient tumor oxygenation,not only enhancing tre-atment outcome of oxygen-consuming cancer therapeutics such as radiotherapy and photody-namic therapy,but also reversing the immunosup-pressive tumor microenvironment to promote cancer immunotherapy.This review focuses on the latest progress in design and fabrication of inno-vative tumor-targeted oxygen nanoshuttles that attenuate tumor hypoxia and enhance cancer therapy.Future perspectives for clinical translation of tumor-targeted oxygen nanoshuttles will be fur-ther discussed.
Liangzhu FengOshra BetzerDanlei TaoTamar SadanRachela PopovtzerZhuang Liu
Needleless electrospinning is a versatile method to produce nanofibers.In particular,the rotary version of this technique has enjoyed widespread use because there is no need to clean the spinneret.The rotation speed is limited by the potential deviation of the jet due to the centrifugal force.Other limitations are due to the fast volatilization of the solvent from the opened spinning system.In order to overcome these drawbacks,here a novel reciprocating system based on a moving spinning-plate is proposed.The spinning process is implemented in a half-closed system with the spinning-plate immersed in the solution tank.When the immersed spinning-plate moves up from the solution tank,multiple jets are ejected from the droplets on the tips of the spinning-plate under the effect of an electric field force.The morphology of the obtained nanofibers has been analyzed by scanning electron microscopy.The results indicated that the obtained fibers are uniform in structures and small in diameters.Both issues of needle clogging and intense solvent evaporation can be mitigated using this alternate approach.
A novel Federated learning classifier is proposed using the Dempster-Shafer(DS)theory for the set-valued classification of industrial IoT Distributed Denial of Service(DDoS)attack traffic.The proposed classifier,referred to as the evidence-based federated learning classifier,employs convolution and pooling layers to extract high-dimensional features of Distributed Denial of Service(DDoS)traffic from the local data of private industrial clients.The characteristics obtained from the various participants are transformed into mass functions and amalgamated utilizing Dempster’s rule within the DS layer,situated on the federated server.Lastly,the set value classification task of attack mode is executed in the expected utility layer.A learning strategy is proposed to update the network parameters in a joint manner according to the weight of mass function value.Experiments on DDoS traffic classification tasks in CIC-DDoS2019 datasets prove that Federated learning,DS layer,and expected utility layer is effective in enhancing the multi-class classification accuracy,especially for those DDoS mixed attack traffic.
Background: In order to uncover the mechanism of significantly reduced insect resistance at the late developmental stage in cotton(Gossypium hirsutum L.),the relationship between boll setting rate under different planting densities and Bacillus thuringiensis(Bt)insecticidal concentrations in the boll wall were investigated in the present study.Two studies were arranged at Yangzhou,China during the 2017–2018 cotton growth seasons.Five planting densities(15000,25000,45000,60000 and 75000 plants per hectare)and the flower-removal treatment were imposed separately on Bt cotton cultivar Sikang3 to arrange different boll setting rates,and the boll setting rates and Bt toxin content were compared.Results: Higher boll setting rate together with lower Bt toxin contents in boll wall was observed under low planting density,whereas lower boll setting rate and higher Bt toxin contents were found under high planting density.Also,higher Bt protein concentration was associated with higher soluble protein content,glutamic-pyruvic transaminase(GPT),and glutamic oxaloacetate transaminase(GOT)activities,but lower amino acid content,and protease and peptidase activities.It was further confirmed that a higher boll setting rate with lower Bt protein content under flower-removal.Conclusions: This study demonstrated that the insecticidal efficacy of boll walls was significantly impacted by boll formation.Reduced protein synthesis and enhanced protein degradation were related to the reduced Bt toxin concentration.
Background：Human immunodeficiency virus （HIV） is spreading rapidly among men who have sex with men （MSM） in China.Anonymous questionnaires or direct interviews have been frequently used to study their behavior.The aim of the study was to describe the behavioral risk profile of the MSM in Beijing using the randomized response techniques （RRTs）.Methods：A cross-sectional survey of sexual behavior among a sample of MSM was conducted in two HIV counseling and testing clinics in Beijing.The survey was carried out with an anonymous questionnaire containing sensitive questions on sexual behavior.To obtain the honest responses to the sensitive questions,three distinctive RRTs were used in the questionnaire：（1） Additive randomized response model for quantitative questions,（2） randomized response model for multiple choice questions,and （3） Simmons randomized response model for binomial questions.Formulae for the point estimate,variance,and confidence interval （CI） were provided for each specific model.Results：Using RRTs in a sample of 659 participants,the mean age at first homosexual encounter was estimated to be 21.7 years （95% CI：21.2-22.2）,and each had sex with about three （2.9,95% CI：2.4-3.4） male partners on average in the past month.The estimated rate for consistent condom use was 56.4% （95% CI：50.1-62.8%）.In addition,condom was estimated to be used among 80.0% （95% CI：74.1-85.9%） of the population during last anal sex with a male partner.Conclusions：Our study employed RRTs in a survey containing questions on sexual behavior among MSM,and the results showed that RRT might be a useful tool to obtain truthful feedback on sensitive information such as sexual behavior from the respondents,especially in traditional Chinese cultural settings.
