Bridged adversarial training.

Adversarial robustness is considered a required property of deep neural networks. In this study, we discover that adversarially trained models might have significantly different characteristics in terms of margin and smoothness, even though they show similar robustness. Inspired by the observation, we investigate the effect of different regularizers and discover the negative effect of the smoothness regularizer on maximizing the margin. Based on the analyses, we propose a new method called bridged adversarial training that mitigates the negative effect by bridging the gap between clean and adversarial examples. We provide theoretical and empirical evidence that the proposed method provides stable and better robustness, especially for large perturbations.

Efficacy and safety of moderate-intensity statin with ezetimibe combination therapy in patients after percutaneous coronary intervention: a post-hoc analysis of the RACING trial.

Background: Moderate-intensity statin role with ezetimibe combination therapy following percutaneous coronary intervention (PCI) has not been thoroughly investigated, particularly compared to high-intensity statin monotherapy. We aimed to investigate the effect of ezetimibe combination with moderate-intensity statin in patients with atherosclerotic cardiovascular disease following PCI. Methods: This was a post-hoc analysis of a subset of patients who underwent PCI in the RACING trial. At 26 centres in South Korea, patients with atherosclerotic cardiovascular disease (ASCVD) were randomly assigned to receive either moderate-intensity statin with ezetimibe combination therapy (rosuvastatin 10 mg with ezetimibe 10 mg) or high-intensity statin monotherapy (rosuvastatin 20 mg). The prespecified endpoints of the RACING trial were used. The primary endpoint was the 3-year composite of cardiovascular death, major cardiovascular events, and nonfatal stroke. Event rates between the two groups were compared using log-rank tests, and hazard ratios (HR) with 95% confidence intervals (CI) were estimated using Cox regression analysis. Consistent with the RACING trial, the primary and secondary efficacy endpoints were evaluated using an intention-to-treatment approach, and the safety endpoints were assessed in the safety population. The RACING trial was registered at ClinicalTrials.gov (NCT03044665). Findings: Between Feb 14, 2017, and Dec 18, 2018, 3780 participants were enrolled in the RACING trial. Prior history of PCI was found in 2497 patients (67%, median 64 years, 79% male), and was associated with higher rates of the primary endpoint (hazard ratio [HR], 1.34; 95% confidence interval [CI], 1.06-1.69; p = 0.014). Among patients with prior PCI, moderate-intensity statin therapy with ezetimibe combination versus high-intensity statin therapy did not increase the risk of the primary endpoint (HR, 0.95; 95% CI, 0.74-1.24; p = 0.781). The proportion of patients with low-density lipoprotein cholesterol (LDL-C) <70 mg/dL at 1, 2, and 3 years was 74%, 76%, and 73%, respectively, in the combination therapy group, and was significantly higher than that in the high-intensity statin monotherapy group (57%, 62%, and 59%, respectively, all p < 0.001). Discontinuation of lipid-lowering drugs occurred less frequently in the combination group (4.2% vs. 7.6%, p = 0.001). Interpretation: The effects of ezetimibe combination therapy observed in the RACING trial were consistently preserved among patients with ASCVD following PCI. Ezetimibe combination could be considered as a suitable therapeutic strategy to achieve strict control of LDL-C and reduce drug intolerance in patients who underwent PCI. Funding: Hanmi Pharmaceutical, Seoul, South Korea.

GradDiv: Adversarial Robustness of Randomized Neural Networks via Gradient Diversity Regularization.

Deep learning is vulnerable to adversarial examples. Many defenses based on randomized neural networks have been proposed to solve the problem, but fail to achieve robustness against attacks using proxy gradients such as the Expectation over Transformation (EOT) attack. We investigate the effect of the adversarial attacks using proxy gradients on randomized neural networks and demonstrate that it highly relies on the directional distribution of the loss gradients of the randomized neural network. We show in particular that proxy gradients are less effective when the gradients are more scattered. To this end, we propose Gradient Diversity (GradDiv) regularizations that minimize the concentration of the gradients to build a robust randomized neural network. Our experiments on MNIST, CIFAR10, and STL10 show that our proposed GradDiv regularizations improve the adversarial robustness of randomized neural networks against a variety of state-of-the-art attack methods. Moreover, our method efficiently reduces the transferability among sample models of randomized neural networks.

The cost of social vulnerability: an integrative conceptual framework and model for assessing financial risks in natural disaster management.

Groups that are unable to prepare for disasters, or to recover from damage on their own, have a high dependency on government services, which inevitably leads to more government spending. Given this, governments can better project the entire cost of disasters and, in turn, effectively manage their finances, by proactively identifying high-vulnerable populations in anticipating financial costs of disasters. However, little attention has been paid to social vulnerability in assessing financial risks in the natural hazards or public finance studies. Thus, this article fills this gap by bringing the concept of social vulnerability from three different fields of study to propose a conceptual framework and corresponding applicable model for estimating disaster costs to inform governmental financial management: the sociological literature on disaster management, economics literature on risk management, and environmental literature of disasters. We review 134 articles on vulnerability from 1990 to 2021, assessing the different conceptualizations of social vulnerability, and the factors affecting vulnerable populations, in each literature. This study contributes to the natural hazards literature on financial and emergency management by integrating the existing literature on social vulnerability into a conceptual framework for measuring social vulnerability and relating it to efforts to assess the financial impact of disasters. Furthermore, based on this conceptual framework, we develop an applicable model for estimating the financial costs of disasters that researchers or governments may apply to assess and develop effective strategies for managing the financial risks associated with disasters. Specifically, the model, which we call the cost of social vulnerability to disasters model (CSVDM), suggests specific indicators from the literature to measure the costs of social vulnerability to more accurately predict the financial impact of disasters.

Pervaporation-assisted in situ formation of nanoporous microchannels with various material and structural properties.

Nanoporous structures are crucial for developing mixed-scale micro-/nanofluidic devices because they facilitate the manipulation of molecule transport along the microfluidic channel networks. Particularly, self-assembled particles have been used for fabricating various nanoporous membranes. However, previous self-assembly mechanisms relied on the material and structural homogeneities of the nanopores. Here, we present a pervaporation-assisted in situ fabrication method that integrates nanoporous membrane structures into microfluidic devices. The microfluidic devices contain a control-channel layer at the top, which induces local and addressable pervaporation, and the main-channel layer, which is present at the bottom with pre-designated locations for nanoporous microchannels; the layers are separated using a gas-permeable film. The target particle suspensions are loaded into the main channels, and their pervaporation is controlled through the gas-permeable film, which successfully assembles the particles at the pre-designated locations. This method yields nanoporous microchannels with various material and structural properties by fabricating heterogeneous nanopore arrays/junctions in series and other diverse structures along the microchannels. We validate the basic working principle of microfluidic devices containing nanoporous microchannels. Furthermore, we theoretically analyze the fundamental experimental results, which suggest the remarkable potential of our strategy to fabricate nanopore networks without using conventional nanofabrication methods.

Constituents of Cryptotaenia japonica Inhibit Melanogenesis via CREB- and MAPK-Associated Signaling Pathways in Murine B16 Melanoma Cells.

Melanin plays an important role in protecting the skin against ultraviolet light and is responsible for skin color. However, overproduction of melanin is related to several skin disorders, such as age spots, freckles, café au lait spots, Becker's nevus and other hyperpigmentation syndromes. The aim of this study was to identify the effects of kaempferol-7-O-ß-d-glucuronide (K7G) and tilianin, isolated from Cryptotaenia japonica, on melanogenesis and their mechanisms of action in murine B16 melanoma cells. The α-melanocyte-stimulating hormone (α-MSH)-induced melanin production was significantly inhibited by K7G and tilianin in a dose-dependent manner. The effects of these compounds on the signaling pathway of melanogenesis were examined. K7G and tilianin downregulated the expression of microphthalmia-associated transcription factor (MITF) and melanocyte-specific enzymes, i.e., tyrosinase and TRP1. These compounds also inhibited the phosphorylation of cyclic adenosine monophosphate (cAMP)-response element binding protein (CREB) in a dose-dependent manner. In addition, these compounds increased the phosphorylation of extracellular signal-regulated kinase (ERK) but decreased the phosphorylation of c-Jun N-terminal kinase (JNK) in B16 cells. Based on the above results, the anti-melanogenic effects of these compounds are caused by suppression of the MAPK signaling pathway through the down-regulation of α-MSH-induced CREB accumulation. This finding suggests that K7G and tilianin may be good candidates for further research to develop therapeutic agents for hyperpigmentation diseases.

Blood Pressure and Cholesterol-lowering Efficacy of a Fixed-dose Combination With Irbesartan and Atorvastatin in Patients With Hypertension and Hypercholesterolemia: A Randomized, Double-blind, Factorial, Multicenter Phase III Study.

PURPOSE: A fixed-dose combination of a stain and an antihypertensive drug may be useful for the treatment of patients with hypertension and hyperlipidemia. It may also improve patient drug compliance to help control risk factors of cardiovascular disease. This study was designed to evaluate the blood pressure-lowering and cholesterol-lowering effect of a fixed-dose combination of irbesartan-atorvastatin compared with monotherapy by either agent over an 8-week treatment period. METHODS: Patients with comorbid hypertension and hypercholesterolemia were screened for this randomized, double-blind, Phase III study. Eligible study patients were randomly assigned to test groups receiving a combination of irbesartan 300 mg and atorvastatin 40 mg or 80 mg (IRB300 + ATO40 and IRB300 + ATO80). Comparator groups comprised monotherapy groups with irbesartan 300 mg (IRB300) or atorvastatin 40 mg (ATO40) or atorvastatin 80 mg (ATO80), or placebo. Patients who were eligible at screening were subjected to a 4- to 6-week washout period before commencing 8 weeks of therapy per their assigned group. The primary efficacy end points were percent change in LDL-C and sitting diastolic blood pressure (DBP) levels from baseline to end of therapy. Tolerability profiles of combination therapy were compared with other groups. FINDINGS: A total of 733 patients with comorbid hypertension and hypercholesterolemia were screened for this study; 230 eligible patients were randomized to treatment. The mean age of patients was 58.9 (8.5) years, and their mean body mass index was 25.8 (3.2) kg/m2. More than two thirds (70.9%) of the study patients were male. Mean LDL-C and sitting DBP levels at baseline were 149.54 (29.19) mg/dL and 92.32 (6.03) mm Hg, respectively. Percent reductions in LDL-C after 8 weeks were 46.74% (2.06%) in the IRB300 + ATO40 group and 48.98% (2.12%) in the IRB300 + ATO80 group; these values were 47.13% (3.21%) and 48.30% (2.98%) in the ATO40 and ATO80 comparator groups. Similarly, a reduction in sitting DBP after 8 weeks was -8.50 (1.06) mm Hg in the IRB300 + ATO40 group and 10.66 (1.08) mm Hg in the IRB300 + ATO80 group compared with 8.40 (1.65) mm Hg in the IRB300 group. The incidence rate for treatment-emergent adverse events was 22.27% and was similar between the monotherapy and combination groups. IMPLICATIONS: A once-daily combination product of irbesartan and atorvastatin provided an effective, safe, and more compliable treatment for patients with coexisting hypertension and hyperlipidemia. ClinicalTrials.gov identifier: NCT01442987.