Search

Many automatic speech recognition (ASR) systems provide functions such as sending messages or opening garage doors. If such a system malfunctions, it could pose a significant threat. In this paper, we propose the attack scenario in which we transfer audio adversarial examples to radio system through the radio signal and the sound received through radio deceives the ASR system. In addition, we propose over-the-air technology using environmental noise so that we can attack easily and effectively. Using this method, we observed that the combined sound was much more robust to environmental noise than the non-combined sound. Finally, we experiment as our scenario on a real environment with adversarial examples created using environmental noise and show the results. Our attack succeeded against commercial ASR systems with a success rate of over 50%.

Unmanned Aerial Vehicles (UAVs) are aerial vehicles that can go to a particular position without human control or with remote human control. It is because unmanned control mainly relies on position estimation that a lot of research has been studied on GPS spoofing attacks. Detection methods against GPS spoofing attacks mainly include monitoring RF signals or IMU sensor values inside UAVs. In this work, we analyze GPS attack and detection in an advanced autopilot system, PX4 without excluding RF detection. Recent studies have shown that GPS spoofing is unable to evade the detection using EKF sensor fusion. Therefore, this paper experiment whether the detection could be evaded by strengthening the attacker model. This paper classifies attacker models according to whether an attacker knows the true position of UAVs or the estimated position by UAV and proposed attack methods depending on each model in PX4. We inject GPS value which our attack method intends to UAV during mission flight in simulation environment. By manipulating the GPS driver code of PX4 controller, we inject GPS value the attack method wants into UAV in physical environment. Finally, we observed the innovation test ratio during GPS spoofing and demonstrate that GPS spoofing attack is possible keeping the innovation test ratio under the anomaly detection criterion. After that, we proposed our approach which scales the EKF’s Kalman gain randomly to increase the detection method’s efficiency and experiment with the attack methods. This detection method has little effect on the test ratio without the attack. But during the attack, the innovation test ratio was increased higher than the anomaly detection criterion so that the attack could be detected.

In recent years, there has been a growing interest in applying knowledge distillation (KD) techniques to the connectionist temporal classification (CTC) framework for training more efficient speech recognition models. Although conventional KD approaches have successfully reduced computational burden, they have limitations in dealing with the inconsistency problem caused by dropout regularization, particularly the gap between the training and inference stages. In the context of KD, this inconsistency may hinder the performance improvement of the student model. To overcome this issue, we propose a novel approach, namely Cons-KD, that combines KD and consistency regularization, where the former trains the student model to benefit from the knowledge of the teacher model, and the latter trains the student model to be more robust to the dropout-induced inconsistency. By directly mitigating the inconsistency problem, our KD framework can further improve the student’s performance compared to the vanilla KD. Experimental results on the LibriSpeech dataset demonstrate that Cons-KD significantly outperforms previous KD methods, improving the word error rate (WER) from 5.10 % to 4.13 % on the test-clean subset and from 12.87 % to 10.32 % on the test-other subset, respectively. These improvements correspond to relative error rate reduction (RERR) of 19.02 % and 19.81 %, respectively, implying notable advancements beyond conventional KD methods. Additionally, we conduct an in-depth analysis to verify the effect of each proposed objective.

BackgroundHeart failure (HF) is one of the most common initial manifestations of cardiovascular disease in patients with type 2 diabetes. Although smoking is an independent risk factor for HF, there is a lack of data for the incidence of HF according to changes in smoking behaviors in patients with type 2 diabetes. ObjectiveWe aimed to examine the association between interval changes in smoking behavior and the risk of HF among patients with type 2 diabetes. MethodsWe conducted a retrospective cohort study using the National Health Insurance Service database. We identified 365,352 current smokers with type 2 diabetes who had 2 consecutive health screenings (2009-2012) and followed them until December 31, 2018, for the incident HF. Based on smoking behavior changes between 2 consecutive health screenings, participants were categorized into quitter, reducer I (≥50% reduction) and II (

Background Genome-wide association studies (GWAS) on type 2 diabetes mellitus (T2DM) have identified more than 400 distinct genetic loci associated with diabetes and nearly 120 loci for fasting plasma glucose (FPG) and fasting insulin level to date. However, genetic risk factors for the longitudinal deterioration of FPG have not been thoroughly evaluated. We aimed to identify genetic variants associated with longitudinal change of FPG over time. Methods We used two prospective cohorts in Korean population, which included a total of 10,528 individuals without T2DM. GWAS of repeated measure of FPG using linear mixed model was performed to investigate the interaction of genetic variants and time, and meta-analysis was conducted. Genome-wide complex trait analysis was used for heritability calculation. In addition, expression quantitative trait loci (eQTL) analysis was performed using the Genotype-Tissue Expression project. Results A small portion (4%) of the genome-wide single nucleotide polymorphism (SNP) interaction with time explained the total phenotypic variance of longitudinal change in FPG. A total of four known genetic variants of FPG were associated with repeated measure of FPG levels. One SNP (rs11187850) showed a genome-wide significant association for genetic interaction with time. The variant is an eQTL for NOC3 like DNA replication regulator (NOC3L) gene in pancreas and adipose tissue. Furthermore, NOC3L is also differentially expressed in pancreatic β-cells between subjects with or without T2DM. However, this variant was not associated with increased risk of T2DM nor elevated FPG level. Conclusion We identified rs11187850, which is an eQTL of NOC3L, to be associated with longitudinal change of FPG in Korean population.

The goal of the “2019 Automatic Speaker Verification Spoofing and Countermeasures Challenge” (ASVspoof) was to make it easier to create systems that could identify voice spoofing attacks with high levels of accuracy. However, model complexity and latency requirements were not emphasized in the competition, despite the fact that they are stringent requirements for implementation in the real world. The majority of the top-performing solutions from the competition used an ensemble technique that merged numerous sophisticated deep learning models to maximize detection accuracy. Those approaches struggle with real-world deployment restrictions for voice assistants which would have restricted resources. We merged skip connection (from ResNet) and max feature map (from Light CNN) to create a compact system, and we tested its performance using the ASVspoof 2019 dataset. Our single model achieved a replay attack detection equal error rate (EER) of 0.30% on the evaluation set using an optimized constant Q transform (CQT) feature, outperforming the top ensemble system in the competition, which scored an EER of 0.39%. We experimented using depthwise separable convolutions (from MobileNet) to reduce model sizes; this resulted in an 84.3 percent reduction in parameter count (from 286K to 45K), while maintaining similar performance (EER of 0.36%). Additionally, we used Grad-CAM to clarify which spectrogram regions significantly contribute to the detection of fake data.

Abstract Background Type 2 diabetes and non-alcoholic fatty liver disease (NAFLD) commonly coexist. However, NAFLD’s effect on mortality in Asian patients with type 2 diabetes awaits full elucidation. Therefore, we examined NAFLD-related all-cause and cause-specific mortality in a nationwide Asian population with type 2 diabetes. Methods We included patients who had undergone general health checkups between 2009 and 2012 using the National Health Insurance Service database linked to death-certificate data. Hepatic steatosis was defined as a fatty liver index (FLI) ≥ 60, and advanced hepatic fibrosis was determined using the BARD score. Findings During the follow-up period of 8.1 years, 222,242 deaths occurred, with a mortality rate of 14.3/1000 person-years. An FLI ≥ 60 was significantly associated with increased risks of all-cause and cause-specific mortality including cardiovascular disease (CVD)-, cancer-, and liver disease (FLI ≥ 60: hazard ratio [HR] = 1.02, 95% confidence interval [CI] 1.01–1.03 for all-cause; 1.07, 1.04–1.10 for CVD; 1.12, 1.09–1.14 for cancer; and 2.63, 2.50–2.77 for liver disease). Those with an FLI ≥ 60 and fibrosis (BARD ≥ 2) exhibited increased risks of all-cause (HR, 95% CI 1.11, 1.10–1.12), CVD- (HR, 95% CI 1.11, 1.09–1.14), cancer- (HR, 95% CI 1.17, 1.15–1.19), and liver disease-related (HR, 95% CI 2.38, 2.29–2.49) mortality. Conclusion Hepatic steatosis and advanced fibrosis were significantly associated with risks of overall and cause-specific mortality in patients with type 2 diabetes. Our results provide evidence that determining the presence of hepatic steatosis and/or fibrosis potentially plays a role in risk stratification of mortality outcomes in patients with type 2 diabetes mellitus.

Abstract Background Metabolic (dysfunction)‐associated fatty liver disease (MAFLD) emphasizes the metabolic dysfunction in nonalcoholic fatty liver disease (NAFLD). Although the relationship between low muscle mass and NAFLD has been suggested, the effect of MAFLD on low muscle mass is yet to be investigated. In this study, we examined the relationship between MAFLD and low muscle mass in an asymptomatic Korean population. Methods Examinees who underwent FibroScan® and bioelectrical impedance analyses on the same day during the period of June 2017 to December 2019 were included. Hepatic steatosis was diagnosed using controlled attenuation parameter (CAP) with two cut‐off values of 248 and 294 dB/m. Low muscle mass was defined based on appendicular skeletal muscle mass/body weight (wt) or body mass index (BMI) ratios of two standard deviations below the sex‐specific mean for healthy young adults. Subjects were divided into four subgroups: diabetic MAFLD (presence of diabetes mellitus [DM]), metabolic dysfunction (MD) MAFLD (≥2 metabolic abnormalities without DM), overweight MAFLD (overweight/obese without DM and

IntroductionClonal hematopoiesis of indeterminate potential (CHIP) is associated with atherosclerosis and cardiovascular disease. It has been suggested that CHIP may be related to diabetes, so we investigated the association between CHIP and new-onset type 2 diabetes.MethodsThis study included 4,047 subjects aged >=40 years without diabetes. To detect CHIP, targeted gene sequencing of genomic DNA from peripheral blood cells was performed. The incidence of new-onset type 2 diabetes during the follow-up period was evaluated.ResultsOf the total subjects, 635 (15.7%) had CHIP. During the median follow-up of 5.1 years, the incidence of new-onset diabetes was significantly higher in CHIP carriers than in subjects without CHIP (11.8% vs. 9.1%, p = 0.039). In a univariate analysis, CHIP significantly increased the risk of new-onset diabetes (HR 1.32, 95% CI 1.02–1.70, p = 0.034), but in a multivariate analysis, it was not significant. The CHIP-related risk of new onset diabetes differed according to LDL cholesterol level. In the hyper-LDL cholesterolemia group, CHIP significantly increased the risk of diabetes (HR 1.64, 95% CI 1.09–2.47, p = 0.018), but it did not increase the risk in the non-hyper-LDL cholesterolemia group. The subjects with CHIP and hyper-LDL-cholesterolemia had approximately twice the risk of diabetes than subjects without CHIP and with low LDL cholesterol (HR 2.05, 95% CI 1.40–3.00, p < 0.001).ConclusionThe presence of CHIP was a significant risk factor for new-onset type 2 diabetes, especially in subjects with high LDL cholesterol. These results show the synergism between CHIP and high LDL cholesterol as a high-risk factor for diabetes.

Cellular agriculture is an emerging research field of agribiotechnology that aims to produce agricultural products using stem cells, without sacrificing animals or cultivating crops. Cultivated meat, as a representative cellular product of cellular agriculture, is being actively researched due to global food insecurity, environmental, and ethical concerns. This review focuses on the application of stem cells, which are the seeds of cellular agriculture, for the production of cultivated meat, with emphasis on deriving and culturing muscle and adipose stem cells for imitating fresh meat. Establishing standards and safety regulations for culturing stem cells is crucial for the market entry of cultured muscle tissue-based biomaterials. Understanding stem cells is a prerequisite for creating reliable cultivated meat and other cellular agricultural biomaterials. The techniques and regulations from the cultivated meat industry could pave the way for new cellular agriculture industries in the future.

Background Epidemiological data have shown that vitamin D deficiency is highly prevalent in Korea. Genetic factors influencing vitamin D deficiency in humans have been studied in Europe but are less known in East Asian countries, including Korea. We aimed to investigate the genetic factors related to vitamin D levels in Korean people using a genome-wide association study (GWAS). Methods We included 12,642 subjects from three different genetic cohorts consisting of Korean participants. The GWAS was performed on 7,590 individuals using linear or logistic regression meta- and mega-analyses. After identifying significant single nucleotide polymorphisms (SNPs), we calculated heritability and performed replication and rare variant analyses. In addition, expression quantitative trait locus (eQTL) analysis for significant SNPs was performed. Results rs12803256, in the actin epsilon 1, pseudogene (ACTE1P) gene, was identified as a novel polymorphism associated with vitamin D deficiency. SNPs, such as rs11723621 and rs7041, in the group-specific component gene (GC) and rs11023332 in the phosphodiesterase 3B (PDE3B) gene were significantly associated with vitamin D deficiency in both meta- and mega-analyses. The SNP heritability of the vitamin D concentration was estimated to be 7.23%. eQTL analysis for rs12803256 for the genes related to vitamin D metabolism, including glutamine-dependent NAD(+) synthetase (NADSYN1) and 7-dehydrocholesterol reductase (DHCR7), showed significantly different expression according to alleles. Conclusion The genetic factors underlying vitamin D deficiency in Korea included polymorphisms in the GC, PDE3B, NADSYN1, and ACTE1P genes. The biological mechanism of a non-coding SNP (rs12803256) for DHCR7/NADSYN1 on vitamin D concentrations is unclear, warranting further investigations.

Objective This study aimed to elucidate the effects of cooking conditions on the physicochemical and sensory characteristics of dry- and wet-aged beef strip loins. Methods Dry- and wet-aged beef aged for 28 days were cooked using different cooking methods (grilling or oven roasting)×cooking temperatures (150°C or 230°C), and their pH, 2-thiobarbituric acid reactive substances (TBARS), volatile compounds, and color were measured. Results Cooking conditions did not affect pH; however, grilling resulted in lower TBARS but higher cooking doneness at the dry-aged beef surface compared to oven roasting (p< 0.05). In descriptive sensory analysis, the roasted flavor of dry-aged beef was significantly stronger when grill-cooked compared to oven roasting. Dry-aged beef grill-cooked at 150°C presented a higher intensity of cheesy flavor, and that grilled at 230°C showed a greater intensity of roasted flavor compared to wet-aged beef at the same condition, respectively. Conclusion Grilling may be effective for enhancing the unique flavor in dry-aged beef.

AimThis study was conducted to evaluate the accuracy of a newly developed multifrequency segmental (MFS) bioelectrical impedance analysis (BIA) method using an additional portable abdominal (PA) impedance analyzer, in the assessment of abdominal visceral fat area (VFA).Materials and methodsOne hundred healthy Korean subjects aged 19 years or over (43 men and 57 women) were recruited, and VFA was estimated by a conventional MFS-BIA machine and a new MFS-BIA machine with a PA-BIA device, indicating MFS-VFA and MFS&PA-VFA, respectively. The accuracy of the VFA values was compared with those evaluated with CT at the level of the umbilicus (CT-VFA).ResultsThe mean age was 41 years and mean body mass index (BMI) was 24.4 kg/m2. The mean ± SD VFAs measured by CT, conventional MFS-BIA, and new MFS&PA-BIA together were 93.4 ± 60.9, 92.7 ± 53.4, and 93.6 ± 55.4 cm2, respectively. Correlation coefficients comparing CT-VFA with MFS-VFA and MFS&PA-VFA were 0.612 and 0.932, respectively (P < 0.001 for both). The mean difference between CT-VFA and MFS&PA-VFA was less affected by age, sex, and BMI compared with that between CT-VFA and MFS-VFA. Intraclass correlation coefficient (95% CI) between CT-VFA and MFS&PA-VFA was also greater than that between CT-VFA and MFS-VFA, 0.96 (0.95–0.98) vs. 0.76 (0.64–0.84), respectively.ConclusionIn this study, application of a newly developed MFS-BIA machine combined with a PA-BIA device significantly improved the correlation with CT-measured VFA without proportional error. This novel approach using advanced technology may be able to provide more reliable estimates of abdominal VFA.

This study investigates how visual appearance of pork cuts affects consumer preference. Images of pork belly, Boston butt, and loin were chosen on the basis of visible fattiness and used to analyze consumers’ perception of the appearance of each pork cut. Meat color and visible fat proportion of images of pork cuts were analyzed by the researchers before conducting the survey. A total of 211 pork eaters evaluated the pork cuts based on appearance (lightness of color, redness, visible fat proportion, and fat distribution), preferability, and overall acceptability. Also, muscle pieces from different pork cuts were taken and the relative area composition of muscle fibers was measured. Based on survey results, correlation between visual traits and preferences of each pork cut was analyzed. The survey results showed that preferred pork appearance varied as per each individual’s favorite pork cut. Also, the respondents evaluated visual characteristics and preference for each pork cut differently possibly due to the different visual characteristics of each cut. Correlation analysis between visual traits and preference indicated that overall acceptability of pork cuts was mainly influenced by fat preference, followed by color preference. Fat and color preferences for each pork cut were affected by various visual attributes including redness, lightness of color, visible fat proportion, and fat distribution, but their effects were considerably varied among different pork cuts. Thus, Korean consumers perceived and assessed pork appearance using various quality cues but the evaluation depended on which cut was being observed.

Background Although obesity differs according to ethnicity, it is globally established as a solid risk factor for cardiovascular disease. However, it is not fully understood how obesity parameters affect the progression of coronary artery calcification (CAC) in Korean population. We sought to evaluate the association of obesity-related parameters including visceral adipose tissue (VAT) measurement and CAC progression. Methods This retrospective observational cohort study investigated 1,015 asymptomatic Korean subjects who underwent serial CAC scoring by computed tomography (CT) with at least 1-year interval and adipose tissue measurement using non-contrast CT at baseline for a routine checkup between 2003 and 2015. CAC progression, the main outcome, was defined as a difference of ≥2.5 between the square roots of the baseline and follow-up CAC scores using Agatston units. Results During follow-up (median 39 months), 37.5% of subjects showed CAC progression of a total population (56.4 years, 80.6% male). Body mass index (BMI) ≥25 kg/m2, increasing waist circumferences (WC), and higher VAT/subcutaneous adipose tissue (SAT) area ratio were independently associated with CAC progression. Particularly, predominance of VAT over SAT at ≥30% showed the strongest prediction for CAC progression (adjusted hazard ratio, 2.20; P

Correlation Power Analysis (CPA) uses a lot of computational resources for deploying side-channel analysis to power traces. For every guessing key, it searches all data points and all power traces to reveal the secret key implemented in the encryption module. In this work, we propose a novel technique to narrow down the target region for applying CPA to the power traces. As byte-wise operations occur in every clock period, the assumed amplitude value based on the power consumption model and the actual amplitude are most correlated at the points where the peaks are observed. We extract the calculation points of each power trace using signal processing and machine learning techniques, and select the region where the CPA will be applied to. Our proposed approach achieves an approximately 20% improvement in accuracy and requires less computational resources compared to conventional CPA.

Background As the genetic variants of trabecular bone microarchitecture are not well-understood, we performed a genome-wide association study to identify genetic determinants of bone microarchitecture analyzed by trabecular bone score (TBS). Methods TBS-associated genes were discovered in the Ansung cohort (discovery cohort), a community-based rural cohort in Korea, and then validated in the Gene-Environment Interaction and Phenotype (GENIE) cohort (validation cohort), consisting of subjects who underwent health check-up programs. In the discovery cohort, 2,451 participants were investigated for 1.42 million genotyped and imputed markers. Results In the validation cohort, identified as significant variants were evaluated in 2,733 participants. An intronic variant in iroquois homeobox 3 (IRX3), rs1815994, was significantly associated with TBS in men (P=3.74E-05 in the discovery cohort, P=0.027 in the validation cohort). Another intronic variant in mitogen-activated protein kinase kinase 5 (MAP2K5), rs11630730, was significantly associated with TBS in women (P=3.05E-09 in the discovery cohort, P=0.041 in the validation cohort). Men with the rs1815994 variant and women with the rs11630730 variant had lower TBS and lumbar spine bone mineral density. The detrimental effects of the rs1815994 variant in men and rs11630730 variant in women were also identified in association analysis (β=−0.0281, β=−0.0465, respectively). Conclusion In this study, the rs1815994 near IRX3 in men and rs11630730 near MAP2K5 in women were associated with deterioration of the bone microarchitecture. It is the first study to determine the association of genetic variants with TBS. Further studies are needed to confirm our findings and identify additional variants contributing to the trabecular bone microarchitecture.

AimA link between low muscle mass and arterial stiffness is not always consistent. In this study, we aimed to evaluate the clinical significance of low skeletal muscle mass in relation to arterial stiffness measured by the cardio-ankle vascular index (CAVI).MethodsA total of 2,561 asymptomatic Korean subjects who underwent bioelectrical impedance analysis (BIA) and CAVI were included for analysis. Using appendicular skeletal muscle mass (ASM), classes I and II sarcopenia were defined as ASM% greater than 1 standard deviation (SD) and 2 SDs below the gender-specific mean of healthy young Korean adults.ResultsCompared to normal, CAVI was significantly higher, but the number of patients with a low ankle-brachial index (ABI) was not significantly different (p < 0.001 for CAVI, p = 0.078 for ABI). Classes I and II sarcopenia showed an independent and significant association with CAVI (estimate 0.148, standard error (SE) 0.043, p < 0.001 and estimate 0.304, SE 0.073, p < 0.001 for classes I and II sarcopenia, respectively, adjusted for age groups, gender, body mass index (BMI) ≥25, hypertension, diabetes, hypercholesterolemia, and smoking).ConclusionLow muscle mass is independently and significantly associated with increased CAVI, and should be considered when managing asymptomatic subjects to assess the risk of atherosclerosis.

The purpose of our study is to figure out the transitions of the cryptocurrency market due to the outbreak of COVID-19 through network analysis, and we studied the complexity of the market from different perspectives. To construct a cryptocurrency network, we first apply a mutual information method to the daily log return values of 102 digital currencies from January 1, 2019, to December 31, 2020, and also apply a correlation coefficient method for comparison. Based on these two methods, we construct networks by applying the minimum spanning tree and the planar maximally filtered graph. Furthermore, we study the statistical and topological properties of these networks. Numerical results demonstrate that the degree distribution follows the power-law and the graphs after the COVID-19 outbreak have noticeable differences in network measurements compared to before. Moreover, the results of graphs constructed by each method are different in topological and statistical properties and the network’s behavior. In particular, during the post-COVID-19 period, it can be seen that Ethereum and Qtum are the most influential cryptocurrencies in both methods. Our results provide insight and expectations for investors in terms of sharing information about cryptocurrencies amid the uncertainty posed by the COVID-19 pandemic.

In recent years, researchers have studied various methods for transferring data in a network-separated environment, and the most representative method is the use of inaudible frequency signals like ultrasonic waves. This method has the advantage of being able to transfer data without other people noticing, but it has the disadvantage that speakers must exist. In a laboratory or company, external speakers may not be attached to each computer. Therefore, this paper presents a new covert channel attack that transfers data using internal speakers on the computer’s motherboard. The internal speaker can also produce a sound of the desired frequency, and, therefore, data can be transferred using high frequency sounds. We encode data into Morse code or binary code and transfer it. Then we record it using a smartphone. At this time, the location of the smartphone can be any distance within 1.5 m when the length per bit is longer than 50 ms, such as on the computer body or on the desk. Data are obtained by analyzing the recorded file. Our results show that data is transferred from a network-separated computer using an internal speaker with 20 bits/s in maximum.

Background : The association between nonalcoholic fatty liver disease (NAFLD) and sarcopenia has been suggested. We investigated the association between sarcopenia and NAFLD independent of visceral adiposity and searched for the clinical characteristics that affect this association. Methods : We performed a retrospective study including of 5,989 subjects (mean age, 53.2 years; men, 57.3%) who underwent bioelectrical impedance analysis (BIA) and abdominal ultrasonography in 2012. The appendicular skeletal muscle mass (ASM) was assessed by BIA method. Sarcopenia was defined as ASM/weight (ASM%)

Background: An association between sarcopenic obesity and cardiovascular disease has been suggested. We investigated the relationship between sarcopenia and coronary atherosclerosis, taking into account the presence or absence of obesity in a health check-up population.Methods: Data were reviewed for subjects who underwent bioelectrical impedance analysis (BIA) and coronary calcium scoring (CAC) computed tomography between January 2017 and December 2018. Appendicular skeletal muscle mass (ASM) was assessed using BIA. Sarcopenia was defined as reduction of muscle mass and calculated as ASM% (ASM/body weight) more than two standard deviations below the sex-specific mean for healthy young adults. CAC scores were dichotomized as low (

This study aims to improve the performance of the Weather Research and Forecasting (WRF) model in the sea breeze circulation using the micro-Genetic Algorithm (micro-GA). We found the optimal combination of four physical parameterization schemes related to the sea breeze system, including planetary boundary layer (PBL), land surface, shortwave radiation, and longwave radiation, in the WRF model coupled with the micro-GA (WRF-μGA system). The optimization was performed with respect to surface meteorological variables (2 m temperature, 2 m relative humidity, 10 m wind speed and direction) and a vertical wind profile (wind speed and direction), simultaneously for three sea breeze cases over the northeastern coast of South Korea. The optimized set of parameterization schemes out of the WRF-μGA system includes the Mellor–Yamada–Nakanishi–Niino level-2.5 (MYNN2) for PBL, the Noah land surface model with multiple parameterization options (Noah-MP) for land surface, and the Rapid Radiative Transfer Model for GCMs (RRTMG) for both shortwave and longwave radiation. The optimized set compared with the various other sets of parameterization schemes for the sea breeze circulations showed up to 29 % for the improvement ratio in terms of the normalized RMSE considering all meteorological variables.

This study investigated the effects of different aging methods on the changes in the concentrations of aroma volatiles of beef. One half (n = 15) of the beef strip loins were dry-aged, and the other half were wet-aged, and both aging processes continued for 28 days. The aroma volatiles from dry- and wet-aged samples were analyzed at seven-day intervals (n = 3 for each aging period). As the aging period increased, dry-aged beef showed higher concentrations of volatile compounds than those in wet-aged beef (p < 0.05). Most changes in the concentrations of aroma volatiles of dry-aged beef were associated with propanal, 2-methylbutanal, 2-methylpropanal, 1-butanamine, trimethylamine, 2-methyl-2-propanethiol, and ethyl propanoate, which were mainly produced by lipid oxidation and/or microbial activity (e.g., proteolysis and lipolysis) during the dry aging period. Therefore, we suggest that the differences in aroma between dry- and wet-aged beef could result from increased lipid oxidation and microbial activity in dry-aged beef possibly owing to its ambient exposure to oxygen.

Recently, generative adversarial networks (GANs) have been successfully applied to speech enhancement. However, there still remain two issues that need to be addressed: (1) GAN-based training is typically unstable due to its non-convex property, and (2) most of the conventional methods do not fully take advantage of the speech characteristics, which could result in a sub-optimal solution. In order to deal with these problems, we propose a progressive generator that can handle the speech in a multi-resolution fashion. Additionally, we propose a multi-scale discriminator that discriminates the real and generated speech at various sampling rates to stabilize GAN training. The proposed structure was compared with the conventional GAN-based speech enhancement algorithms using the VoiceBank-DEMAND dataset. Experimental results showed that the proposed approach can make the training faster and more stable, which improves the performance on various metrics for speech enhancement.

Secure computation, a methodology of computing on encrypted data, has become a key factor in machine learning. Homomorphic encryption (HE) enables computation on encrypted data without leaking any information to untrusted servers. In machine learning, the model selection method is a crucial algorithm that determines the performance and reduces the fitting problem. Despite the importance of finding the optimal model, none of the previous studies have considered model selection when performing data analysis through the HE scheme. The HE-based model selection we proposed finds the optimal complexity that best describes given data that is encrypted and whose distribution is unknown. Since this process requires a matrix calculation, we constructed the matrix multiplication and inverse of the matrix based on the bitwise operation. Based on these, we designed the model selection of the HE cross-validation approach and the HE Bayesian approach for homomorphic machine learning. Our focus was on evidence approximation for linear models to find goodness-of-fit that maximizes the evidence. We conducted an experiment on a dataset of age and Body Mass Index (BMI) from Kaggle to compare the capabilities and our model showed that encrypted data can regress homomorphically without decrypting it.

BackgroundThe study aimed to investigate the influence of hyperglycemia on muscle quality in older men with type 2 diabetes.MethodsThis was a subsidiary study of the Korean Longitudinal Study of Health and Aging. Among 326 older men consenting to tests of body composition and muscle strength, 269 men were ultimately analyzed after the exclusion because of stroke (n=30) and uncertainty about the diagnosis of diabetes (n=27). Body composition was measured using dual-energy X-ray absorptiometry and computed tomography. Muscle strength for knee extension was measured using an isokinetic dynamometer. Muscle quality was assessed from the ratio of leg strength to the entire corresponding leg muscle mass.ResultsThe muscle mass, strength, and quality in patients with type 2 diabetes did not differ significantly from controls. However, when patients with diabetes were subdivided according to their glycemic control status, patients with a glycosylated hemoglobin (HbA1c) level of ≥8.5% showed significantly decreased leg muscle quality by multivariate analysis (odds ratio, 4.510; P=0.045) after adjustment for age, body mass index, smoking amount, alcohol consumption, physical activity, and duration of diabetes. Physical performance status was also impaired in subjects with an HbA1c of ≥8.5%.ConclusionPoor glycemic control in these older patients with diabetes was associated with significant risk of decreased muscle quality and performance status. Glycemic control with an HbA1c of

In this paper, we propose a new automated way to find out the secret exponent from a single power trace. We segment the power trace into subsignals that are directly related to recovery of the secret exponent. The proposed approach does not need the reference window to slide, templates nor correlation coefficients compared to previous manners. Our method detects change points in the power trace to explore the locations of the operations and is robust to unexpected noise addition. We first model the change point detection problem to catch the subsignals irrelevant to the secret and solve this problem with Markov Chain Monte Carlo (MCMC) which gives a global optimal solution. After separating the relevant and irrelevant parts in signal, we extract features from the segments and group segments into clusters to find the key exponent. Using single power trace indicates the weakest power level of attacker where there is a very slight chance of acquiring as many power traces as needed for breaking the key. We empirically show the improvement in accuracy even with presence of high level of noise.

OBJECTIVES:Debates whether metabolically healthy obesity (MHO) increases the cardiovascular risk might be due to the metabolic instability of MHO or the absence of a perfect definition of MHO. Therefore, we aimed to investigate the influence of the MHO phenotype on the coronary artery calcium score (CACS) progression according to definition of MHO. METHODS:We analyzed a retrospective cohort with a CACS of 0 at baseline and available serial CACS measurements taken ≥ 12 months apart (n = 1,218). Obesity was defined as BMI ≥ 25 kg/m2, and MHO was defined as obesity accompanied by ≤ 1 (MHO class I) or 0 (MHO class II) components of metabolic syndrome (MetS). RESULTS:During a median follow-up of 45 months, 32.2% of MHO class I and 10.2% of MHO class II subjects developed MetS. Compared to non-obese/metabolically healthy subjects (reference group), hazard ratios (HR) for development of MetS were 2.174 (95% confidence interval [CI]: 1.513-3.124) and 1.166 (95% CI: 0.434-3.129) for MHO class I and II subjects, respectively. The MHO class I subjects showed a significantly increased risk of CACS progression as compared to the reference group (HR: 1.653; 95% CI: 1.144-2.390), whereas MHO class II subjects did not (HR: 1.195; 95% CI: 0.514-2.778). Among subjects with MHO class I, no significant CACS progression was observed in the subjects who maintained metabolic health during follow-up (HR: 1.448; 95% CI: 0.921-2.278). CONCLUSIONS:The risks of metabolic deterioration and CACS progression were significant in subjects with MHO class I, but not in those with MHO class II.

In the Guidance for Industry from the Food and Drug Administration in 2008, excess cardiovascular risk should be ruled out in trials of all new antidiabetic drugs; however, relatively few studies have focused on cardiovascular safety with antidiabetic drug use. We aimed to examine mortality and cardiovascular risk using a network meta-analysis. We searched the Medline, Embase, Cochrane, and ClinicalTrials.gov registry databases in March 2016 to identify randomized controlled trials reporting cardiovascular risk with the following oral antidiabetic drugs: metformin, sulfonylureas, thiazolidinedione (TZD), dipeptidyl peptidase-4 (DPP4) inhibitors, and sodium-glucose co-transporter-2 (SGLT2) inhibitors. We assessed the differences in the risks of all-cause mortality, cardiovascular-related mortality, acute coronary syndrome (ACS), and myocardial infarction (MI) among antidiabetic drugs with fixed effect models for direct pairwise comparisons and Bayesian network meta-analyses to integrate direct and indirect comparisons. Of the 101,183 patients in 73 randomized controlled trials, 3,434 (3.4%) died. The relative risks of all-cause mortality with SGLT2 inhibitor use were 0.68 (95% credible interval: 0.57-0.80), 0.74 (0.49-1.10), 0.63 (0.46-0.87), 0.71 (0.55-0.90), and 0.65 (0.54-0.78), compared with placebo, metformin, sulfonylurea, TZD, and DPP4 inhibitor, respectively. The relative risks of cardiovascular-related mortality with SGLT2 inhibitor use were 0.61 (0.50-0.76), 0.81(0.36-1.90), 0.52(0.31-0.88), 0.66(0.49-0.91), and 0.61(0.48-0.77), compared with placebo, metformin, sulfonylurea, TZD, and DPP4 inhibitor, respectively. The relative risks of ACS with SGLT2 inhibitor use was consistent with that of all-cause mortality. SGLT2 inhibitor use was associated with a lower risk of ACS than the other OADs and placebo. The relative risks of MI with SGLT2 inhibitor use were 0.77 (0.63-0.93) and 0.75 (0.60-0.94), compared with placebo and DPP4 inhibitor, respectively. The currently available data provide the evidence of cardiovascular benefit from use of SGLT2 inhibitors to patients with type 2 diabetes, although additional results from ongoing studies will be pivotal.

BackgroundVarious diabetes risk scores composed of non-laboratory parameters have been developed, but only a few studies performed cross-validation of these scores and a comparison with laboratory parameters. We evaluated the performance of diabetes risk scores composed of non-laboratory parameters, including a recently published Korean risk score (KRS), and compared them with laboratory parameters.MethodsThe data of 26,675 individuals who visited the Seoul National University Hospital Healthcare System Gangnam Center for a health screening program were reviewed for cross-sectional validation. The data of 3,029 individuals with a mean of 6.2 years of follow-up were reviewed for longitudinal validation. The KRS and 16 other risk scores were evaluated and compared with a laboratory prediction model developed by logistic regression analysis.ResultsFor the screening of undiagnosed diabetes, the KRS exhibited a sensitivity of 81%, a specificity of 58%, and an area under the receiver operating characteristic curve (AROC) of 0.754. Other scores showed AROCs that ranged from 0.697 to 0.782. For the prediction of future diabetes, the KRS exhibited a sensitivity of 74%, a specificity of 54%, and an AROC of 0.696. Other scores had AROCs ranging from 0.630 to 0.721. The laboratory prediction model composed of fasting plasma glucose and hemoglobin A1c levels showed a significantly higher AROC (0.838, P < 0.001) than the KRS. The addition of the KRS to the laboratory prediction model increased the AROC (0.849, P = 0.016) without a significant improvement in the risk classification (net reclassification index: 4.6%, P = 0.264).ConclusionsThe non-laboratory risk scores, including KRS, are useful to estimate the risk of undiagnosed diabetes but are inferior to the laboratory parameters for predicting future diabetes.

In this study, the effects of autologous islet transplantation (ITx) were compared to those of oral antidiabetic drugs (OAD) after distal pancreatectomy (NCT01922492). We enrolled nondiabetic patients who underwent distal pancreatectomy for benign tumors. In the ITx group, islets were isolated from the normal part of the resected pancreas and implanted via the portal vein. Patients who did not receive ITx were regularly monitored and were enrolled in the OAD group if diabetes mellitus developed. The OAD group was treated with metformin with or without vildagliptin. Metabolic parameters were monitored for 12 months postoperatively. Nine patients in the ITx group and 10 in the OAD group were included in the analysis. After 12 months, hemoglobin A1c significantly increased by 5% of the baseline in each group. Area under the curve for blood glucose (AUC glucose ) of the 75-g oral glucose tolerance test increased similarly in the immediate postoperative period in both groups but significantly reduced only in the ITx group thereafter. Insulinogenic index (INSindex) significantly decreased from 25.6 ± 18.9 to 4.7 ± 3.7 in the OAD group, while no significant change was observed in the ITx group (from 15.0 ± 4.5 to 11.0 ± 8.2). In the multiple regression analysis, ITx was an independent factor for changes in AUC glucose and INSindex. In addition, changes in INSindex in the ITx group after postoperative 6 months were associated with the efficacy of islet isolation, amount of grafts, and peak serum HMGB1 and VEGF levels after ITx. ITx was superior to OAD in maintaining insulin secretory capacity and glucose tolerance after distal pancreatectomy.

Using high performance parallel and distributed computing systems, we can collect, generate, handle, and transmit ever-increasing amounts of data. However, these technical advancements also allow malicious individuals to obtain high computational power to attack cryptosystems. Traditional cryptosystem countermeasures have been somewhat passive in response to this change, because they simply increase computational costs by increasing key lengths. Cryptosystems that use the conventional countermeasures cannot preserve secrecy against various cryptanalysis approaches, including side channel analysis and brute-force attacks. Therefore, two new countermeasures have recently been introduced to address this problem: honey encryption and the structural code scheme. Both methods look different; however, they have similar security goals and they both feature distribution transforming encoders based on statistical schemes. We unify them into a statistical code scheme in this study. After explaining the statistical code scheme, we describe the structural code scheme, which has not been adopted as widely as the honey encryption.

BackgroundRecently, it has been suggested that enhancement of incretin effect improves cardiac function. We investigated the effect of a DPP-IV inhibitor, des-fluoro-sitagliptin, in reducing occurrence of restenosis in carotid artery in response to balloon injury and the related mechanisms.Methods and findingsOtsuka Long-Evans Tokushima Fatty rats were grouped into four: control (normal saline) and sitagliptin 100, 250 and 500 mg/kg per day (n = 10 per group). Sitagliptin or normal saline were given orally from 1 week before to 2 weeks after carotid injury. After 3 weeks of treatment, sitagliptin treatment caused a significant and dose-dependent reduction in intima-media ratio (IMR) in obese diabetic rats. This effect was accompanied by improved glucose homeostasis, decreased circulating levels of high-sensitivity C-reactive protein (hsCRP) and increased adiponectin level. Moreover, decreased IMR was correlated significantly with reduced hsCRP, tumor necrosis factor-α and monocyte chemoattractant protein-1 levels and plasminogen activator inhibitor-1 activity. In vitro evidence with vascular smooth muscle cells (VSMCs) demonstrated that proliferation and migration were decreased significantly after sitagliptin treatment. In addition, sitagliptin increased caspase-3 activity and decreased monocyte adhesion and NFκB activation in VSMCs.ConclusionsSitagliptin has protective properties against restenosis after carotid injury and therapeutic implications for treating macrovascular complications of diabetes.

EGb761, a standardized Ginkgo biloba extract, has antioxidant and antiplatelet aggregation and thus might protect against atherosclerosis. However, molecular and functional properties of EGb761 and its major subcomponents have not been well characterized. We investigated the effect of EGb761 and its major subcomponents (bilobalide, kaemferol, and quercetin) on preventing atherosclerosis in vitro, and in a rat model of type 2 diabetes.EGb761 (100 and 200 mg/kg) or normal saline (control) were administered to Otsuka Long-Evans Tokushima Fatty rats, an obese insulin-resistant rat model, for 6 weeks (from 3 weeks before to 3 weeks after carotid artery injury). Immunohistochemical staining was performed to investigate cell proliferation and apoptosis in the injured arteries. Cell migration, caspase-3 activity and DNA fragmentation, monocyte adhesion, and ICAM-1/VCAM-1 levels were explored in vitro. Treatment with EGb761 dose-dependently reduced intima-media ratio, proliferation of vascular smooth muscle cells (VSMCs) and induced greater apoptosis than the controls. Proliferation and migration of VSMCs in vitro were also decreased by the treatment of EGb761. Glucose homeostasis and circulating adiponectin levels were improved, and plasma hsCRP concentrations were decreased in the treatment groups. Caspase-3 activity and DNA fragmentation increased while monocyte adhesion and ICAM-1/VCAM-1 levels decreased significantly. Among subcomponents of EGb761, kaemferol and quercetin reduced VSMC migration and increased caspase activity.EGb761 has a protective role in the development of atherosclerosis and is a potential therapeutic agent for preventing atherosclerosis.

BackgroundFat accumulation in android compartments may confer increased metabolic risk. The incremental utility of measuring regional fat deposition in association with metabolic syndrome (MS) has not been well described particularly in an elderly population.Methods and findingsAs part of the Korean Longitudinal Study on Health and Aging, which is a community-based cohort study of people aged more than 65 years, subjects (287 male, 75.9±8.6 years and 278 female, 76.0±8.8 years) with regional body composition data using Dual energy X-ray absorptiometry for android/gynoid area, computed tomography for visceral/subcutaneous adipose tissue (VAT/SAT), and cardiometabolic markers including adiponectin and high-sensitivity CRP were enrolled. We investigated the relationship between regional body composition and MS in multivariate regression models. Mean VAT and SAT area was 131.4±65.5 cm(2) and 126.9±55.2 cm(2) in men (P = 0.045) and 120.0±46.7 cm(2) and 211.8±65.9 cm(2) in women (PConclusionsOur findings are consistent with the hypothesized role of android fat as a pathogenic fat depot in the MS. Measurement of android fat may provide a more complete understanding of metabolic risk associated with variations in fat distribution.