2995 research outputs found
Sort by
Glyph: Fast and Accurately Training Deep Neural Networks on Encrypted Data
No full textBig data is one of the cornerstones to enabling and training deep neural networks (DNNs). Because of the lack of expertise, to gain benefits from their data, average users have to rely on and upload their private data to big data companies they may not trust. Due to the compliance, legal, or privacy constraints, most users are willing to contribute only their encrypted data, and lack interests or resources to join the training of DNNs in cloud. To train a DNN on encrypted data in a completely non-interactive way, a recent work proposes a fully homomorphic encryption (FHE)-based technique implementing all activations in the neural network by \textit{Brakerski-Gentry-Vaikuntanathan (BGV)}-based lookup tables. However, such inefficient lookup-table-based activations significantly prolong the training latency of privacy-preserving DNNs. In this paper, we propose, Glyph, a FHE-based scheme to fast and accurately train DNNs on encrypted data by switching between TFHE (Fast Fully Homomorphic Encryption over the Torus) and BGV cryptosystems. Glyph uses logic-operation-friendly TFHE to implement nonlinear activations, while adopts vectorial-arithmetic-friendly BGV to perform multiply-accumulation (MAC) operations. Glyph further applies transfer learning on the training of DNNs to improve the test accuracy and reduce the number of MAC operations between ciphertext and ciphertext in convolutional layers. Our experimental results show Glyph obtains the state-of-the-art test accuracy, but reduces the training latency by 99% over the prior FHE-based technique on various encrypted datasets
Psychological threat avoidance as a barrier to HIV testing in gay/bisexual men
No full textThe present study examined how three psychosocial barriers—anticipated HIV stigma, HIV infectiousness-reduction beliefs, and optimism about available HIV treatments—related to HIV testing history and acceptance of an at-home HIV test among men who have sex with men. We also examined the mediating role of a variable that affects medical screening for other health conditions but has not yet been investigated in HIV contexts: the tendency to avoid psychologically threatening information. Volunteers completed a paper and pencil survey and were offered a free at-home HIV test during the 2015 Atlanta Pride Festival in Atlanta, GA. Anticipated HIV stigma, infectiousness beliefs, and treatment optimism were inconsistently related to HIV testing history and acceptance of an at-home HIV test, but all had direct effects on the desire to avoid HIV information. In a mediation model, each of these psychosocial barriers had indirect effects on both HIV testing outcomes via information avoidance. These findings suggest that information avoidance is an important proximal HIV testing barrier, thus providing a novel target for interventions and information campaigns
Structure of the Large Extracellular Loop of FtsX and Its Interaction with the Essential Peptidoglycan Hydrolase PcsB in Streptococcus pneumoniae
No full textStreptococcus pneumoniae is a leading killer of infants and immunocompromised adults and has become increasingly resistant to major antibiotics. Therefore, the development of new antibiotic strategies is desperately needed. Targeting bacterial cell division is one such strategy, specifically by targeting proteins that are essential for the synthesis and breakdown of peptidoglycan. One complex important to this process is FtsEX. FtsEX comprises a cell division-regulating integral membrane protein (FtsX) and a cytoplasmic ATPase (FtsE) that resembles an ATP-binding cassette (ABC) transporter. Here, we present nuclear magnetic resonance (NMR) solution structural and crystallographic models of the large extracellular domain of FtsX, denoted extracellular loop 1 (ECL1). The structure of ECL1 reveals an upper extended β-hairpin and a lower α-helical lobe, each extending from a mixed α-β core. The helical lobe mediates a physical interaction with the peptidoglycan hydrolase PcsB via the coiled-coil domain of PcsB (PscBCC). Characterization of S. pneumoniae strain D39-derived strains harboring mutations in the α-helical lobe shows that this subdomain is essential for cell viability and required for proper cell division of S. pneumoniae
Developments in the Laws Affecting Electronic Payments and Financial Services
No full textThis short article surveys developments in the law affecting electronic payments and financial services from June 1, 2017 to June 1, 2018. During this period, significant developments occurred that affected the regulation of initial coin offerings (ICOs), the Office of the Comptroller of the Currency’s proposal to issue “special purpose national bank charters” to FinTech companies, the CFPB’s final regulation of prepaid, general-purpose cards, state regulation of payroll cards, and how lawyers taking cryptocurrencies from clients as payment for services or for safekeeping should protect them. The survey also presents newly issued BitLicenses under the New York Department of Financial Services’ 2015 “BitLicense” regulation
More Steps Toward Fully Electronic Interbank Check Collection and Return: Amendments to Federal Reserve Board Regulation CC and a Regulatory Resolution of a Circuit Split
No full textThis article analyzes two actions in 2017 and 2018, respectively, by the Board of Governors of the Federal Reserve System that amend Regulation CC, which governs expedited deposit availability and collection of checks generally and implements the Expedited Funds Availability Act of 1987 and the Check Clearing for the 21st Century Act of 2003. It selects examples from the two sets of amendments that highlight regulatory strategies being used by the Board to facilitate faster payments through the movement of electronic images of checks or electronic information among banks in the check-collection process. Those strategies involve creation of new forms of instruments that can be treated as checks or used to return checks to depositary banks, and, in the 2018 amendments, establishment of a presumption that reallocates risks in collection long a feature of Anglo-American payments law, the doctrine of Price v. Neal. The strategies offer evidence of the surgical precision with which the Board has acted over the past 30 years to modernize check collection regulation with little use of its authority to preempt state laws, primarily Articles 3 and 4 of the Uniform Commercial Code (“UCC”). This article does not describe or evaluate all provisions of Regulation CC that the Board revised or added in 2017 and 2018
Competition between Normative and Drug-Induced Virus Self-Assembly Observed with Single-Particle Methods
No full textDisruption of virus capsid assembly has compelling antiviral potential that has been applied to hepatitis B virus (HBV). HBV core protein assembly can be modulated by heteroaryldihydropyrimidines (HAPs), and such molecules are collectively termed core protein allosteric modulators (CpAMs). Although the antiviral effects of CpAMs are acknowledged, the mechanism of action remains an open question. Challenging aspects of characterizing misdirected assembly are the large size and nonuniform nature of the final particles. In this study of HBV assembly, we observed a competition between normative and CpAM-induced aberrant assembly with electron microscopy and resistive-pulse sensing on nanofluidic devices. This competition was a function of the strength of the association energy between individual core proteins, which is proportional to ionic strength. At strong association energy, assembly reactions primarily yielded morphologically normal HBV capsids, despite the presence of HAP-TAMRA. At weak association energy, HAP-TAMRA led to increased assembly product size and disrupted morphology. The smallest particles were T = 4 icosahedra, whereas the larger particles were defective spheres, ellipsoids, and bacilliform cylinders, with regions of T = 4 geometry interspersed with flat regions. Deviation from spherical geometry progressively increased with particle size, which is consistent with the interpretation of a competition between two alternative assembly pathways
Machine Learning for Parameter Auto-tuning in Molecular Dynamics Simulations: Efficient Dynamics of Ions near Polarizable Nanoparticles
No full textSimulating the dynamics of ions near polarizable nanoparticles (NPs) using coarse-grained models is extremely challenging due to the need to solve the Poisson equation at every simulation timestep. Recently, a molecular dynamics (MD) method based on a dynamical optimization framework bypassed this obstacle by representing the polarization charge density as virtual dynamic variables and evolving them in parallel with the physical dynamics of ions. We highlight the computational gains accessible with the integration of machine learning (ML) methods for parameter prediction in MD simulations by demonstrating how they were realized in MD simulations of ions near polarizable NPs. An artificial neural network–based regression model was integrated with MD simulation and predicted the optimal simulation timestep and optimization parameters characterizing the virtual system with 94.3% success. The ML-enabled auto-tuning of parameters generated accurate dynamics of ions for ≈ 10 million steps while improving the stability of the simulation by over an order of magnitude. The integration of ML-enhanced framework with hybrid Open Multi-Processing / Message Passing Interface (OpenMP/MPI) parallelization techniques reduced the computational time of simulating systems with thousands of ions and induced charges from thousands of hours to tens of hours, yielding a maximum speedup of ≈ 3 from ML-only acceleration and a maximum speedup of ≈ 600 from the combination of ML and parallel computing methods. Extraction of ionic structure in concentrated electrolytes near oil–water emulsions demonstrates the success of the method. The approach can be generalized to select optimal parameters in other MD applications and energy minimization problems
Lot-to-lot Variation in Adeno-associated Virus Serotype 9 (AAV9) Preparations
No full textViral vectors are complex drugs that pose a particular challenge for manufacturing. Previous studies have shown that, unlike small-molecule drugs, vector preparations do not yield a collection of identical particles. Instead, a mixture of particles that vary in capsid stoichiometry and impurities is created, which may differ from lot to lot. The consequences of this are unclear, but conflicting reports regarding the biological properties of vectors, including transduction patterns, suggest that this variability may have an effect. However, other variables, including differences in animal strains and techniques, make it difficult to identify a cause. Here, we report lot-to-lot variation in spinal cord gray matter transduction following intrathecal delivery of self-complementary adeno-associated virus serotype 9 vectors. Eleven lots of vector were evaluated from six vector cores, including one preclinical/Good Laboratory Practice lot. Eight of the lots, including the preclinical lot, failed to transduce the gray matter, whereas the other three provided robust transduction. The cause for this variation is unknown, but it did not correlate with vector titer, buffer, or purification method. These results highlight the need to identify the cause of this variation and to develop improved production and quality control methods to ensure lot-to-lot consistency of vector potency
A Brief Report of Sexual Violence among Universities with NCAA Division I Athletic Programs
No full textViolence against women on college campuses continues to be a pervasive public health problem with approximately one in five women experiencing sexual assault and one in nine women experiencing rape while in college. The current study examined relationship and sexual violence among National Collegiate Athletic Association (NCAA) Division I universities. Based on previous research, Division I universities seem to report higher rates of sexual assault, but within-group differences have yet to be examined. The data include 1422 four-year private and public institutions with at least 1000 students who submitted Clery data (2014) on rape, domestic and dating violence, and stalking. Division I campuses reported significantly higher reports of violence against women compared to Division II, III, and universities with no athletic programs. There were no differences in violence reported across the three subdivisions within Division I, however, certain conferences reported significantly higher relationship and sexual violence within the football bowl and football championship subdivisions. These findings have important implications for targeting higher risk campuses, such as the Big 10, Big 12, Ivy League, Pac-12, and SEC with much needed sexual assault prevention programs