Adversarial training is considered the most powerful approach for robustness against attacks on deep neural networks involving adversarial examples. However, recent works have shown that the similar robustness level c...
详细信息
ISBN:
(纸本)9781665423281
Adversarial training is considered the most powerful approach for robustness against attacks on deep neural networks involving adversarial examples. However, recent works have shown that the similar robustness level can be achieved by other means, namely interpretability-based regularization. We evaluate these interpretability-based approaches on real-world ResNet models trained on CIFAR-10 and ImageNet datasets. Our results show that interpretability can marginally improve robustness when combined with adversarial training, however, they bring additional computational complexity making these approaches questionable for such models and datasets.
In this paper we present preliminary results of the multi-fractal analysis for describing the behavior of turbulence evolution in the wake behind two model wind turbines using the program ImaCal. The wind energy is an...
详细信息
In this paper we present preliminary results of the multi-fractal analysis for describing the behavior of turbulence evolution in the wake behind two model wind turbines using the program ImaCal. The wind energy is an important component of renewable energy sources. Nowadays the research community focuses on studying the turbulent wakes behind wind turbines and its interactions with the Atmospheric Boundary Layer. For finding the fractal dimension a multi-fractal analysis was performed using the results of a numerical simulation of the flow over two model wind turbines. The latter was carried out by means of the Large Eddy method with the Lagrangian Dynamic Smagorinsky subgrid model. The SOWFA library was employed as a part of OpenFoam software to perform the simulation.
Cloud storages delivered as services are made available to the general public. As cloud storage provider prices become low, they have moved into the mainstream of storage technology. However, there are various factors...
详细信息
ISBN:
(数字)9781728144849
ISBN:
(纸本)9781728144856
Cloud storages delivered as services are made available to the general public. As cloud storage provider prices become low, they have moved into the mainstream of storage technology. However, there are various factors that cause many potential users do not use it intensively. There exist high risks for confidentiality, integrity, and availability violation associated with the loss of information, denial of access, technical failures, etc. In this article, we propose a two-level secret sharing scheme (TL-SSS) based on a residue number system (RNS) for a configurable, reliable, and secure distributed data storage. RNS moduli of a special type increase the reliability of the data storage system and reduce the computational complexity of the data encoding and decoding from linear-logarithmic to linear. TL-SSS is the weighted data access structure. It creates and distributes data shares according to the characteristics of the cloud storages under various scenarios. We provide a solution that improves system reliability without reduction of the security level. In contrast to classical solutions, it can restore the data with less available shares than the state-of-the-art approaches.
The paper discusses the capabilities of the iceFoam solver for modeling the ice accretion processes for various swept wing profiles and 3D swept wing. The Euler-Lagrangian approach is used to describe the gas-droplet ...
The paper discusses the capabilities of the iceFoam solver for modeling the ice accretion processes for various swept wing profiles and 3D swept wing. The Euler-Lagrangian approach is used to describe the gas-droplet medium. To determine the shape of ice, the model of a liquid film according to the shallow water theory is used. Three different version of solver iceFoam are discussed. The features of the solver algorithm for parallel computations are explained. The numerical cases with swept wing were run for “Rime Ice” and “Glaze Ice”. The distribution of ice thickness and temperature were defined for swept wing with GLC-305 airfoil using iceDyMFoam3. The calculation results are compared with the data of experiments by M. Papadakis. The computations were run on the high-performance cluster UNIHUB of ISP RAS, HPC Kurchatov institute.
The study of icing processes on elements of an airplane, helicopter, UAV, wind turbines, and power line wires is an urgent task in connection with flight safety and reliable operation of ground equipment. The features...
The study of icing processes on elements of an airplane, helicopter, UAV, wind turbines, and power line wires is an urgent task in connection with flight safety and reliable operation of ground equipment. The features of the numerical simulation of multicomponent reactive mixture flow using the iceFoam solver were considered. The Euler-Lagrangian approach was used to describe the behavior of the two-phase flow with air and liquid droplets. The specifics of the construction of the project on the basis of the computation of the liquid film model for the conditions of the experimental setup for the 2D cylinder and NACA0012 airfoil are discussed. The results of computations of the main parameters of the ice accretion process are given, the efficiency of the iceFoam solver for parallel simulation on two different grids was shown. The computations were run on the high-performance cluster UNIHUB of ISP RAS.
We address dynamic elasticity issues of VM provisioning in a heterogeneous distributed computing environment that integrates resources of a data center. We consider the scenario when the center includes dedicated reso...
详细信息
ISBN:
(纸本)9781538678800
We address dynamic elasticity issues of VM provisioning in a heterogeneous distributed computing environment that integrates resources of a data center. We consider the scenario when the center includes dedicated resources (private cloud) for providing virtualized service and non-dedicated resources for Grid-computing. Existing platforms for the resource virtualization does not support management of such an infrastructure. We propose an approach to a job management based on the dynamic elasticity of virtual machines provisioning using resources of both types. We develop the multi¬agent job scheduler for dedicated resources and hypervisor shell to launch virtual machines through queues of resource management systems in non-dedicated resources. The scheduler provides a dynamic elasticity of virtual machine provisioning. Advantages of the offered approach to the resource virtualization are demonstrated by an example of a job flow management for a scalable application to solve the complicated practical problem. It is related to the energy security of Vietnam. Provided experiments show that using the developed tools together with the platform for the resource virtualization enables agents to significantly speed up the problem-solving process.
Features of numerical simulation of multicomponent reactive mixture flows using the OpenFOAM package reactingCentralFoam solver are considered. The specifics of the construction of the project on the basis of the calc...
Features of numerical simulation of multicomponent reactive mixture flows using the OpenFOAM package reactingCentralFoam solver are considered. The specifics of the construction of the project on the basis of the calculation of the combustion of the oxygen-hydrogen mixture for the conditions of the experimental setup “Penn State Pre-burner Combustor” are discussed. The results of computations of the main parameters of the combustion process are given, the efficiency of the applied solver is shown.
Various models of solubility of gas in two-phase ows of gas and liquid are considered. Such phenomena play an important role in various technical devices especially at modern nuclear power plants. Mathematical models ...
Various models of solubility of gas in two-phase ows of gas and liquid are considered. Such phenomena play an important role in various technical devices especially at modern nuclear power plants. Mathematical models of bubble ows with gas dissolution, gas phase model, and model of liquid phase dissolution were considered and discussed.
Results of QGD (quasi-gas dynamics) equations approximation using OpenFOAM Finite Volume Method library are presented. The new developed solver QGDFoam compared to standard OpenFOAM solver rhoCentralFoam. Two approach...
详细信息
Results of QGD (quasi-gas dynamics) equations approximation using OpenFOAM Finite Volume Method library are presented. The new developed solver QGDFoam compared to standard OpenFOAM solver rhoCentralFoam. Two approaches for the approximation of QGD terms are considered and compared - (1) least square method and (2) face normal derivative. Qualitative analysis showed that the first approach is more accurate compared with the second one and is preferable for reproducing complex shock interactions during unsteady flow. It was shown that the second approach produces the solution, similar to rhoCentralFoam solver with upwind interpolation. Being more diffusive than rhoCentralFoam solver with TVD schemes, QGDFoam, allows, however, to reproduce flows at wider Mach number range without introducing too much numerical diffusion comparing to upwind interpolation. Parallel scalability of developed solver has been checked. It scales for meshes down to approximately 1000 cells per node.
The wind energy is an important part of renewable energy. The wind farms can operate in various climatic conditions on a large territory of Russian Federation. The features of numerical simulation of turbulent flows u...
The wind energy is an important part of renewable energy. The wind farms can operate in various climatic conditions on a large territory of Russian Federation. The features of numerical simulation of turbulent flows using the OpenFOAM package and SOWFA library are considered. The mathematical model has the equations for mass, momentum and energy conservation for incompressible flow. Large-eddy simulation has been applied in the context of wind turbines operation. Lagrangian-averaged scale-independent dynamic Smagorinsky model is used. Two solvers ABLSolver and pisoFoamTurbine have been used for simulations. The results of computations for two test cases (2 and 12 wind turbines) with definition of the main flow parameters are given, the efficiency of the applied solver is shown.
暂无评论