Repository UNSAP (Universitas Sebelas April)
Not a member yet
1145 research outputs found
Sort by
Putting low-level vision into global context: Why vision cannot be reduced to basic circuits
No full textTo cope with the complexity of vision, most models in neuroscience and computer vision are of hierarchical and feedforward nature. Low-level vision, such as edge and motion detection, is explained by basic low-level neural circuits, whose outputs serve as building blocks for more complex circuits computing higher level features such as shape and entire objects. There is an isomorphism between states of the outer world, neural circuits, and perception, inspired by the positivistic philosophy of the mind. Here, we show that although such an approach is conceptually and mathematically appealing, it fails to explain many phenomena including crowding, visual masking, and non-retinotopic processing. (C) 2015 Elsevier Ltd. All rights reserved.LPS
Aerodynamic characterization of a delta-wing UAV based on real flight data
No full textVehicle Dynamic Model based navigation represents a novel approach to autonomous navigation for small UAVs. It significantly improves the navigation solution under GNSS outage condition without adding additional sensors to the platform. This paper proposes an alternative method for determination of the aerodynamic coefficients characterizing the system dynamics needed to enable VDM-based navigation. It relies on the usage of post-processed traditional INS/GNSS navigation solution for the estimation of initial values of the aerodynamic coefficients. The importance of optimal selection of experimental data is highlighted, proposing an evaluation criterion whose benefit is the increase of the statistical significance of the coefficient estimation.TOP
Asynchronous Byzantine Machine Learning (the case of SGD)
No full textAsynchronous distributed machine learning solutions have proven very effective so far, but always assuming perfectly functioning workers. In practice, some of the workers can however exhibit Byzantine behavior, caused by hardware failures, software bugs, corrupt data, or even malicious attacks. We introduce Kardam, the first distributed asynchronous stochastic gradient descent (SGD) algorithm that copes with Byzantine workers. Kardam consists of two complementary components: a filtering and a dampening component. The first is scalar-based and ensures resilience against 1/3 Byzantine workers. Essentially, this filter leverages the Lipschitzness of cost functions and acts as a self-stabilizer against Byzantine workers that would attempt to corrupt the progress of SGD. The dampening component bounds the convergence rate by adjusting to stale information through a generic gradient weighting scheme. We prove that Kardam guarantees almost sure convergence in the presence of asynchrony and Byzantine behavior, and we derive its convergence rate. We evaluate Kardam on the CIFAR100 and EMNIST datasets and measure its overhead with respect to non Byzantine-resilient solutions. We empirically show that Kardam does not introduce additional noise to the learning procedure but does induce a slowdown (the cost of Byzantine resilience) that we both theoretically and empirically show to be less than f/n, where f is the number of Byzantine failures tolerated and n the total number of workers. Interestingly, we also empirically observe that the dampening component is interesting in its own right for it enables to build an SGD algorithm that outperforms alternative staleness-aware asynchronous competitors in environments with honest workers.DC
Intracellular nanomanipulation by a photonic-force microscope with real-time acquisition of a 3D stiffness matrix
No full textA traditional photonic-force microscope (PFM) results in huge sets of data, which requires tedious numerical analysis. In this paper, we propose instead an analog signal processor to attain real-time capabilities while retaining the richness of the traditional PFM data. Our system is devoted to intracellular measurements and is fully interactive through the use of a haptic joystick. Using our specialized analog hardware along with a dedicated algorithm, we can extract the full 3D stiffness matrix of the optical trap in real time, including the off-diagonal cross-terms. Our system is also capable of simultaneously recording data for subsequent offline analysis. This allows us to check that a good correlation exists between the classical analysis of stiffness and our real-time measurements. We monitor the PFM beads using an optical microscope. The force-feedback mechanism of the haptic joystick helps us in interactively guiding the bead inside living cells and collecting information from its ( possibly anisotropic) environment. The instantaneous stiffness measurements are also displayed in real time on a graphical user interface. The whole system has been built and is operational; here we present early results that confirm the consistency of the real-time measurements with offline computations.LPMVLI
Analysis of walking in five Swiss cities: a quantitative and spatial approach
No full textMany European cities are experiencing an apparent shift in mobility patterns. Walking and cycling are becoming increasingly popular, while many households are deciding not to have a car. These trends feed into a new concept of urban proximity, within which walking is to be considered as a transport mode in its own right. However, little is known about the social and spatial determinants of urban walking, which is why the present study seeks to ask the two following questions: What is the profile of urban walkers in the five largest conurbations in Switzerland: Basel, Bern, Geneva, Lausanne and Zurich? And to what extent can the level of walking in a given area be explained by urban density or urban function (mainly residential; mainly employment; mixed residential and employment)? To answer these questions, we analysed data from the 2010 Swiss transport micro-census using a quantitative and spatial approach, introducing sub-sectors which we characterised according to jobs/residents ratios and measures of urban density. Results show that urban density and urban function do play a role in stimulating or impeding walking behaviours, but this effect is weak compared to the effect of individual characteristics. Intriguing differences were found between the French-speaking cities, Geneva and Lausanne, which display significantly more walking, and the three German-speaking cities, which have less walking but better public transportation systems. The article concludes with a contribution to the research agenda: that the link between walking and public transport use in medium-sized cities should be investigated at the European level.LASU
Numerical Analysis of the Effects of the Magnetic Self-Field on the Transport Properties of a Multilayer HTS Cable
No full textIn this paper the effects of the magnetic self-field on the transport properties of a multi-layer high-Tc superconducting (HTS) cable are investigated by means of 2D finite element method (FEM) simulations. Analyzed is a 3-layer HTS cable, but the developed methods can be used for a different number of layers. The superconductor is described by the non-linear power-law relation E=Ec(J/Jc)^n, where the parameters Jc and n depend on the magnetic field experienced by the material. This dependence decreases the global transport capacity of the superconductor, enhancing its AC losses. It is shown that, especially at high transport currents, the AC losses are considerably higher than in the case where the dependence on the magnetic field is neglected. A simple electrical model, considering the cable from macroscopic point of view, has been proposed for finding the optimal winding pitches, leading to a uniform current repartition. The use of this electrical model allows to overcome the difficulties of direct 3D FEM computations. In addition, the rapidity of solutions by the electric model gives the possibility of testing quickly many geometrical configurations in order to find the ones leading to an even current repartition. This optimization process would not be possible with detailed FEM simulations.LANOSSUPR
Why are the Interaction Energies of Charge-Transfer Complexes Challenging for DFT?
No full textThe description of ground state charge-transfer complexes is highly challenging. Illustrative examples include large overestimations of charge-transfer by local and semi-local density functional approximations as well as inaccurate binding energies. It is demonstrated here that standard density functionals fail to accurately describe interaction energies of charge-transfer complexes not only because of the missing long-range exchange as generally assumed, but also as a result of the neglect of weak interactions. Thus accounting for the missing van der Waals interactions is of key importance. These assertions, based on the evaluation of the extent of stabilization due to dispersion using both DFT coupled with our recent density-dependent dispersion correction (dDsC) and high-level ab initio computations, reflect the imperfect error-cancellation between the overestimation of charge-transfer and the missing long-range interactions. An in-depth energy decomposition analysis of an illustrative series of four small ambidentate molecules (HCN, HNC, HF and ClF) bound together with NF3 provides the main conclusions, which are validated on a prototypical organic charge-transfer complex (i.e., tetrathiafulvalene-tetracyanoquinodimethane, TTF-TCNQ). We establish that the interaction energies for charge-transfer complexes can only be properly described when using well-balanced functionals such as PBE0-dDsC, M06-2X and LC-BOP-LRD.LCM
Unraveling the Relevance of Graphene-Fluid Hydrodynamic Coupling on the Exfoliation of Graphite in Water
No full textLiquid-phase exfoliation via shear flow is a widely adopted technique for the large-scale production of graphene. However, the underlying nano- and microscale exfoliation mechanisms remain poorly understood. In this work, we address this issue by performing hybrid nonequilibrium hydrodynamic simulations of coarse-grained defect-free graphite nanoplatelets immersed in a mesoscopic water fluid via the lattice Boltzmann method. This approach enables us to investigate graphene exfoliation up to 100 nm in length. Nonequilibrium effects, such as tumbling, alignment, and bending, are demonstrated. In particular, we reveal that due to the graphene-fluid hydrodynamic coupling, the graphite dynamics distorts the surrounding shear flow and reduces the local shear stress, thereby leading to an increase in the critical shear rate by a factor of 2 ∼ 4. This statement is fully supported by a theoretical analysis using a force-based criterion, i.e., overcoming the maximum interlayer van der Waals attraction, and hierarchical simulations: athermal and no coupling; athermal and hydrodynamic coupling; and thermal and hydrodynamic coupling. Our work unravels the paramount relevance of hydrodynamic coupling on graphene exfoliation and paves the way toward achieving large-scale nonequilibrium graphene simulations reminiscent of experiments.EPF
