Search

This paper aims to develop the study of historical printed ornaments with modern unsupervised computer vision. We highlight three complex tasks that are of critical interest to book historians: clustering, element discovery, and unsupervised change localization. For each of these tasks, we introduce an evaluation benchmark, and we adapt and evaluate state-of-the-art models. Our Rey's Ornaments dataset is designed to be a representative example of a set of ornaments historians would be interested in. It focuses on an XVIIIth century bookseller, Marc-Michel Rey, providing a consistent set of ornaments with a wide diversity and representative challenges. Our results highlight the limitations of state-of-the-art models when faced with real data and show simple baselines such as k-means or congealing can outperform more sophisticated approaches on such data. Our dataset and code can be found at https://printed-ornaments.github.io/.

Light emission of europium (Eu3+) ions placed in the vicinity of optically resonant nanoantennas is usually controlled by tailoring the local density of photon states (LDOS). We show that the polarization and shape of the excitation beam can also be used to manipulate light emission, as azimuthally or radially polarized cylindrical vector beam offers to spatially shape the electric and magnetic fields, in addition to the effect of silicon nanorings (Si-NRs) used as nanoantennas. The photoluminescence mappings of the Eu3+ transitions and the Si phonon mappings are strongly dependent of both the excitation beam and the Si-NR dimensions. The experimental results of Raman scattering and photoluminescence are confirmed by numerical simulations of the near-field intensity in the Si nanoantenna and in the Eu3+-doped film, respectively. The branching ratios obtained from the experimental PL maps also reveal a redistribution of the electric and magnetic emission channels. Our results show that it is possible to spatially control both electric and magnetic dipolar emission of Eu3+ ions by switching the laser beam polarization, hence the near-field at the excitation wavelength, and the electric and magnetic LDOS at the emission wavelength. This paves the way for optimized geometries taking advantage of both excitation and emission processes., Comment: 8 pages, 4 figures

This article focuses on the multi-objective optimization of stochastic simulators with high output variance, where the input space is finite and the objective functions are expensive to evaluate. We rely on Bayesian optimization algorithms, which use probabilistic models to make predictions about the functions to be optimized. The proposed approach is an extension of the Pareto Active Learning (PAL) algorithm for the estimation of Pareto-optimal solutions that makes it suitable for the stochastic setting. We named it Pareto Active Learning for Stochastic Simulators (PALS). The performance of PALS is assessed through numerical experiments over a set of bi-dimensional, bi-objective test problems. PALS exhibits superior performance when compared to other scalarization-based and random-search approaches.

We report the resonantly enhanced radiative emission from a single SiGe quantum dot (QD), which is deterministically embedded into a bichromatic photonic crystal resonator (PhCR) at the position of its largest modal electric field by a scalable method. By optimizing our molecular beam epitaxy (MBE) growth technique, we were able to reduce the amount of Ge within the whole resonator to obtain an absolute minimum of exactly one QD, accurately positioned by lithographic methods relative to the PhCR, and an otherwise flat, a few monolayer thin, Ge wetting layer (WL). With this method, record quality (Q) factors for QD-loaded PhCRs up to $Q\sim 10^5$ are achieved. A comparison with control PhCRs on samples containing a WL but no QDs is presented, as well as a detailed analysis of the dependence of the resonator-coupled emission on temperature, excitation intensity, and emission decay after pulsed excitation. Our findings undoubtedly confirm a single QD in the center of the resonator as a potentially novel photon source in the telecom spectral range.

Morphological neural networks allow to learn the weights of a structuring function knowing the desired output image. However, those networks are not intrinsically robust to lighting variations in images with an optical cause, such as a change of light intensity. In this paper, we introduce a morphological neural network which possesses such a robustness to lighting variations. It is based on the recent framework of Logarithmic Mathematical Morphology (LMM), i.e. Mathematical Morphology defined with the Logarithmic Image Processing (LIP) model. This model has a LIP additive law which simulates in images a variation of the light intensity. We especially learn the structuring function of a LMM operator robust to those variations, namely : the map of LIP-additive Asplund distances. Results in images show that our neural network verifies the required property.

In addition to signature-based and heuristics-based detection techniques, machine learning (ML) is widely used to generalize to new, never-before-seen malicious software (malware). However, it has been demonstrated that ML models can be fooled by tricking the classifier into returning the incorrect label. These studies, for instance, usually rely on a prediction score that is fragile to gradient-based attacks. In the context of a more realistic situation where an attacker has very little information about the outputs of a malware detection engine, modest evasion rates are achieved. In this paper, we propose a method using reinforcement learning with DQN and REINFORCE algorithms to challenge two state-of-the-art ML-based detection engines (MalConv \& EMBER) and a commercial AV classified by Gartner as a leader AV. Our method combines several actions, modifying a Windows portable execution (PE) file without breaking its functionalities. Our method also identifies which actions perform better and compiles a detailed vulnerability report to help mitigate the evasion. We demonstrate that REINFORCE achieves very good evasion rates even on a commercial AV with limited available information.

Monolayers (MLs) of transition metal dichalcogenides (TMDs) such as WSe2 and MoSe2 can be placed by dry stamping directly on broadband dielectric resonators, which have the ability to enhance the spontaneous emission rate and brightness of solid-state emitters at room temperature. We show strongly enhanced emission and directivity modifications in room temperature photoluminescence mapping experiments. By varying TMD material (WSe2 versus MoSe2) transferred on silicon nanoresonators with various designs (planarized versus non-planarized), we experimentally separate the different physical mechanisms that govern the global light emission enhancement. For WSe2 and MoSe2 we address the effects of Mie Resonances and strain in the monolayer. For WSe2 an important additional contribution comes from out-of-plane exciton dipoles. This paves the way for more targeted designs of TMD-Si nanoresonator structures for room temperature applications., Comment: 10 pages, 5 figures

High refractive index dielectric nanostructures provide original optical properties thanks to the occurrence of size- and shape-dependent optical resonance modes. These modes commonly present a spectral overlap of broad, low-order modes (\textit{e.g}. dipolar modes) and much narrower, higher-order modes. The latter are usually characterized by a rapidly varying frequency-dependent phase, which - in superposition with the lower order mode of approximately constant phase - leads to typical spectral features known as Fano resonances. Interestingly, such Fano resonances occur in dielectric nanostructures of the simplest shapes. In spheroidal nanoparticles, interference between broad magnetic dipole and narrower electric dipole modes can be observed. In high aspect-ratio structures like nanowires, either the electric or the magnetic dipolar mode (depending on the illumination conditions) interferes with higher order multipole contributions of the same nature (electric or magnetic). Using the analytical Mie theory, we analyze the occurrence of Fano resonances in high-index dielectric nanowires and discuss their consequences like unidirectional scattering. By means of numerical simulations, we furthermore study the impact on those Fano resonances of the shape of the nanowire cross-sections as well as the coupling of two parallel nanowires. The presented results show that all-dielectric nanostructures, even of simple shapes, provide a reliable low-loss alternative to plasmonic nanoantennas., Comment: 30 pages, 8 figures. To appear as a chapter in the book "Fano resonances in optics and microwaves: Physics and application", Springer (2018). Ed. by E. O. Kamenetskii, A. Sadreev, and A. Miroshnichenko

It has been experimentally demonstrated only recently that a simultaneous excitation of interfering electric and magnetic resonances can lead to uni-directional scattering of visible light in zero-dimensional dielectric nanoparticles. We show both theoretically and experimentally, that strongly anisotropic scattering also occurs in individual dielectric nanowires. The effect occurs even under either pure transverse electric or pure transverse magnetic polarized normal illumination. This allows for instance to toggle the scattering direction by a simple rotation of the incident polarization. Finally, we demonstrate that directional scattering is not limited to cylindrical cross-sections, but can be further tailored by varying the shape of the nanowires., Comment: 12 pages, 6 figures + supporting informations of 12 pages, 19 figures

Common-path digital in-line holography is considered as a valuable 3D diagnostic techniques for a wide range of applications. This configuration is cost effective and relatively immune to variation in the experimental environment. Nevertheless, due to its common-path geometry, the signal to noise-ratio of the acquired hologram is weak as most of the detector (i.e. CCD/CMOS sensor) dynamics is occupied by the reference field signal, whose energy is orders of magnitude higher than the field scattered by the imaged object. As it is intrinsically impossible to modify the ratio of energy of reference to the object field, we propose a co-design approach (Optics/Data Processing) to tackle this issue. The reference to object field ratio is adjusted by adding a 4-f device to a conventional in-line holographic setup, making it possible to reduce the weight of the reference field while keeping the object field almost constant. Theoretical analysis of the Cr\`amer-Rao lower bounds of the corresponding imaging model illustrate the advantages of this approach. These lower bounds can be asymptotically reached using a parametric inverse problems reconstruction. This implementation results in a 60 % gain in axial localization accuracy (for of 100 $\mu$m diameter spherical objects) compared to a classical in-line holography setup .

Asteroid modeling efforts in the last decade resulted in a comprehensive dataset of almost 400 convex shape models and their rotation states. This amount already provided a deep insight into physical properties of main-belt asteroids or large collisional families. We aim to increase the number of asteroid shape models and rotation states. Such results are an important input for various further studies such as analysis of asteroid physical properties in different populations, including smaller collisional families, thermophysical modeling, and scaling shape models by disk-resolved images, or stellar occultation data. This provides, in combination with known masses, bulk density estimates, but constrains also theoretical collisional and evolutional models of the Solar System. We use all available disk-integrated optical data (i.e., classical dense-in-time photometry obtained from public databases and through a large collaboration network as well as sparse-in-time individual measurements from a few sky surveys) as an input for the convex inversion method, and derive 3D shape models of asteroids, together with their rotation periods and orientations of rotation axes. The key ingredient is the support of more that one hundred observers who submit their optical data to publicly available databases. We present updated shape models for 36 asteroids, for which mass estimates are currently available in the literature or their masses will be most likely determined from their gravitational influence on smaller bodies, which orbital deflection will be observed by the ESA Gaia astrometric mission. This was achieved by using additional optical data from recent apparitions for the shape optimization. Moreover, we also present new shape model determinations for 250 asteroids, including 13 Hungarias and 3 near-Earth asteroids., Comment: Accepted for publication in A&A

In-line digital holography is a valuable tool for sizing, locating and tracking micro- or nano-objects in a volume. When a parametric imaging model is available, Inverse Problems approaches provide a straightforward estimate of the object parameters by fitting data with the model, thereby allowing accurate reconstruction. As recently proposed and demonstrated, combining pixel super-resolution techniques with Inverse Problems approaches improves the estimation of particle size and 3D-position. Here we demonstrate the accurate tracking of colloidal particles in Brownian motion. Particle size and 3D-position are jointly optimized from video holograms acquired with a digital holographic microscopy set up based on a "low-end" microscope objective ($\times 20$, $\rm NA\ 0.5$). Exploiting information redundancy makes it possible to characterize particles with a standard deviation of 15 nm in size and a theoretical resolution of 2 x 2 x 5 nm$^3$ for position under additive white Gaussian noise assumption.

Comparing weighted networks in neuroscience is hard, because the topological properties of a given network are necessarily dependent on the number of edges of that network. This problem arises in the analysis of both weighted and unweighted networks. The term density is often used in this context, in order to refer to the mean edge weight of a weighted network, or to the number of edges in an unweighted one. Comparing families of networks is therefore statistically difficult because differences in topology are necessarily associated with differences in density. In this review paper, we consider this problem from two different perspectives, which include (i) the construction of summary networks, such as how to compute and visualize the mean network from a sample of network-valued data points; and (ii) how to test for topological differences, when two families of networks also exhibit significant differences in density. In the first instance, we show that the issue of summarizing a family of networks can be conducted by adopting a mass-univariate approach, which produces a statistical parametric network (SPN). In the second part of this review, we then highlight the inherent problems associated with the comparison of topological functions of families of networks that differ in density. In particular, we show that a wide range of topological summaries, such as global efficiency and network modularity are highly sensitive to differences in density. Moreover, these problems are not restricted to unweighted metrics, as we demonstrate that the same issues remain present when considering the weighted versions of these metrics. We conclude by encouraging caution, when reporting such statistical comparisons, and by emphasizing the importance of constructing summary networks., Comment: 16 pages, 5 figures

In this paper we put forward a systematic and unifying approach to construct gauge invariant composite fields out of connections. It relies on the existence in the theory of a group valued field with a prescribed gauge transformation. As an illustration, we detail some examples. Two of them are based on known results: the first one provides a reinterpretation of the symmetry breaking mechanism of the electroweak part of the Standard Model of particle physics; the second one is an application to Einstein's theory of gravity described as a gauge theory in terms of Cartan connections. The last example depicts a new situation: starting with a gauge field theory on Atiyah Lie algebroids, the gauge invariant composite fields describe massive vector fields. Some mathematical and physical discussions illustrate and highlight the relevance and the generality of this approach., Comment: 22 pages, revised version to appear in Int. J. of Geometric Methods in Modern Physics

Electrically tunable g-factors in quantum dots are highly desirable for applications in quantum computing and spintronics. We report giant modulation of the hole g-factor in a SiGe nanocrystal when an electric field is applied to the nanocrystal along its growth direction. We derive a contribution to the g-factor that stems from an orbital effect of the magnetic field, which lifts the Kramers degeneracy in the nanocrystal by altering the mixing between the heavy and the light holes. We show that the relative displacement between the heavy- and light-hole wave functions, occurring upon application of the electric field, has an effect on the mixing strength and leads to a strong non-monotonic modulation of the g-factor. Despite intensive studies of the g-factor since the late 50's, this mechanism of g-factor control has been largely overlooked in the literature., Comment: 9 pages, 6 figures

In this paper we introduce and study some mathematical structures on top of transitive Lie algebroids in order to formulate gauge theories in terms of generalized connections and their curvature: metrics, Hodge star operator and integration along the algebraic part of the transitive Lie algebroid (its kernel). Explicit action functionals are given in terms of global objects and in terms of their local description as well. We investigate applications of these constructions to Atiyah Lie algebroids and to derivations on a vector bundle. The obtained gauge theories are discussed with respect to ordinary and to similar non-commutative gauge theories., Comment: 30 pages. Final version. arXiv admin note: substantial text overlap with arXiv:1109.4282

Studies of functional MRI data are increasingly concerned with the estimation of differences in spatio-temporal networks across groups of subjects or experimental conditions. Unsupervised clustering and independent component analysis (ICA) have been used to identify such spatio-temporal networks. While these approaches have been useful for estimating these networks at the subject-level, comparisons over groups or experimental conditions require further methodological development. In this paper, we tackle this problem by showing how self-organizing maps (SOMs) can be compared within a Frechean inferential framework. Here, we summarize the mean SOM in each group as a Frechet mean with respect to a metric on the space of SOMs. We consider the use of different metrics, and introduce two extensions of the classical sum of minimum distance (SMD) between two SOMs, which take into account the spatio-temporal pattern of the fMRI data. The validity of these methods is illustrated on synthetic data. Through these simulations, we show that the three metrics of interest behave as expected, in the sense that the ones capturing temporal, spatial and spatio-temporal aspects of the SOMs are more likely to reach significance under simulated scenarios characterized by temporal, spatial and spatio-temporal differences, respectively. In addition, a re-analysis of a classical experiment on visually-triggered emotions demonstrates the usefulness of this methodology. In this study, the multivariate functional patterns typical of the subjects exposed to pleasant and unpleasant stimuli are found to be more similar than the ones of the subjects exposed to emotionally neutral stimuli. Taken together, these results indicate that our proposed methods can cast new light on existing data by adopting a global analytical perspective on functional MRI paradigms., Comment: 23 pages, 5 figures, 4 tables. Submitted to Neuroimage

In this paper we study the local description of spaces of forms on transitive Lie algebroids. We use this local description to introduce global structures like metrics, $\ast$-Hodge operation and integration along the algebraic part of the transitive Lie algebroid (its kernel). We construct a \v{C}ech-de Rham bicomplex with a Leray-Serre spectral sequence. We apply the general theory to Atiyah Lie algebroids and to derivations on a vector bundle.

Spin-selective tunneling of holes in SiGe nanocrystals contacted by normal-metal leads is reported. The spin selectivity arises from an interplay of the orbital effect of the magnetic field with the strong spin-orbit interaction present in the valence band of the semiconductor. We demonstrate both experimentally and theoretically that spin-selective tunneling in semiconductor nanostructures can be achieved without the use of ferromagnetic contacts. The reported effect, which relies on mixing the light and heavy holes, should be observable in a broad class of quantum-dot systems formed in semiconductors with a degenerate valence band., Comment: 8 pages, 5 figures

The epitaxial growth of germanium on silicon leads to the self-assembly of SiGe nanocrystals via a process that allows the size, composition and position of the nanocrystals to be controlled. This level of control, combined with an inherent compatibility with silicon technology, could prove useful in nanoelectronic applications. Here we report the confinement of holes in quantum-dot devices made by directly contacting individual SiGe nanocrystals with aluminium electrodes, and the production of hybrid superconductorsemiconductor devices, such as resonant supercurrent transistors, when the dot is strongly coupled to the electrodes. Charge transport measurements on weakly coupled quantum dots reveal discrete energy spectra, with the confined hole states displaying anisotropic gyromagnetic factors and strong spin-orbit coupling strength with pronounced gate-voltage and magnetic-field dependence., Comment: 35 pages, 6 figures