Feature selection in learning to rank has recently emerged as a crucial issue. Whereas several preprocessing approaches have been proposed, only a few works have been focused on integrating the feature selection into the learning process. In this wor
k, we propose a general framework for feature selection in learning to rank using SVM with a sparse regularization term. We investigate both classical convex regularizations such as $ell_1$ or weighted $ell_1$ and non-convex regularization terms such as log penalty, Minimax Concave Penalty (MCP) or $ell_p$ pseudo norm with $ptextless{}1$. Two algorithms are proposed, first an accelerated proximal approach for solving the convex problems, second a reweighted $ell_1$ scheme to address the non-convex regularizations. We conduct intensive experiments on nine datasets from Letor 3.0 and Letor 4.0 corpora. Numerical results show that the use of non-convex regularizations we propose leads to more sparsity in the resulting models while prediction performance is preserved. The number of features is decreased by up to a factor of six compared to the $ell_1$ regularization. In addition, the software is publicly available on the web.
During the pre-Swift era, a clustering of light curves was observed in the X-ray, optical and infrared afterglow of gamma-ray bursts. We used a sample of 254 GRB X-ray afterglows to check this fact in the Swift era. We corrected fluxes for distance,
time dilation and losses of energy due to cosmological effects. With all our data in hand, we faced with a problem: our data were scattered. We investigated 3 possibilities to explain this, namely: the clustering does not exist, there are problems during calibration of data, and there are instrumental problems. We finally confirm that our sample is consistent with Dainotti correlation.
Astronomical optical interferometers (OI) sample the Fourier transform of the intensity distribution of a source at the observation wavelength. Because of rapid atmospheric perturbations, the phases of the complex Fourier samples (visibilities) canno
t be directly exploited , and instead linear relationships between the phases are used (phase closures and differential phases). Consequently, specific image reconstruction methods have been devised in the last few decades. Modern polychromatic OI instruments are now paving the way to multiwavelength imaging. This paper presents the derivation of a spatio-spectral (3D) image reconstruction algorithm called PAINTER (Polychromatic opticAl INTErferometric Reconstruction software). The algorithm is able to solve large scale problems. It relies on an iterative process, which alternates estimation of polychromatic images and of complex visibilities. The complex visibilities are not only estimated from squared moduli and closure phases, but also from differential phases, which help to better constrain the polychromatic reconstruction. Simulations on synthetic data illustrate the efficiency of the algorithm.
We study the decay $trightarrow cgamma$ with flavor-changing neutral interactions in scalar sector of the type III Two Higgs Doublet Model (THDM-III) with mixing between neutral scalar fields as a result of considering the most general scalar potenti
al. The branching ratio of the decay $Br(trightarrow cgamma)$ is calculated as function of the mixing parameters and masses of the neutral scalar fields. We obtain a $Br(trightarrow cgamma)$ of the order of $10^{-8}$ for the considered regions of the mixing parameters. Finally, one upper bound for the possible events is estimated to be $n=18$ by assuming a expected luminosity of the order of 300 $fb^{-1}$.
We discuss the formalism of two Higgs doublet model type III with CP violation from CP-even CP-odd mixing in the neutral Higgs bosons. The flavor changing interactions among neutral Higgs bosons and fermions are presented at tree level in this type o
f model. These assumptions allow the study rare top decays mediated by neutral Higgs bosons, particularly we are interested in $trightarrow c l^+l^-$. For this process we estimated upper bounds of the branching ratios $textrm{Br}(trightarrow c tau^+tau^-)$ of the order of $10^{-9}sim 10^{-7}$ for a neutral Higgs boson mass of 125 GeV and $tanbeta=1$, 1.5, 2, 2.5. For the case of $trightarrow c tau^+tau^-$ the number of possible events is estimated from 1 to 10 events which could be observed in future experiments at LHC with a luminosity of 300 $textrm{fb}^{-1}$ and 14 GeV for the energy of the center of mass. Also we estimate that the number of events for the process $trightarrow c l^+l^-$ in different scenarios is of order of $2500$.
Distances from the Gaia mission will no doubt improve our understanding of stellar physics by providing an excellent constraint on the luminosity of the star. However, it is also clear that high precision stellar properties from, for example, asteros
eismology, will also provide a needed input constraint in order to calibrate the methods that Gaia will use, e.g. stellar models or GSP_phot. For solar-like stars (F, G, K IV/V), asteroseismic data delivers at the least two very important quantities: (1) the average large frequency separation <Delta_nu> and (2) the frequency corresponding to the maximum of the modulated-amplitude spectrum nu_max. Both of these quantities are related directly to stellar parameters (radius and mass) and in particular their combination (gravity and density). We show how the precision in <Delta_nu>, nu_max, and atmospheric parameters T_eff and [Fe/H] affect the determination of gravity (log g) for a sample of well-known stars. We find that log g can be determined within less than 0.02 dex accuracy for our sample while considering precisions in the data expected for V<12 stars from Kepler data. We also derive masses and radii which are accurate to within 1sigma of the accepted values. This study validates the subsequent use of all of the available asteroseismic data on main sequence solar-like stars from the Kepler field (>500 IV/V stars) in order to provide a very important constraint for Gaia calibration of GSP_phot through the use of log g. We note that while we concentrate on IV/V stars, both the CoRoT and Kepler fields contain asteroseismic data on thousands of giant stars which will also provide useful calibration measures.
The relevance of non-thermal cluster studies and the importance of combining observations of future radio surveys with WFXT data are discussed in this paper.
We discuss the most general formulation of the Two-Higgs doublet model, which incorporates flavor changing neutral scalar interactions (FCNSI) and CP violation (CPV) from several sources. CP violation can arise either from Yukawa terms or from the Hi
ggs potential, be it explicit or spontaneous. We show how the model, which is denoted as 2HDM-X, reduces to so
- Context: The triple stellar system delta Vel system presents a significant infrared excess, whose origin is still being debated. A large infrared bow shock has been discovered using Spitzer/MIPS observations. Although it appears as a significant co
ntributor to the measured IR excess, the possibility exists that a circumstellar IR excess is present around the stars of the system. - Aims: The objective of the present VISIR and NACO observations is to identify whether one of the stars of the delta Vel system presents a circumstellar photometric excess in the thermal IR domain and to quantify it. - Methods: We observed delta Vel using the imaging modes of the ESO/VLT instruments VISIR (in BURST mode) and NACO to resolve the A-B system (0.6 separation) and obtain the photometry of each star. We also obtained one NACO photometry epoch precisely at the primary (annular) eclipse of delta Vel Aa by Ab. - Results: Our photometric measurements with NACO (2.17 mic), complemented by the existing visible photometry allowed us to reconstruct the spectral energy distribution of the three stars. We then compared the VISIR photometry (8.6-12.8 mic) to the expected photospheric emission from the three stars at the corresponding wavelengths. - Conclusions: We can exclude the presence of a circumstellar thermal infrared excess around delta Vel A or B down to a few percent level. This supports the conclusions of Gaspar et al. (2008) that the IR excess of delta Vel has an interstellar origin, although a cold circumstellar disk could still be present. In addition, we derive the spectral types of the three stars Aa, Ab, and B (respectively A2IV, A4V and F8V), and we estimate the age of the system around 400-500 Myr.
We derive the one loop renormalization group equations for the Cabibbo-Kobayashi-Maskawa matrix for the Standard Model, its two Higgs extension and the minimal supersymmetric extension in a novel way. The derived equations depend only on a subset of
the model parameters of the renormalization group equations for the quark Yukawa couplings so the CKM matrix evolution cannot fully test the renormalization group evolution of the quark Yukawa couplings. From the derived equations we obtain the invariant of the renormalization group evolution for three models which is the angle $alpha$ of the unitarity triangle. For the special case of the Standard Model and its extensions with $v_{1}approx v_{2}$ we demonstrate that also the shape of the unitarity triangle and the Buras-Wolfenstein parameters $bar{rho}=(1-{1/2}lambda^{2})rho$ and $bar{eta}=(1-{1/2}lambda^{2})eta$ are conserved. The invariance of the angles of the unitarity triangle means that it is not possible to find a model in which the CKM matrix might have a simple, special form at asymptotic energies.
