Variable Sparsity Kernel Learning

This paper(1) presents novel algorithms and applications for a particular class of mixed-norm regularization based Multiple Kernel Learning (MKL) formulations. The formulations assume that the given kernels are grouped and employ l(1) norm regularization for promoting sparsity within RKHS norms of each group and l(s), s >= 2 norm regularization for promoting non-sparse combinations across groups. Various sparsity levels in combining the kernels can be achieved by varying the grouping of kernels-hence we name the formulations as Variable Sparsity Kernel Learning (VSKL) formulations. While previous attempts have a non-convex formulation, here we present a convex formulation which admits efficient Mirror-Descent (MD) based solving techniques. The proposed MD based algorithm optimizes over product of simplices and has a computational complexity of O (m(2)n(tot) log n(max)/epsilon(2)) where m is no. training data points, n(max), n(tot) are the maximum no. kernels in any group, total no. kernels respectively and epsilon is the error in approximating the objective. A detailed proof of convergence of the algorithm is also presented. Experimental results show that the VSKL formulations are well-suited for multi-modal learning tasks like object categorization. Results also show that the MD based algorithm outperforms state-of-the-art MKL solvers in terms of computational efficiency.

Non-Fermi-liquid theory for disordered metals near two dimensions

We consider the Finkelstein action describing a system of spin-polarized or spinless electrons in 2+2epsilon dimensions, in the presence of disorder as well as the Coulomb interactions. We extend the renormalization-group analysis of our previous work and evaluate the metal-insulator transition of the electron gas to second order in an epsilon expansion. We obtain the complete scaling behavior of physical observables like the conductivity and the specific heat with varying frequency, temperature, and/or electron density. We extend the results for the interacting electron gas in 2+2epsilon dimensions to include the quantum critical behavior of the plateau transitions in the quantum Hall regime. Although these transitions have a very different microscopic origin and are controlled by a topological term in the action (theta term), the quantum critical behavior is in many ways the same in both cases. We show that the two independent critical exponents of the quantum Hall plateau transitions, previously denoted as nu and p, control not only the scaling behavior of the conductances sigma(xx) and sigma(xy) at finite temperatures T, but also the non-Fermi-liquid behavior of the specific heat (c(v)proportional toT(p)). To extract the numerical values of nu and p it is necessary to extend the experiments on transport to include the specific heat of the electron gas.

Application of a Class of Nano fluids to improve the Loadability of Power Transformers

It is generally known that addition of conducting or insulating particles to mineral transformer oil, lowers its breakdown strength, E-d. However, if the particulates are of molecular dimensions, or nanoparticles, (NPs), as they are called, the breakdown strength is seen to increase considerably. Recent experiments by the authors on oil cooled power equipment such as transformers showed that, nanofluids comprising NPs of selected oxides of iron, such as Fe(3)o(4), called magnetite, added to transformer oil increased the breakdown voltage of the virgin oil and more importantly a remarkable enhancement in the thermal conductivity and the viscosity and hence an increased loadability of the transformer for a given top oil temperature (TOT).

Photoluminescence, photocatalysis and Judd-Ofelt analysis of Eu3+-activated layered BiOCl phosphors

Eu3+-activated layered BiOCl phosphors were synthesized by the conventional solid-state method at relatively low temperature and shorter duration (400 degrees C for 1 h). All the samples were crystallized in the tetragonal structure with the space group P4/nmm (no. 129). Field emission scanning electron microscopy (FE-SEM) studies confirmed the plate-like morphology. Photoluminescence spectra exhibit characteristic luminescent D-5(0) -> F-7(J) (J = 0-4) intra-4f shell Eu3+ ion transitions. The electric dipole transition located at 620 nm (D-5(0) -> F-7(2)) was stronger than the magnetic dipole transition located at 594 nm (D-5(0) -> F-7(1)). The evaluated Commission International de l'Eclairage (CIE) color coordinates of Eu3+-activated BiOCl phosphors were close to the commercial Y2O3:Eu3+ and Y2O2S:Eu3+ red phosphors. Intensity parameters (Omega(2), Omega(4)) and various radiative properties such as transition probability (A(tot)), radiative lifetime (tau(rad)), stimulated emission cross-section (sigma(e)), gain bandwidth (sigma(e) x Delta lambda(eff)) and optical gain (sigma(e) x tau(rad)) were calculated using the Judd-Ofelt theory. The experimental decay curves of the D-5(0) level in Eu3+-activated BiOCl have a single exponential profile. In comparison with other Eu3+ doped materials, Eu3+-activated BiOCl phosphors have a long lifetime (tau(exp)), low non-radiative relaxation rate (W-NR), high quantum efficiency (eta) and better optical gain (sigma(e) x tau(rad)). The determined radiative properties revealed the usefulness of Eu3+-activated BiOCl in developing red lasers as well as optical display devices. Further, these samples showed efficient photocatalytic activity for the degradation of rhodamine B (RhB) dye under visible light irradiation. These photocatalysts are useful for the removal of toxic and non-biodegradable organic pollutants in water.