The Supervised IBP: Neighbourhood Preserving Infinite Latent Feature Models

We propose a probabilistic model to infer supervised latent variables in the Hamming space from observed data. Our model allows simultaneous inference of the number of binary latent variables, and their values. The latent variables preserve neighbourhood structure of the data in a sense that objects in the same semantic concept have similar latent values, and objects in different concepts have dissimilar latent values. We formulate the supervised infinite latent variable problem based on an intuitive principle of pulling objects together if they are of the same type, and pushing them apart if they are not. We then combine this principle with a flexible Indian Buffet Process prior on the latent variables. We show that the inferred supervised latent variables can be directly used to perform a nearest neighbour search for the purpose of retrieval. We introduce a new application of dynamically extending hash codes, and show how to effectively couple the structure of the hash codes with continuously growing structure of the neighbourhood preserving infinite latent feature space.

Distribution of 25 semi-volatile organic compounds of two urban lakes in Wuhan, China

In this study, the levels of 25 semi-volatile organic compounds (SVOCs) were measured in samples of water, suspended particulate matter (SPM) and sediment from two urban lakes in Wuhan, China. The total concentrations of 25 SVOCs varied from 529.4 to 2168.9 ng/L, 120.7 to 22543.7 ng/g dry weight and 1577.3 to 61579.6 ng/g dry wt. in water, SPM and sediment, respectively. The concentration of SVOCs in SPM was 9-10 times higher than that in water, and the concentration of SVOCs in sediment 1.5-2 times higher than that in SPM. The level of total SVOC25 in the samples from Moshuihu Lake was higher than that in Yuehu Lake. Among the 25 SVOCs, phthalate compounds were on the highest level in all observed samples ranging between 441.9-1831.2 ng/L, 116.3-17566.8 ng/g, dry wt. and 6432.8-48177.6 ng/g dry wt. in water, SPM and sediment, respectively. Bis(2-ethylhexyl)phthalate, the predominant component of the analyzed pollutants, was in the range from 246.7 to 537.5 ng/l, 51.2 to 15540.0 ng/g dry wt. and 468.2 to 45010.3 ng/g dry wt. in water, SPM and sediment, respectively. The content of PAHs, dinitrotoluene and isophoton in sediment was higher than that in water and SPM at most of the locations. The possible sources of the pollutants and their inter-relation with human activities were discussed.

Semantic transform: Weakly supervised semantic inference for relating visual attributes

Relative (comparative) attributes are promising for thematic ranking of visual entities, which also aids in recognition tasks. However, attribute rank learning often requires a substantial amount of relational supervision, which is highly tedious, and apparently impractical for real-world applications. In this paper, we introduce the Semantic Transform, which under minimal supervision, adaptively finds a semantic feature space along with a class ordering that is related in the best possible way. Such a semantic space is found for every attribute category. To relate the classes under weak supervision, the class ordering needs to be refined according to a cost function in an iterative procedure. This problem is ideally NP-hard, and we thus propose a constrained search tree formulation for the same. Driven by the adaptive semantic feature space representation, our model achieves the best results to date for all of the tasks of relative, absolute and zero-shot classification on two popular datasets. © 2013 IEEE.

Magnetic coupling properties of mn-doped ZnO nanowires: First-principles calculations

Based on the density functional theory, we study the magnetic coupling properties of Mn-doped ZnO nanowires. For the nanowires with passivated surfaces, the antiferromagnetic state is found and the Mn atoms have a clustering tendency. When the distance between two Mn atoms is large, the system energetically favors the paramagnetic or spin-glass state. For the nanowires with unpassivated surfaces, the ferromagnetic (FM) coupling states appear between the two nearest Mn atoms, and the zinc vacancies can further stabilize the FM states between them. The electrons with enough concentration possibly mediate the FM coupling due to the negative exchange splitting of conduction band minimum induced by the s-d coupling, which could be useful in nanomaterial design for spintronics. (C) 2008 American Institute of Physics.

Integratable and High Speed Complex-Coupled MQW-DFB Lasers Fabricated on Semi-Insulating Substrates

A novel integratable and high speed InGaAsP multi-quantum well (MQW) complex-coupled distributed feedback (DFB) laser is successfully fabricated on a semi-insulating substrate. The fabricated ridge DFB laser exhibits a threshold current of 26 mA, a slope efficiency of 0.14 W.A(-1) and a side mode suppression ratio of 40 dB together with a 3 dB bandwidth of more than 8 GHz. The device is suitable for 10 Gbit/s optical fiber communication.

Influence of deep level defects on electrical compensation in semi-insulating InP materials

In this paper, we analyze and compare electrical compensation and deep level defects in semi-insulating ( SI) materials prepared by Fe-doping and high temperature annealing of undoped InP. Influence of deep level defects in the SI-InP materials on the electrical compensation has been studied thermally stimulated current spectroscopy (TSC). Electrical property of the Fe-doped SI-InP is deteriorated due to involvement of a high concentration of deep level defects in the compensation. In contrast, the concentration of deep defects is very low in high temperature annealed undoped SI-InP in which Fe acceptors formed by diffusion act as the only compensation centre to pin the Fermi level, resulting in excellent electrical performance. A more comprehensive electrical compensation model of SI-InP has been given based on the research results.

A comparison between contactless electroreflectance and photoreflectance spectra from n-doped GaAs on a semi-insulating GaAs substrate

Contactless electroreflectance (CER) and photoreflectance (PR) measurements have been performed on samples with the structure of an n-doped GaAs epitaxial layer on a semi- insulating GaAs substrate. Modulated reflectance signals from the n-GaAs surface and those from the n-GaAs/SI-GaAs interface are superposed in PR spectra. For the case of CER measurement, however, Franz-Keldysh oscillations (FKOs) from the interface, which are observed in PR spectra, cannot be detected. This discrepancy is attributed to different modulation mechanisms of CER and PR. In CER experiments, the electric field modulation cannot be added to the interfacial electric field because of the effective screening by the fast response of carriers across the interface. FKOs from the interface without any perturbation by the surface signals are extracted by subtracting CER spectra from PR spectra.

Preparation and AFM characterization of self-ordered porous alumina films on semi-insulated gaas substrate

Self-ordered porous alumina films on a semi-insulated GaAs substrate were prepared in oxalic acid aqueous solutions by three-step anodization. The I-t curve of anodization process was recorded to observe time effects of anodization. Atomic force microscopy was used to investigate structure and morphology of alumina films. It was revealed that the case of oxalic acid resulted in a self-ordered porous structure, with the pore diameters of 60-70 nm, the pore density of the order of about 10(10) pore cm(-2), and interpore distances of 95-100nm. At the same time the pore size and shape change with the pore widening time. Field-enhanced dissolution model and theory of deformation relaxation combined were brought forward to be the cause of self-ordered pore structure according to I-t curve of anodization and structure characteristics of porous alumina films. (c) 2006 Elsevier Ltd. All rights reserved.

Growth and characterization of semi-insulating GaN films grown by MOCVD

High resistivity unintentionally doped GaN films were grown on (0001) sapphire substrates by metalorganic chemical vapor deposition. The surface morphology of the layer was measured by both atomic force microscopy and scanning electron microscopy. The results show that the films have mirror-like surface morphology with root mean square of 0.3 nm. The full width at half maximum of double crystal X-ray diffraction rocking curve for (0002) GaN is about 5.22 arc-min, indicative of high crystal quality. The resistivity of the GaN epilayers at room temperature and at 250 degrees C was measured to be approximate 10(9) and 10(6) Omega(.)cm respectively, by variable temperature Hall measurement. Deep level traps in the GaN epilayers were investigated by thermally stimulated current and resistivity measurements.

Growth of high quality semi-insulating InP single crystal by suppression of compensation defects

Deep level defects in as-grown and annealed SI-InP samples were investigated by thermally stimulated current spectroscopy. Correlations between electrical property, compensation ratio, thermal stability and deep defect concentration in SI-InP were revealed. An optimized crystal growth condition for high quality SI-InP was demonstrated based on the experimental results.

Annealing and activation of silicon implanted in semi-insulating InP substrates

We have investigated the annealing and activation of silicon implanted in both as-grown Fe-doped semi-insulating (SI) InP substrate and undoped SI InP substrate obtained by annealing high purity conductive InP wafer (wafer-annealed). Si implantations were performed at an energy of 500 keV and a dose of 1 X 10(15) cm(-2). Following the implantations, rapid thermal annealing (RTA) cycles were carried out for 30 s at different temperatures. The results of Raman measurements show that for 700degreesC/30s RTA, the two Si-implanted SI InP substrates have acquired a high degree of lattice recovery and electrical activation. However, further Hall measurements indicate that the carrier concentration of the wafer-annealed SI InP substrate is about three times higher than that of the as-grown Fe-doped SI InP substrate. The difference can be ascribed to the low Fe concentration of the wafer-annealed SI InP substrate.These experimental data imply that the use of the wafer-annealed SI InP substrate can be conducive to the improvement of InP-based device performances. (C) 2003 Elsevier Ltd. All rights reserved.