This article provides a step by step development of designing a Object Detection scheme using the HOG based Feature Pyramid aligned with the concept of Template Matching.
https://arxiv.org/abs/1406.7120
The magnetic anisotropy of a planar array of Ga${x}$Fe${4-x}$N nanocrystals (NCs) embedded in a GaN host is studied by ferromagnetic resonance. X-ray diffraction and transmission electron microscopy are employed to determine the phase and distribution of the nanocrystals. The magnetic anisotropy is found to be primarily uniaxial with the hard axis normal to the NCs plane and to have a comparably weak in-plane hexagonal symmetry. The origin of the magnetic anisotropy is discussed taking into consideration the morphology of the nanocrystals, the epitaxial relations, strain effects and magnetic coupling between the NCs.
https://arxiv.org/abs/1405.0906
We show that in the generic situation where a biological network, e.g. a protein interaction network, is in fact a subnetwork embedded in a larger “bulk” network, the presence of the bulk causes not just extrinsic noise but also memory effects. This means that the dynamics of the subnetwork will depend not only on its present state, but also its past. We use projection techniques to get explicit expressions for the memory functions that encode such memory effects, for generic protein interaction networks involving binary and unary reactions such as complex formation and phosphorylation, respectively. Remarkably, in the limit of low intrinsic copy-number noise such expressions can be obtained even for nonlinear dependences on the past. We illustrate the method with examples from a protein interaction network around epidermal growth factor receptor (EGFR), which is relevant to cancer signalling. These examples demonstrate that inclusion of memory terms is not only important conceptually but also leads to substantially higher quantitative accuracy in the predicted subnetwork dynamics.
https://arxiv.org/abs/1402.0749
While low-level image features have proven to be effective representations for visual recognition tasks such as object recognition and scene classification, they are inadequate to capture complex semantic meaning required to solve high-level visual tasks such as multimedia event detection and recognition. Recognition or retrieval of events and activities can be improved if specific discriminative objects are detected in a video sequence. In this paper, we propose an image representation, called Detection Bank, based on the detection images from a large number of windowed object detectors where an image is represented by different statistics derived from these detections. This representation is extended to video by aggregating the key frame level image representations through mean and max pooling. We empirically show that it captures complementary information to state-of-the-art representations such as Spatial Pyramid Matching and Object Bank. These descriptors combined with our Detection Bank representation significantly outperforms any of the representations alone on TRECVID MED 2011 data.
https://arxiv.org/abs/1405.7102
Deep convolutional neural networks have recently proven extremely competitive in challenging image recognition tasks. This paper proposes the epitomic convolution as a new building block for deep neural networks. An epitomic convolution layer replaces a pair of consecutive convolution and max-pooling layers found in standard deep convolutional neural networks. The main version of the proposed model uses mini-epitomes in place of filters and computes responses invariant to small translations by epitomic search instead of max-pooling over image positions. The topographic version of the proposed model uses large epitomes to learn filter maps organized in translational topographies. We show that error back-propagation can successfully learn multiple epitomic layers in a supervised fashion. The effectiveness of the proposed method is assessed in image classification tasks on standard benchmarks. Our experiments on Imagenet indicate improved recognition performance compared to standard convolutional neural networks of similar architecture. Our models pre-trained on Imagenet perform excellently on Caltech-101. We also obtain competitive image classification results on the small-image MNIST and CIFAR-10 datasets.
https://arxiv.org/abs/1406.2732
Detecting objects becomes difficult when we need to deal with large shape deformation, occlusion and low resolution. We propose a novel approach to i) handle large deformations and partial occlusions in animals (as examples of highly deformable objects), ii) describe them in terms of body parts, and iii) detect them when their body parts are hard to detect (e.g., animals depicted at low resolution). We represent the holistic object and body parts separately and use a fully connected model to arrange templates for the holistic object and body parts. Our model automatically decouples the holistic object or body parts from the model when they are hard to detect. This enables us to represent a large number of holistic object and body part combinations to better deal with different “detectability” patterns caused by deformations, occlusion and/or low resolution. We apply our method to the six animal categories in the PASCAL VOC dataset and show that our method significantly improves state-of-the-art (by 4.1% AP) and provides a richer representation for objects. During training we use annotations for body parts (e.g., head, torso, etc), making use of a new dataset of fully annotated object parts for PASCAL VOC 2010, which provides a mask for each part.
https://arxiv.org/abs/1406.2031
We demonstrate THz intersubband absorption (15.6-26.1 meV) in m-plane AlGaN/GaN quantum wells. We find a trend of decreasing peak energy with increasing quantum well width, in agreement with theoretical expectations. However, a blue-shift of the transition energy of up to 14 meV was observed relative to the calculated values. This blue-shift is shown to decrease with decreasing charge density and is therefore attributed to many-body effects. Furthermore, a ~40% reduction in the linewidth (from roughly 8 to 5 meV) was obtained by reducing the total sheet density and inserting undoped AlGaN layers that separate the wavefunctions from the ionized impurities in the barriers.
https://arxiv.org/abs/1406.1772
The Robo-AO Kepler Planetary Candidate Survey is designed to observe every Kepler planet candidate host star with laser adaptive optics imaging to search for blended nearby stars, which may be physically associated companions and/or responsible for transit false positives. In this paper we present the results from the 2012 observing season, searching for stars close to 715 representative Kepler planet candidate hosts. We find 53 companions, 44 of which are new discoveries. We detail the Robo-AO survey data reduction methods including a method of using the large ensemble of target observations as mutual point-spread-function references, along with a new automated companion-detection algorithm designed for large adaptive optics surveys. Our survey is sensitive to objects from 0.15” to 2.5” separation, with contrast ratios up to delta-m~6. We measure an overall nearby-star-probability for Kepler planet candidates of 7.4% +/- 1.0%, and calculate the effects of each detected nearby star on the Kepler-measured planetary radius. We discuss several KOIs of particular interest, including KOI-191 and KOI-1151, which are both multi-planet systems with detected stellar companions whose unusual planetary system architecture might be best explained if they are “coincident multiple” systems, with several transiting planets shared between the two stars. Finally, we detect 2.6-sigma evidence for <15d-period giant planets being 2-3 times more likely be found in wide stellar binaries than smaller close-in planets and all sizes of further-out planets.
https://arxiv.org/abs/1312.4958
This paper presents an architecture of an information retrieval system that use the advantages offered by mobile agents to collect information from different sources and bring the result to the calling user. Mobile agent technology will be used for determine the traceability of a product and also for searching information about a specific entity.
https://arxiv.org/abs/1406.0296
In this paper, forward and backward propagating waves and reflectivity in an optical waveguide structure namely the fiber Bragg reflector also considered as a one dimensional photonic crystal, are analytically computed using coupled mode theory for different grating lengths and coupling conditions. AlxGa1-xN/GaN material composition is considered as unit block of the periodic organization, and refractive index of AlxGa1-xN is taken to be dependent on material composition, bandgap and operating wavelength following Adachis’ model. The structure being considered is the Bragg grating where increase in grating length enhances the reflection of electromagnetic wave, and strong coupling provides larger bandgap spectral width. Input wavelength is made different from Bragg wavelength to study the characteristics of propagating waves. A suitable combination of grating length and coupling coefficient is helpful in designing the photonic bandgap at 1550 nm wavelength. These characteristic curves can be utilized to study how waves propagate through the optical waveguides which have a special place in optical communications
https://arxiv.org/abs/1312.4442
In this paper, the forward and backward propagating modes in an optical waveguide structure namely the fiber Bragg filter also considered as a one dimensional photonic crystal, are analytically computed as a function of grating length for coupled optical modes. AlxGa1-xN/GaN material composition is considered as a unit block of the periodic organization, and refractive index of AlxGa1-xN/GaN is taken to be dependent on material composition, bandgap and operating wavelength following Adachi’s model. Expressions of propagating wave are derived using coupled mode theory. Simulated results help us to study the propagation of forward and backward wave propagating modes inside fiber and waveguide devices.
https://arxiv.org/abs/1312.4762
Adsorption of ammonia at NH3/NH2/H covered GaN(0001) surface was analyzed using results of ab initio calculations. The whole configuration space of partially NH3/NH2/H covered GaN(0001) surface was divided into zones differently pinned Fermi level: at Ga broken bond state for dominantly bare surface (region I), at VBM for NH2 and H covered (region II), and at CBM for NH3 covered surface (region III). The extensive ab intio calculations show validity of electron counting rule (ECR) for all mixed coverage, for bordering these three regions. The adsorption was analyzed using newly identified dependence of the adsorption energy on the charge transfer at the surface. For region I and II ammonia adsorb dissociatively, disintegrating into H adatom and HN2 radical for large fraction of vacant sites while for high coverage the ammonia adsorption is molecular. The dissociative adsorption energy strongly depends on the Fermi level at the surface (pinned) and in the bulk (unpinned) while the molecular adsorption energy is determined by bonding to surface only, in accordance to the recently published theory. The molecular adsorption is determined by the energy of covalent bonding to the surface. Ammonia adsorption in region III (Fermi level pinned at CBM) leads to unstable configuration both molecular and dissociative which is explained by the fact that Ga-broken bond sites are doubly occupied by electrons. The adsorbing ammonia brings 8 electrons to the surface, necessitating transfer of the electrons from Ga-broken bond state to Fermi level, energetically costly process. Adsorption of ammonia at H-covered site leads to creation of NH2 radical at the surface and escape of H2 molecule. The process energy is close to 0.12 eV, thus not large, but the inverse process is not possible due to escape of the hydrogen molecule.
https://arxiv.org/abs/1405.6309
We present a comprehensive computational study of the electronic, thermal, and thermoelectric (TE) properties of gallium nitride nanowires (NWs) over a wide range of thicknesses (3–9 nm), doping densities ($10^{18}$–$10^{20}$ cm$^{-3}$), and temperatures (300–1000 K). We calculate the low-field electron mobility based on ensemble Monte Carlo transport simulation coupled with a self-consistent solution of the Poisson and Schrödinger equations. We use the relaxation-time approximation and a Poisson-Schrodinger solver to calculate the electron Seebeck coefficient and thermal conductivity. Lattice thermal conductivity is calculated using a phonon ensemble Monte Carlo simulation, with a real-space rough surface described by a Gaussian autocorrelation function. Throughout the temperature range, the Seebeck coefficient increases while the lattice thermal conductivity decreases with decreasing wire cross section, both boding well for TE applications of thin GaN NWs. However, at room temperature these benefits are eventually overcome by the detrimental effect of surface roughness scattering on the electron mobility in very thin NWs. The highest room-temperature $ZT$ of 0.2 is achieved for 4-nm-thick NWs, while further downscaling degrades it. In contrast, at 1000 K, the electron mobility varies weakly with the NW thickness owing to the dominance of polar optical phonon scattering and multiple subbands contributing to transport, so $ZT$ increases with increasing confinement, reaching 0.8 for optimally doped 3-nm-thick NWs. The $ZT$ of GaN NWs increases with increasing temperature beyond 1000 K, which further emphasizes their suitability for high-temperature TE applications.
https://arxiv.org/abs/1405.4942
A critical component in the implementation of a concurrent tabling system is the design of the table space. One of the most successful proposals for representing tables is based on a two-level trie data structure, where one trie level stores the tabled subgoal calls and the other stores the computed answers. In this work, we present a simple and efficient lock-free design where both levels of the tries can be shared among threads in a concurrent environment. To implement lock-freedom we took advantage of the CAS atomic instruction that nowadays can be widely found on many common architectures. CAS reduces the granularity of the synchronization when threads access concurrent areas, but still suffers from low-level problems such as false sharing or cache memory side-effects. In order to be as effective as possible in the concurrent search and insert operations over the table space data structures, we based our design on a hash trie data structure in such a way that it minimizes potential low-level synchronization problems by dispersing as much as possible the concurrent areas. Experimental results in the Yap Prolog system show that our new lock-free hash trie design can effectively reduce the execution time and scale better than previous designs.
https://arxiv.org/abs/1405.2850
Interestingness,as the composition of Relevance and Unexpectedness, has been tested by means of Web search cases studies and led to promising results. But for thorough investigation and routine practical application one needs a flexible and robust tool. This work describes such an Interestingness based search tool, its software architecture and actual implementation. One of its flexibility traits is the choice of Interestingness functions: it may work with Match-Mismatch and Tf-Idf, among other functions. The tool has been experimentally verified by application to various domains of interest. It has been validated by comparison of results with those of commercial search engines and results from differing Interestingness functions.
https://arxiv.org/abs/1405.3557
A signature often found in non-minimal Higgs sectors is Higgs decay to a new gauge-singlet scalar, followed by decays of the singlets into Standard Model fermions through small mixing angles. The scalar decay can naturally be displaced from the primary vertex. The present experimental constraints on such models are extremely weak, due to low (or zero) trigger rates for the resulting low $p_T$ displaced jets. In this letter, we highlight the advantages of integrating into the trigger system massively parallel computing and coprocessors based on Graphics Processing Units (GPUs) or the Many Integrated Core (MIC) architecture. In particular, if such coprocessors are added to the LHC experiments’ high level trigger systems, a fast Hough transform based triggers performed on this hardware would result in significant improvement to displaced searches, sufficient to discover long-lived Higgs models with a small amount of luminosity in Run II at the 14 TeV LHC.
https://arxiv.org/abs/1405.2082
(In,Ga)N insertions embedded in self-assembled GaN nanowires are of current interest for applications in solid state light emitters. Such structures exhibit a notoriously broad emission band. We use cathodoluminescence spectral imaging in a scanning electron microscope and micro-photoluminescence spectroscopy on single nanowires to learn more about the mechanisms underlying this emission. We observe a shift of the emission energy along the stack of six insertions within single nanowires that may be explained by compositional pulling. Our results also corroborate reports that the localization of carriers at potential fluctuations within the insertions plays a crucial role for the luminescence of these nanowire based emitters. Furthermore, we resolve contributions from both structural and point defects in our measurements.
https://arxiv.org/abs/1405.1507
We report on the carrier dynamics in InGaN/GaN disk-in-a-wire quantum dots with precisely controlled location and structural parameters, including diameter, thickness and material composition. We measured the time-integrated and time-resolved spectra and the second-order correlation function of the photoluminescence from quantum dots with diameters ranging from 19 nm to 33 nm at temperatures of 10 K to 120 K. The influence of the small fluctuations in structural parameters, most importantly the quantum dot thickness, on the optical properties are also investigated through statistical correlations among multiple optical properties of many individual quantum dots. We found that in a single dot the strain-induced polarization field and the strain relaxation at the sidewall form a potential barrier to protect the exciton from reaching the sidewall surface. However, the exciton can overcome this potential barrier and recombine nonradiatively at the surface through two mechanisms: tunnelling through the barrier quantum mechanically and hopping over the barrier by attaining sufficient thermal energy. The former (latter) mechanism is temperature insensitive (sensitive) and dominates nonradiaitve exciton decay at low (high) temperatures. We also found that despite the good uniformities in structural parameters, all optical properties still exhibit inhomogeneities from dot to dot. However, all these inhomogeneities can be modeled by simply varying the potential barrier height, which also explains the observed correlation curves among all optical properties. Finally, we found that the biexciton-to-exciton quantum efficiency ratio, which determines the probability of multi-photon emission, can be tuned by adjusting the potential barrier height and the temperature, suggesting a new way to achieve single photon emission at high temperatures.
https://arxiv.org/abs/1309.4081
A new platform for fabricating polariton lasers operating at room temperature is introduced: nitride-based distributed Bragg reflectors epitaxially grown on patterned silicon substrates. The patterning allows for an enhanced strain relaxation thereby enabling to stack a large number of crack-free AlN/AlGaN pairs and achieve cavity quality factors of several thousands with a large spatial homogeneity. GaN and ZnO active regions are epitaxially grown thereon and the cavities are completed with top dielectric Bragg reflectors. The two structures display strong-coupling and polariton lasing at room temperature and constitute an intermediate step in the way towards integrated polariton devices.
https://arxiv.org/abs/1404.7743
We study here the problem of determining the majority type in an arbitrary connected network, each vertex of which has initially two possible types. The vertices may have a few additional possible states and can interact in pairs only if they share an edge. Any (population) protocol is required to stabilize in the initial majority. We first present and analyze a protocol with 4 states per vertex that always computes the initial majority value, under any fair scheduler. As we prove, this protocol is optimal, in the sense that there is no population protocol that always computes majority with fewer than 4 states per vertex. However this does not rule out the existence of a protocol with 3 states per vertex that is correct with high probability. To this end, we examine a very natural majority protocol with 3 states per vertex, introduced in [Angluin et al. 2008] where its performance has been analyzed for the clique graph. We study the performance of this protocol in arbitrary networks. We prove that, when the two initial states are put uniformly at random on the vertices, this protocol converges to the initial majority with probability higher than the probability of converging to the initial minority. In contrast, we present an infinite family of graphs, on which the protocol can fail whp, even when the difference between the initial majority and the initial minority is $n - \Theta(\ln{n})$. We also present another infinite family of graphs in which the protocol of Angluin et al. takes an expected exponential time to converge. These two negative results build upon a very positive result concerning the robustness of the protocol on the clique. Surprisingly, the resistance of the clique to failure causes the failure in general graphs. Our techniques use new domination and coupling arguments for suitably defined processes whose dynamics capture the antagonism between the states involved.
https://arxiv.org/abs/1404.7671
During the last decade, network approaches became a powerful tool to describe protein structure and dynamics. Here, we describe first the protein structure networks of molecular chaperones, then characterize chaperone containing sub-networks of interactomes called as chaperone-networks or chaperomes. We review the role of molecular chaperones in short-term adaptation of cellular networks in response to stress, and in long-term adaptation discussing their putative functions in the regulation of evolvability. We provide a general overview of possible network mechanisms of adaptation, learning and memory formation. We propose that changes of network rigidity play a key role in learning and memory formation processes. Flexible network topology provides “learning competent” state. Here, networks may have much less modular boundaries than locally rigid, highly modular networks, where the learnt information has already been consolidated in a memory formation process. Since modular boundaries are efficient filters of information, in the “learning competent” state information filtering may be much smaller, than after memory formation. This mechanism restricts high information transfer to the “learning competent” state. After memory formation, modular boundary-induced segregation and information filtering protect the stored information. The flexible networks of young organisms are generally in a “learning competent” state. On the contrary, locally rigid networks of old organisms have lost their “learning competent” state, but store and protect their learnt information efficiently. We anticipate that the above mechanism may operate at the level of both protein-protein interaction and neuronal networks.
https://arxiv.org/abs/1206.0094
First result of experiment for searching of 2K-capture of Xe-124 with the large-volume copper proportional counter is given. The 12 litre sample with 63.3% (44 g) of Xe-124 was used in measurements. The limit on the half-life of Xe-124 with regard to 2K(2\nu)-capture for the ground state of Te-124 has been found: T_{1/2} > 4.67x10^{20} y (90% C.L.). A sample with volume 52 L comprising of Xe-124 (10.6 L - 58.6 g) and Xe-126 (14.1 L - 79.3 g) will used at the next step of the experiment to increase a sensitivity of 2K-caption of Xe-124 registration. In this case sensitivity to the investigated process will be at the level of S=1.46 x 10^{21} y (90% C.L.) for 1 year measurement.
给出了用大容量铜正比计数器搜索Xe-124的2K捕获实验的初步结果。测量中使用具有63.3％（44g）Xe-124的12升样品。已经发现对于Te-124的基态的2K（2’） - 捕获的Xe-124的半衰期的限制已经被发现：T_ {1/2}> 4.67x10 ^ {20} y（90 ％CL）。在实验的下一个步骤中使用包含Xe-124（10.6L-58.6g）和Xe-126（14.1L-79.3g）的体积52L的样品以增加Xe-124的2K-标题的灵敏度注册。在这种情况下，对被调查过程的敏感度将在S = 1.46×10 ^ {21} y（90％C.L.）的水平下进行1年的测量。
https://arxiv.org/abs/1404.5530
When an object moves smoothly across a field of view, the identify of the object is unchanged, but the activation pattern of the photoreceptors on the retina changes drastically. One of the major computational roles of our visual system is to manage selectivity for different objects and tolerance to such identity-preserving transformations as translations or rotations. This study demonstrates that a hierarchical neural network, whose synaptic connectivities are learned competitively with Hebbian plasticity operating within a local spatiotemporal pooling range, is capable of gradually achieving feature selectivity and transformation tolerance, so that the top level neurons carry higher mutual information about object categories than a single-level neural network. Furthermore, when genetic algorithm is applied to search for a network architecture that maximizes transformation-invariant object recognition performance, in conjunction with the associative learning algorithm, it is found that deep networks outperform shallower ones.
https://arxiv.org/abs/1404.5373
Activity-driven modeling has been recently proposed as an alternative growth mechanism for time varying networks, displaying power-law degree distribution in time-aggregated representation. This approach assumes memoryless agents developing random connections, thus leading to random networks that fail to reproduce two-nodes degree correlations and the high clustering coefficient widely observed in real social networks. In this work we introduce these missing topological features by accounting for memory effects on the dynamic evolution of time-aggregated networks. To this end, we propose an activity-driven network growth model including a triadic-closure step as main connectivity mechanism. We show that this mechanism provides some of the fundamental topological features expected for social networks. We derive analytical results and perform extensive numerical simulations in regimes with and without population growth. Finally, we present two cases of study, one comprising face-to-face encounters in a closed gathering, while the other one from an online social friendship network.
https://arxiv.org/abs/1312.3496
The market trends towards the use of smaller dish antennas for TV satellite receivers, as well as the growing density of broadcasting satellites in orbit require the application of robust adjacent satellite interference (ASI) cancellation algorithms at the receivers. The wider beamwidth of a small size dish and the growing number of satellites in orbit impose an overloaded scenario, i.e., a scenario where the number of transmitting satellites exceeds the number of receiving antennas. For such a scenario, we present a two stage receiver to enhance signal detection from the satellite of interest, i.e., the satellite that the dish is pointing to, while reducing interference from neighboring satellites. Towards this objective, we propose an enhanced List-based Group-wise Search Detection (LGSD) receiver architecture that takes into account the spatially correlated additive noise and uses the signal-to-interference-plus noise ratio (SINR) maximization criterion to improve detection performance. Simulations show that the proposed receiver structure enhances the performance of satellite systems in the presence of ASI when compared to existing methods.
https://arxiv.org/abs/1404.4443
This paper addresses the challenge of establishing a bridge between deep convolutional neural networks and conventional object detection frameworks for accurate and efficient generic object detection. We introduce Dense Neural Patterns, short for DNPs, which are dense local features derived from discriminatively trained deep convolutional neural networks. DNPs can be easily plugged into conventional detection frameworks in the same way as other dense local features(like HOG or LBP). The effectiveness of the proposed approach is demonstrated with the Regionlets object detection framework. It achieved 46.1% mean average precision on the PASCAL VOC 2007 dataset, and 44.1% on the PASCAL VOC 2010 dataset, which dramatically improves the original Regionlets approach without DNPs.
https://arxiv.org/abs/1404.4316
We propose a new approach for metric learning by framing it as learning a sparse combination of locally discriminative metrics that are inexpensive to generate from the training data. This flexible framework allows us to naturally derive formulations for global, multi-task and local metric learning. The resulting algorithms have several advantages over existing methods in the literature: a much smaller number of parameters to be estimated and a principled way to generalize learned metrics to new testing data points. To analyze the approach theoretically, we derive a generalization bound that justifies the sparse combination. Empirically, we evaluate our algorithms on several datasets against state-of-the-art metric learning methods. The results are consistent with our theoretical findings and demonstrate the superiority of our approach in terms of classification performance and scalability.
http://arxiv.org/abs/1404.4105
Single GaN nanowires formed spontaneously on a given substrate represent nanoscopic single crystals free of any extended defects. However, due to the high area density of thus formed GaN nanowire ensembles, individual nanowires coalesce with others in their immediate vicinity. This coalescence process may introduce strain and structural defects, foiling the idea of defect-free material due to the nanowire geometry. To investigate the consequences of this process, a quantitative measure of the coalescence of nanowire ensembles is required. We derive objective criteria to determine the coalescence degree of GaN nanowire ensembles. These criteria are based on the area-perimeter relationship of the cross-sectional shapes observed, and in particular on their circularity. Employing these criteria, we distinguish single nanowires from coalesced aggregates in an ensemble, determine the diameter distribution of both, and finally analyze the coalescence degree of nanowire ensembles with increasing fill factor.
https://arxiv.org/abs/1402.5252
In this work, we present an off-axis holography study of GaN/AlN heterostructured nanowires grown by plasma-assisted molecular-beam epitaxy. We discuss the sample preparation of nanowire samples for electron holography and combine potential profiles obtained using holography with theoretical calculations of the projected potential in order to gain understanding of the potential distribution in these nanostructures. The effects of surface states are discussed
https://arxiv.org/abs/1404.2823
The photocurrent and the small-signal photoconductance of InGaN/GaN multiple-quantum-well structures were studied at the temperature range from 10 to 300 K. The optical excitation was carried out at the quantum wells intrinsic absorption wavelengths. Regardless of the temperature the experimental plots of direct photocurrent vs. reverse voltage were step-like, which is related to the sequential quantum wells passage from quasi-neutrality into the p-n-junction space charge region. In addition, under optical excitation near the quantum wells material absorption edge we observed the photocurrent declines with increasing reverse bias, i.e. the negative differential photoconductance. This phenomenon is associated with a blue shift of the InGaN quantum well absorption edge arising due to compensation of its build-in piezoelectric field by the p-n-junction electric field. Furthermore, it was experimentally shown that each quantum well corresponds to two peaks in the small-signal photoconductance vs. reverse voltage dependence. The temperature changes in the amplitude and position of these peaks indicate that they probably related to the charge carriers thermal emission and thermally activated tunneling from the quantum well.
https://arxiv.org/abs/1404.2391
The tracking algorithm performance depends on video content. This paper presents a new multi-object tracking approach which is able to cope with video content variations. First the object detection is improved using Kanade- Lucas-Tomasi (KLT) feature tracking. Second, for each mobile object, an appropriate tracker is selected among a KLT-based tracker and a discriminative appearance-based tracker. This selection is supported by an online tracking evaluation. The approach has been experimented on three public video datasets. The experimental results show a better performance of the proposed approach compared to recent state of the art trackers.
https://arxiv.org/abs/1404.2005
Human face as a physical human recognition can be used as a unique identity for computer to recognize human by transforming human face with face algorithm as simple text number which can be primary key for human. Human face as single identity for human will be done by making a huge and large world centre human face database, where the human face around the world will be recorded from time to time and from generation to generation. Architecture database will be divided become human face image database which will save human face images and human face output code which will save human face output code as a transformation human face image with face algorithm. As an improvement the slightly and simple human face output code database will make human face searching process become more fast. Transaction with human face as a transaction without card can make human no need their card for the transaction and office automation and banking system as an example for implementation architecture. As an addition suspect human face database can be extended for fighting crime and terrorism by doing surveillance and searching suspect human face around the world.
https://arxiv.org/abs/1405.6168
The top-performing systems for billion-scale high-dimensional approximate nearest neighbor (ANN) search are all based on two-layer architectures that include an indexing structure and a compressed datapoints layer. An indexing structure is crucial as it allows to avoid exhaustive search, while the lossy data compression is needed to fit the dataset into RAM. Several of the most successful systems use product quantization (PQ) for both the indexing and the dataset compression layers. These systems are however limited in the way they exploit the interaction of product quantization processes that happen at different stages of these systems. Here we introduce and evaluate two approximate nearest neighbor search systems that both exploit the synergy of product quantization processes in a more efficient way. The first system, called Fast Bilayer Product Quantization (FBPQ), speeds up the runtime of the baseline system (Multi-D-ADC) by several times, while achieving the same accuracy. The second system, Hierarchical Bilayer Product Quantization (HBPQ) provides a significantly better recall for the same runtime at a cost of small memory footprint increase. For the BIGANN dataset of billion SIFT descriptors, the 10% increase in Recall@1 and the 17% increase in Recall@10 is observed.
https://arxiv.org/abs/1404.1831
We study strain relaxation and surface damage of GaN nanopillar arrays fabricated using inductively coupled plasma (ICP) etching and post etch wet chemical treatment. We controlled the shape and surface damage of such nanopillar structures through selection of etching parameters. We compared different substrate temperatures and different chlorine-based etch chemistries to fabricate high quality GaN nanopillars. Room temperature photoluminescence and Raman scattering measurements were carried to study the presence of surface defect and strain relaxation on these nanostructures, respectively. We found that wet KOH etching can remove the side wall damages caused by dry plasma etching, leading to better quality of GaN nanopillars arrays. The Si material underneath the GaN pillars was removed by KOH wet etching, leaving behind a fine Si pillar to support the GaN structure. Substantial strain relaxations were observed in these structures from room temperature Raman spectroscopy measurements. Room temperature Photoluminescence spectroscopy shows the presence of whispering gallery modes from these the nano disks structures.
https://arxiv.org/abs/1311.0321
Under endoscopic assumptions about $L$-packets of unitary groups, we prove the local Gan-Gross-Prasad conjecture for tempered representations of unitary groups over $p$-adic fields. Roughly, this conjecture says that branching laws for $U(n-1)\subset U(n)$ can be computed using epsilon factors.
https://arxiv.org/abs/1212.0951
Huge amount of data in the form of strings are being handled in bio-computing applications and searching algorithms are quite frequently used in them. Many methods utilizing on both software and hardware are being proposed to accelerate processing of such data. The typical hardware-based acceleration techniques either require special hardware such as general purpose graphics processing units (GPGPUs) or need building a new hardware such as an FPGA based design. On the other hard, software-based acceleration techniques are easier since they only require some changes in the software code or the software architecture. Typical software-based techniques make use of computers connected over a network, also known as a network grid to accelerate the processing. In this paper, we test the hypothesis that multi-core architectures should provide better performance in this kind of computation, but still it would depend on the algorithm selected as well as the programming model being utilized. We present the acceleration of a string-searching algorithm on a multi-core CPU via a POSIX thread based implementation. Our implementation on an 8-core processor (that supports 16-threads) resulted in 9x throughput improvement compared to a single thread implementation.
https://arxiv.org/abs/1403.7294
InGaN/GaN tunnel junction contacts were grown on top of an InGaN/GaN blue (450 nm) light emitting diode wafer using plasma assisted molecular beam epitaxy. The tunnel junction contacts enable low spreading resistance n-GaN top contact layer thereby requiring less top metal contact coverage on the surface. A voltage drop of 5.3 V at 100 mA, forward resistance of 2 x 10-2 ohm cm2 and a higher light output power are measured in tunnel junction LED. A low resistance of 5 x 10-4 ohm cm2 was measured in a MBE grown tunnel junction on GaN PN junction device, indicating that the tunnel junction LED device resistance is limited by the regrowth interface and not by the intrinsic tunneling resistance.
https://arxiv.org/abs/1403.3932
We perform correlated studies of individual GaN nanowires in scanning electron microscopy combined to low temperature cathodoluminescence, microphotoluminescence, and scanning transmission electron microscopy. We show that some nanowires exhibit well localized regions emitting light at the energy of a stacking fault bound exciton (3.42 eV) and are able to observe the presence of a single stacking fault in these regions. Precise measurements of the cathodoluminescence signal in the vicinity of the stacking fault give access to the exciton diffusion length near this location.
https://arxiv.org/abs/1403.3886
Interbank markets are fundamental for bank liquidity management. In this paper, we introduce a model of interbank trading with memory. Our model reproduces features of preferential trading patterns in the e-MID market recently empirically observed through the method of statistically validated networks. The memory mechanism is used to introduce a proxy of trust in the model. The key idea is that a lender, having lent many times to a borrower in the past, is more likely to lend to that borrower again in the future than to other borrowers, with which the lender has never (or has in- frequently) interacted. The core of the model depends on only one parameter representing the initial attractiveness of all the banks as borrowers. Model outcomes and real data are compared through a variety of measures that describe the structure and properties of trading networks, including number of statistically validated links, bidirectional links, and 3-motifs. Refinements of the pairing method are also proposed, in order to capture finite memory and reciprocity in the model. The model is implemented within the Mason framework in Java.
https://arxiv.org/abs/1403.3638
This paper presents results on the memory capacity of a generalized feedback neural network using a circulant matrix. Children are capable of learning soon after birth which indicates that the neural networks of the brain have prior learnt capacity that is a consequence of the regular structures in the brain’s organization. Motivated by this idea, we consider the capacity of circulant matrices as weight matrices in a feedback network.
https://arxiv.org/abs/1403.3115
We demonstrate a new method to measure absorption coefficients in any family of nanowires, provided they are grown on a substrate having considerable difference in permittivity with the nanowire-air matrix. In the case of high crystal quality, strain-free GaN nanowires, grown on Si (111) substrates with a density of ~1010 cm-2, the extracted absorption coefficients do not exhibit any enhancement compared to bulk GaN values, unlike relevant claims in the literature. This may be attributed to the relatively small diameters, short heights, and high densities of our nanowire arrays.
https://arxiv.org/abs/1402.7318
The WorldWideWeb (WWW) is a huge conservatory of web pages. Search Engines are key applications that fetch web pages for the user query. In the current generation web architecture, search engines treat keywords provided by the user as isolated keywords without considering the context of the user query. This results in a lot of unrelated pages or links being displayed to the user. Semantic Web is based on the current web with a revised framework to display a more precise result set as response to a user query. The current web pages need to be annotated by finding relevant meta data to be added to each of them, so that they become useful to Semantic Web search engines. Semantic Look explores the context of user query by processing the Semantic information recorded in the web pages. It is compared with an existing algorithm called OntoLook and it is shown that Semantic Look is a better optimized search engine by being more than twice as fast as OntoLook.
https://arxiv.org/abs/1402.7200
Broadband visible light emitting, three-dimensional hexagonal annular microstructures with InGaN/GaN multiple quantum wells (MQWs) are fabricated via selective-area epitaxial growth. The single hexagonal annular structure is composed of not only polar facet of (0001) on top surface but also semi-polar facets of {10-11} and {11-22} in inner and outer sidewalls, exhibiting multi-color visible light emission from InGaN/GaN MQWs formed on the different facets. The InGaN MQWs on (0001) facet emits the longer wavelength (green color) due to the larger well thickness and the higher In composition, while those on semi-polar facets of {10-11} and {11-22} had highly-efficient shorter wavelength (violet to blue color) emission caused by smaller well thickness and smaller In composition. By combining the multiple color emission depending on different facets, high efficiency broadband visible light emission could be achieved. The emission color can be changed with excitation power density owing to the built-in electric field on the (0001) facet, which is confirmed by time-resolved luminescence experiments. The hexagonal annular structures can be a critical building block for highly efficient broadband visible light emitting sources, providing a solution to previous problems related to the fabrication issues for phosphor-free white light emitting devices.
https://arxiv.org/abs/1402.5917
We investigate the computation of mappings from a set S^n to itself with “in situ programs”, that is using no extra variables than the input, and performing modifications of one component at a time, hence using no memory. In this paper, we survey this problem introduced in previous papers by the authors, we detail its close relation with rearrangeable multicast networks, and we provide new results for both viewpoints. A bijective mapping can be computed by 2n - 1 component modifications, that is by a program of length 2n - 1, a result equivalent to the rearrangeability of the concatenation of two reversed butterfly networks. For a general arbitrary mapping, we give two methods to build a program with maximal length 4n-3. Equivalently, this yields rearrangeable multicast routing methods for the network formed by four successive butterflies with alternating reversions. The first method is available for any set S and practically equivalent to a known method in network theory. The second method, a refinement of the first, described when |S| is a power of 2, is new and allows more flexibility than the known method. For a linear mapping, when S is any field, or a quotient of an Euclidean domain (e.g. Z/sZ for any integer s), we build a program with maximal length 2n - 1. In this case the assignments are also linear, thereby particularly efficient from the algorithmic viewpoint, and giving moreover directly a program for the inverse when it exists. This yields also a new result on matrix decompositions, and a new result on the multicast properties of two successive reversed butterflies. Results of this flavour were known only for the boolean field Z/2Z.
https://arxiv.org/abs/1310.5380
We present maps of a large number of dense molecular gas tracers across the Central Molecular Zone of our Galaxy. The data were taken with the CSIRO/CASS Mopra telescope in Large Projects in the 1.3cm, 7mm, and 3mm wavelength regimes. Here, we focus on the brightness of the shock tracers SiO and HNCO, molecules that are liberated from dust grains under strong (SiO) and weak (HNCO) shocks. The shocks may have occurred when the gas enters the bar regions and the shock differences could be due to differences in the moving cloud mass. Based on tracers of ionizing photons, it is unlikely that the morphological differences are due to selective photo-dissociation of the molecules. We also observe direct heating of molecular gas in strongly shocked zones, with a high SiO/HNCO ratios, where temperatures are determined from the transitions of ammonia. Strong shocks appear to be the most efficient heating source of molecular gas, apart from high energy emission emitted by the central supermassive black hole Sgr A* and the processes within the extreme star formation region Sgr B2.
https://arxiv.org/abs/1402.5066
We present a large-scale, interferometric survey of ammonia (1,1) and (2,2) toward the Galactic Center observed with the Australia Telescope Compact Array (ATCA). The survey covers Delta l ~1degree (~150pc) at an assumed distance of 8.5 kpc) and Delta b ~0.2degree (~30pc) which spans the region between the supermassive black hole SgrA* and the massive star forming region SgrB2. The resolution is ~20’’ (~0.8pc) and emission at scales >~2’ (>~3.2pc) is filtered out due to missing interferometric short spacings. Consequently, the data represent the denser, compact clouds and disregards the large scale, diffuse gas. Many of the clumps align with the 100 pc dust ring and mostly anti-correlate with 1.2cm continuum emission. We present a kinetic temperature map of the dense gas. The temperature distribution peaks at ~38K with a width at half maximum between 18K and 61K (measurements sensitive within Tkin~10-80K). Larger clumps are on average warmer than smaller clumps which suggests internal heating sources. Our observations indicate that the circumnuclear disk ~1.5 pc around SgrA* is supplied with gas by the 20km/s molecular cloud. This gas is substantially cooler than gas ~3-15pc away from SgrA*. We find a strong temperature gradient across SgrB2. Ammonia column densities correlate well with SCUBA 850um fluxes, but the relation is shifted from the origin, which may indicate a requirement for a minimum amount of dust to form and shield ammonia. Around the Arches and Quintuplet clusters we find shell morphologies with UV-influenced gas in their centers, followed by ammonia and radio continuum layers.
https://arxiv.org/abs/1402.4531
We report on electrostatic screening of polarization-induced internal electric fields in AlN/GaN nanowire heterostructures with Germanium-doped GaN nanodiscs embedded between AlN barriers. The incorporation of Germanium at concentrations above $10^{20}\,\text{cm}^{-3}$ shifts the photoluminescence emission energy of GaN nanodiscs to higher energies accompanied by a decrease of the photoluminescence decay time. At the same time, the thickness-dependent shift in emission energy is significantly reduced. In spite of the high donor concentration a degradation of the photoluminescence properties is not observed.
https://arxiv.org/abs/1402.3081
The need for appropriate ways to measure the distance or similarity between data is ubiquitous in machine learning, pattern recognition and data mining, but handcrafting such good metrics for specific problems is generally difficult. This has led to the emergence of metric learning, which aims at automatically learning a metric from data and has attracted a lot of interest in machine learning and related fields for the past ten years. This survey paper proposes a systematic review of the metric learning literature, highlighting the pros and cons of each approach. We pay particular attention to Mahalanobis distance metric learning, a well-studied and successful framework, but additionally present a wide range of methods that have recently emerged as powerful alternatives, including nonlinear metric learning, similarity learning and local metric learning. Recent trends and extensions, such as semi-supervised metric learning, metric learning for histogram data and the derivation of generalization guarantees, are also covered. Finally, this survey addresses metric learning for structured data, in particular edit distance learning, and attempts to give an overview of the remaining challenges in metric learning for the years to come.
http://arxiv.org/abs/1306.6709
Long Short-Term Memory (LSTM) is a recurrent neural network (RNN) architecture that has been designed to address the vanishing and exploding gradient problems of conventional RNNs. Unlike feedforward neural networks, RNNs have cyclic connections making them powerful for modeling sequences. They have been successfully used for sequence labeling and sequence prediction tasks, such as handwriting recognition, language modeling, phonetic labeling of acoustic frames. However, in contrast to the deep neural networks, the use of RNNs in speech recognition has been limited to phone recognition in small scale tasks. In this paper, we present novel LSTM based RNN architectures which make more effective use of model parameters to train acoustic models for large vocabulary speech recognition. We train and compare LSTM, RNN and DNN models at various numbers of parameters and configurations. We show that LSTM models converge quickly and give state of the art speech recognition performance for relatively small sized models.
https://arxiv.org/abs/1402.1128
Edges characterize boundaries and are therefore a problem of practical importance in remote this http URL this paper a comparative study of various edge detection techniques and band wise analysis of these algorithms in the context of object extraction with regard to remote sensing satellite images from the Indian Remote Sensing Satellite (IRS) sensors LISS 3, LISS 4 and Cartosat1 as well as Google Earth is presented.
https://arxiv.org/abs/1405.6132