arXiv subjects

Wei Tang

Publications and source records attributed to Wei Tang.

193 records · Page 11Linked to original sources

arXiv2018

A Compositional Textual Model for Recognition of Imperfect Word Images

Printed text recognition is an important problem for industrial OCR systems. Printed text is constructed in a standard procedural fashion in most settings. We develop a mathematical model for this process that can be applied to the backward inference problem of text recognition from an image. Through ablation experiments we show that this model is realistic and that a multi-task objective setting can help to stabilize estimation of its free parameters, enabling use of conventional deep learning methods. Furthermore, by directly modeling the geometric perturbations of text synthesis we show that our model can help recover missing characters from incomplete text regions, the bane of multicomponent OCR systems, enabling recognition even when the detection returns incomplete information.

cs.CV↗

arXiv2018

On spinodal decomposition in alnico---a transmission electron microscopy and atom probe tomography study

Alnico is a prime example of a finely tuned nanostructure whose magnetic properties are intimately connected to magnetic annealing (MA) during spinodal transformation and subsequent lower temperature annealing (draw) cycles. Using a combination of transmission electron microscopy and atom probe tomography, we show how these critical processing steps affect the local composition and nanostructure evolution with impact on magnetic properties. The nearly 2-fold increase of intrinsic coercivity ($H_\text{ci}$) during the draw cycle is not adequately explained by chemical refinement of the spinodal phases. Instead, increased Fe-Co phase ($α_1$) isolation, development of Cu-rich spheres/rods/blades and additional $α_1$ rod precipitation that occurs during the MA and draw, likely play a key role in $H_\text{ci}$ enhancement. Chemical ordering of the Al-Ni-phase ($α_2$) and formation of Ni-rich ($α_3$) may also contribute. Unraveling of the subtle effect of these nano-scaled features is crucial to understanding on how to improve shape anisotropy in alnico magnets.

cond-mat.mtrl-sci↗

arXiv2017

ReabsNet: Detecting and Revising Adversarial Examples

Though deep neural network has hit a huge success in recent studies and applica- tions, it still remains vulnerable to adversarial perturbations which are imperceptible to humans. To address this problem, we propose a novel network called ReabsNet to achieve high classification accuracy in the face of various attacks. The approach is to augment an existing classification network with a guardian network to detect if a sample is natural or has been adversarially perturbed. Critically, instead of simply rejecting adversarial examples, we revise them to get their true labels. We exploit the observation that a sample containing adversarial perturbations has a possibility of returning to its true class after revision. We demonstrate that our ReabsNet outperforms the state-of-the-art defense method under various adversarial attacks.

cs.LG↗

arXiv2017

Study on a Poisson's Equation Solver Based On Deep Learning Technique

In this work, we investigated the feasibility of applying deep learning techniques to solve Poisson's equation. A deep convolutional neural network is set up to predict the distribution of electric potential in 2D or 3D cases. With proper training data generated from a finite difference solver, the strong approximation capability of the deep convolutional neural network allows it to make correct prediction given information of the source and distribution of permittivity. With applications of L2 regularization, numerical experiments show that the predication error of 2D cases can reach below 1.5\% and the predication of 3D cases can reach below 3\%, with a significant reduction in CPU time compared with the traditional solver based on finite difference methods.

physics.comp-ph↗

arXiv2017

Search for Massive Bosons Decaying to Wg and Zg Using the ATLAS Detector

The Standard Model is by far the most encompassing physics theory. With the recent discovery of the Higgs Boson, the Standard Model has performed extremely well against experimental data. However, the theory is intrinsically not complete, it does not address physical phenomenon such as gravity and dark matter. Many proposals for physics beyond the Standard Model predict new massive bosons from additional gauge fields or extended Higgs sectors. This paper examines a preliminary search of possible V gamma decay channels of proposed new bosons.

hep-ph↗

arXiv2017

Universal Boundary Entropies in Conformal Field Theory: A Quantum Monte Carlo Study

Recently, entropy corrections on nonorientable manifolds such as the Klein bottle are proposed as a universal characterization of critical systems with an emergent conformal field theory (CFT). We show that entropy correction on the Klein bottle can be interpreted as a boundary effect via transforming the Klein bottle into an orientable manifold with nonlocal boundary interactions. The interpretation reveals the conceptual connection of the Klein bottle entropy with the celebrated Affleck-Ludwig entropy in boundary CFT. We propose a generic scheme to extract these universal boundary entropies from quantum Monte Carlo calculation of partition function ratios in lattice models. Our numerical results on the Affleck-Ludwig entropy and Klein bottle entropy for the $q$-state quantum Potts chains with $q=2,3$ show excellent agreement with the CFT predictions. For the quantum Potts chain with $q=4$, the Klein bottle entropy slightly deviates from the CFT prediction, which is possibly due to marginally irrelevant terms in the low-energy effective theory.

cond-mat.str-el↗

arXiv2017

Microstructural and magnetic property evolution with different heat-treatment conditions in an alnico alloy

Further property enhancement of alnico, an attractive near-term, non-rare-earth permanent magnet alloy system, primarily composed of Al, Ni, Co, and Fe, relies on improved morphology control and size refinement of its complex spinodally decomposed nanostructure that forms during heat-treatment. Using a combination of transmission electron microscopy and atom probe tomography techniques, this study evaluates the magnetic properties and microstructures of an isotropic 32.4Fe-38.1Co-12.9Ni-7.3Al-6.4Ti-3.0Cu (wt.$\%$) alloy in terms of processing parameters such as annealing temperature, annealing time, application of an external magnetic field, as well as low-temperature "draw" annealing. Optimal spinodal morphology and spacing is formed within a narrow temperature and time range ($\sim 840 \unicode{x2103}$ and 10 min during thermal-magnetic annealing (MA). The ideal morphology is a mosaic structure consisting of periodically arrayed $\sim 40$ nm diameter (Fe-Co)-rich rods ($α_1$ phase) embedded in an (Al-Ni)-rich ($α_2$ phase) matrix. A Cu-enriched phase with a size of $\sim$ 3-5 nm is located at the corners of two adjacent $\{110\}$ facets of the $α_1$ phase. The MA process significantly increased remanence ($B_\text{r}$) ($\sim$ 40-70 $\%$) of the alloy due to biased elongation of the $α_1$ phase along the $\langle100\rangle$ crystallographic direction, which is closest in orientation to the applied magnetic field. The optimum magnetic properties of the alloy with an intrinsic coercivity ($H_\text{cj}$) of 1845 Oe and a maximum energy product ($BH_\text{max}$) of 5.9 MGOe were attributed to the uniformity of the mosaic structure.

cond-mat.mtrl-sci↗

arXiv2017

Simulation of alnico coercivity

Micromagnetic simulations of alnico show substantial deviations from Stoner-Wohlfarth behavior due to the unique size and spatial distribution of the rod-like Fe-Co phase formed during spinodal decomposition in an external magnetic field. The maximum coercivity is limited by single-rod effects, especially deviations from ellipsoidal shape, and by interactions between the rods. Both the exchange interaction between connected rods and magnetostatic interaction between rods are considered, and the results of our calculations show good agreement with recent experiments. Unlike systems dominated by magnetocrystalline anisotropy, coercivity in alnico is highly dependent on size, shape, and geometric distribution of the Fe-Co phase, all factors that can be tuned with appropriate chemistry and thermal-magnetic annealing.

cond-mat.mtrl-sci↗

arXiv2016

A 20-Liter Test Stand with Gas Purification for Liquid Argon Research

We describe the design of a 20-liter test stand constructed to study fundamental properties of liquid argon (LAr). This system utilizes a simple, cost-effective gas argon (GAr) purification to achieve high purity, which is necessary to study electron transport properties in LAr. An electron drift stack with up to 25 cm length is constructed to study electron drift, diffusion, and attachment at various electric fields. A gold photocathode and a pulsed laser are used as a bright electron source. The operational performance of this system is reported.

physics.ins-det↗

arXiv2016

Measurement of Longitudinal Electron Diffusion in Liquid Argon

We report the measurement of longitudinal electron diffusion coefficients in liquid argon for electric fields between 100 and 2000 V/cm with a gold photocathode as a bright electron source. The measurement principle, apparatus, and data analysis are described. Our results, which are consistent with previous measurements in the region between 100 to 350 V/cm [1] , are systematically higher than the prediction of Atrazhev-Timoshkin[2], and represent the world's best measurement in the region between 350 to 2000 V/cm. The quantum efficiency of the gold photocathode, the drift velocity and longitudinal diffusion coefficients in gas argon are also presented.

physics.ins-det↗

arXiv2015

Recent Results from the Daya Bay Neutrino Experiment

The Daya Bay neutrino experiment has recently updated the oscillation analysis results with 621 days of data in 2015, which has 3.6 times more statistics than the previous publication in 2014. The relative $\barν_{e}$ rate and spectrum measurement between the near and far detectors yielded the best fit values of $\sin^{2}2θ_{13}$ = 0.084 $\pm$ 0.005 and $|Δm_{ee}^{2}|$ = (2.42 $\pm$ 0.11) $\times$ 10$^{-3}$ eV$^{2}$. This is currently the most precise measurement of $\sin^{2}2θ_{13}$ in the world. The measurement of $|Δm_{ee}^{2}|$ also has a precision that is comparable to the measurements from MINOS and T2K experiments in 2014. Daya Bay also performed several other analyses such as the search for the light sterile neutrino in the 3+1 neutrino framework, and the measurements of the absolute reactor anti-neutrino flux and spectrum.

hep-ex↗

arXiv2014

Applications of Compressed Sensing in Communications Networks

This paper presents a tutorial for CS applications in communications networks. The Shannon's sampling theorem states that to recover a signal, the sampling rate must be as least the Nyquist rate. Compressed sensing (CS) is based on the surprising fact that to recover a signal that is sparse in certain representations, one can sample at the rate far below the Nyquist rate. Since its inception in 2006, CS attracted much interest in the research community and found wide-ranging applications from astronomy, biology, communications, image and video processing, medicine, to radar. CS also found successful applications in communications networks. CS was applied in the detection and estimation of wireless signals, source coding, multi-access channels, data collection in sensor networks, and network monitoring, etc. In many cases, CS was shown to bring performance gains on the order of 10X. We believe this is just the beginning of CS applications in communications networks, and the future will see even more fruitful applications of CS in our field.

cs.NI↗

arXiv2011

Structural and Magnetic Properties of Sm(Co0.7Fe0.1Ni0.12Zr0.04B0.04)7.5 Melt Spun Isotropic and Anisotropic Ribbons

We have investigated the structural and magnetic properties of Sm(Co0.7Fe0.1Ni0.12Zr0.04B0.04)7.5 melt spun ribbons. Samples were arc melted then melt spun at 37 m/s up to 55 m/s to obtain ribbon for powdering. Annealing has been performed in argon atmosphere for (30 to 75) min at (600 to 870) oC. In as-spun ribbons the hexagonal SmCo7 (TbCu7-type of structure) of crystal structure has been determined from x-ray diffraction patterns, while fcc-Co has been identified as a secondary phase. After annealing, the 1:7 phase of the as-spun ribbons transforms into 2:17 and 1:5 phases. X-ray patterns for as-milled powders exhibit very broad peaks making it difficult to identify a precise structure but represent the 1:7 structure after annealing at low temperature (650 oC). TEM analysis shows a homogeneous nanocrystalline microstructure with average grain size of (30 to 80) nm. Coercivity values of (15 to 27) kOe are obtained from hysteresis loops traced up to a field of 5 T. The coercivity decreases as temperature increases, but it maintains values higher than 5 kOe at 380 oC. The maximum energy product at room temperature increases, as high as 7.2 MGOe, for melt-spun isotropic ribbons produced at higher wheel speeds. Anisotropic ribbons have a maximum energy product close to 12 MGOe.

cond-mat.mtrl-sci↗