arXiv subjects

Ye Yuan

Publications and source records attributed to Ye Yuan.

At least 325 records · Page 18Linked to original sources

arXiv2018

Understanding Convolution for Semantic Segmentation

Recent advances in deep learning, especially deep convolutional neural networks (CNNs), have led to significant improvement over previous semantic segmentation systems. Here we show how to improve pixel-wise semantic segmentation by manipulating convolution-related operations that are of both theoretical and practical value. First, we design dense upsampling convolution (DUC) to generate pixel-level prediction, which is able to capture and decode more detailed information that is generally missing in bilinear upsampling. Second, we propose a hybrid dilated convolution (HDC) framework in the encoding phase. This framework 1) effectively enlarges the receptive fields (RF) of the network to aggregate global information; 2) alleviates what we call the "gridding issue" caused by the standard dilated convolution operation. We evaluate our approaches thoroughly on the Cityscapes dataset, and achieve a state-of-art result of 80.1% mIOU in the test set at the time of submission. We also have achieved state-of-the-art overall on the KITTI road estimation benchmark and the PASCAL VOC2012 segmentation task. Our source code can be found at https://github.com/TuSimple/TuSimple-DUC .

cs.CV↗

arXiv2018

Blind Identification of Fully Observed Linear Time-Varying Systems via Sparse Recovery

Discrete-time linear time-varying (LTV) systems form a powerful class of models to approximate complex dynamical systems with nonlinear dynamics for the purpose of analysis, design and control. Motivated by inference of spatio-temporal dynamics in breast cancer research, we propose a method to efficiently solve an identification problem for a specific class of discrete-time LTV systems, in which the states are fully observed and there is no access to system inputs. In addition, it is assumed that we do not know on which states the inputs act, which can change between time steps, and that the total number of inputs is sparse over all states and over time. The problem is formulated as a compressive sensing problem, which incorporates the effect of measurement noise and which has a solution with a partially sparse support. We derive sufficient conditions for the unique recovery of the system model and input values, which lead to practical conditions on the number of experiments and rank conditions on system outputs. Synthetic experiments analyze the method's sensitivity to noise for randomly generated models.

eess.SY↗

arXiv2018

Link prediction via linear optimization

Link prediction is an elemental challenge in network science, which has already found applications in guiding laboratorial experiments, digging out drug targets, recommending friends in social networks, probing mechanisms in network evolution, and so on. With a simple assumption that the likelihood of the existence of a link between two nodes can be unfolded by a linear summation of neighboring nodes' contributions, we obtain the analytical solution of the optimal likelihood matrix, which shows remarkably better performance in predicting missing links than the state-of-the-art algorithms for not only simple networks, but also weighted and directed networks. To our surprise, even some degenerated local similarity indices from the solution outperform well-known local indices, which largely refines our knowledge, for example, the direct count of the number of 3-hop paths between two nodes more accurately predicts missing links than the number of 2-hop paths (i.e., the number of common neighbors), while in the previous studies, as indicated by the local path index and Katz index, the statistics on longer paths are always considered to be complementary to but less important than those on shorter paths.

physics.soc-ph↗

arXiv2018

The pc-scale radio structure of MIR-observed radio galaxies

We investigated the relationship between the accretion process and jet properties by ultilizing the VLBA and mid-infrared (MIR) data for a sample of 45 3CRR radio galaxies selected with a flux density at 178 MHz $>16.4$ Jy, 5 GHz VLA core flux density $\geq$ 7 mJy, and MIR observations. The pc-scale radio structure at 5 GHz are presented by using our VLBA observations for 21 sources in February, 2016, the analysis on the archival data for 16 objects, and directly taking the measurements for 8 radio galaxies available in literatures. The accretion mode is constrained from the Eddington ratio with a dividing value of 0.01, which is estimated from the MIR-based bolometric luminosity and the black hole masses. While most FRII radio galaxies have higher Eddington ratio than FRIs, we found that there is indeed no single correspondence between the FR morphology and accretion mode with eight FRIIs at low accretion and two FRIs at high accretion rate. There is a significant correlation between the VLBA core luminosity at 5 GHz and the Eddington ratio. Various morphologies are found in our sample, including core only, single-sided core-jet, and two-sided core-jet structures. We found that the higher accretion rate may be more likely related with the core-jet structure, thus more extended jet. These results imply that the higher accretion rates are likely able to produce more powerful jets. There is a strong correlation between the MIR luminosity at 15 $μ$m and VLBA 5 GHz core luminosity, in favour of the tight relation between the accretion disk and jets. In our sample, the core brightness temperature ranges from $10^{9}$ to $10^{13.38}$ K with a median value of $10^{11.09}$ K indicating that systematically the beaming effect may not be significant....

astro-ph.GA↗

arXiv2018

SFace: An Efficient Network for Face Detection in Large Scale Variations

Face detection serves as a fundamental research topic for many applications like face recognition. Impressive progress has been made especially with the recent development of convolutional neural networks. However, the issue of large scale variations, which widely exists in high resolution images/videos, has not been well addressed in the literature. In this paper, we present a novel algorithm called SFace, which efficiently integrates the anchor-based method and anchor-free method to address the scale issues. A new dataset called 4K-Face is also introduced to evaluate the performance of face detection with extreme large scale variations. The SFace architecture shows promising results on the new 4K-Face benchmarks. In addition, our method can run at 50 frames per second (fps) with an accuracy of 80% AP on the standard WIDER FACE dataset, which outperforms the state-of-art algorithms by almost one order of magnitude in speed while achieves comparative performance.

cs.CV↗

arXiv2018

Interactive Co-Design of Form and Function for Legged Robots using the Adjoint Method

Our goal is to make robotics more accessible to casual users by reducing the domain knowledge required in designing and building robots. Towards this goal, we present an interactive computational design system that enables users to design legged robots with desired morphologies and behaviors by specifying higher level descriptions. The core of our method is a design optimization technique that reasons about the structure, and motion of a robot in coupled manner in order to achieve user-specified robot behavior, and performance. We are inspired by the recent works that also aim to jointly optimize robot's form and function. However, through efficient computation of necessary design changes, our approach enables us to keep user-in-the-loop for interactive applications. We evaluate our system in simulation by automatically improving robot designs for multiple scenarios. Starting with initial user designs that are physically infeasible or inadequate to perform the user-desired task, we show optimized designs that achieve user-specifications, all while ensuring an interactive design flow.

cs.RO↗

arXiv2018

CMOS-compatible controlled hyperdoping of silicon nanowires

Hyperdoping consists of the intentional introduction of deep-level dopants into a semiconductor in excess of equilibrium concentrations. This causes a broadening of dopant energy levels into an intermediate band between the valence and conduction bands.[1,2] Recently, bulk Si hyperdoped with chalcogens or transition metals has been demonstrated to be an appropriate intermediate-band material for Si-based short-wavelength infrared photodetectors.[3-5] Intermediate-band nanowires could potentially be used instead of bulk materials to overcome the Shockley-Queisser limit and to improve efficiency in solar cells,[6-9] but fundamental scientific questions in hyperdoping Si nanowires require experimental verification. The development of a method for obtaining controlled hyperdoping levels at the nanoscale concomitant with the electrical activation of dopants is, therefore, vital to understanding these issues. Here, we show a CMOS-compatible technique based on non-equilibrium processing for the controlled doping of Si at the nanoscale with dopant concentrations several orders of magnitude greater than the equilibrium solid solubility. Through the nanoscale spatially controlled implantation of dopants, and a bottom-up template-assisted solid phase recrystallization of the nanowires with the use of millisecond-flash lamp annealing, we form Se-hyperdoped Si/SiO2 core/shell nanowires that have a room-temperature sub-band gap optoelectronic photoresponse when configured as a photoconductor device.

cond-mat.mtrl-sci↗

arXiv2018

Electronic phase separation in insulating (Ga,Mn)As with low compensation: Super-paramagnetism and hopping conduction

In the present work, low compensated insulating (Ga,Mn)As with 0.7% Mn is obtained by ion implantation combined with pulsed laser melting. The sample shows variable-range hopping transport behavior with a Coulomb gap in the vicinity of the Fermi energy, and the activation energy is reduced by an external magnetic field. A blocking super-paramagnetism is observed rather than ferromagnetism. Below the blocking temperature, the sample exhibits a colossal negative magnetoresistance. Our studies confirm that the disorder-induced electronic phase separation occurs in (Ga,Mn)As samples with a Mn concentration in the insulator-metal transition regime, and it can account for the observed superparamagnetism and the colossal magnetoresistance.

cond-mat.mtrl-sci↗

arXiv2017

Chiral liquid crystal colloids

Colloidal particles disturb the alignment of rod-like molecules of liquid crystals, giving rise to long-range interactions that minimize the free energy of distorted regions. Particle shape and topology are known to guide this self-assembly process. However, how chirality of colloidal inclusions affects these long-range interactions is unclear. Here we study the effects of distortions caused by chiral springs and helices on the colloidal self-organization in a nematic liquid crystal using laser tweezers, particle tracking and optical imaging. We show that chirality of colloidal particles interacts with the nematic elasticity to predefine chiral or racemic colloidal superstructures in nematic colloids. These findings are consistent with numerical modelling based on the minimization of Landau-de Gennes free energy. Our study uncovers the role of chirality in defining the mesoscopic order of liquid crystal colloids, suggesting that this feature may be a potential tool to modulate the global orientated self-organization of these systems.

cond-mat.soft↗

arXiv2017

Face Attention Network: An Effective Face Detector for the Occluded Faces

The performance of face detection has been largely improved with the development of convolutional neural network. However, the occlusion issue due to mask and sunglasses, is still a challenging problem. The improvement on the recall of these occluded cases usually brings the risk of high false positives. In this paper, we present a novel face detector called Face Attention Network (FAN), which can significantly improve the recall of the face detection problem in the occluded case without compromising the speed. More specifically, we propose a new anchor-level attention, which will highlight the features from the face region. Integrated with our anchor assign strategy and data augmentation techniques, we obtain state-of-art results on public face detection benchmarks like WiderFace and MAFA. The code will be released for reproduction.

cs.CV↗

arXiv2017

Investigation of a possible electronic phase separation in the magnetic semiconductors Ga$_{1-x}$Mn$_{x}$As and Ga$_{1-x}$Mn$_{x}$P by means of fluctuation spectroscopy

We present systematic temperature-dependent resistance noise measurements on a series of ferromagnetic Ga$_{1-x}$Mn$_{x}$As epitaxial thin films covering a large parameter space in terms of the Mn content $x$ and other variations regarding sample fabrication. We infer that the electronic noise is dominated by switching processes related to impurities in the entire temperature range. While metallic compounds with $x>2$ % do not exhibit any significant change in the low-frequency resistance noise around the Curie temperature $T_\mathrm{C}$, we find indications for an electronic phase separation in films with $x<2$ % in the vicinity of $T_\mathrm{C}$, manifesting itself in a maximum in the noise power spectral density. These results are compared with noise measurements on an insulating Ga$_{1-x}$Mn$_{x}$P reference sample, for which the evidence for an electronic phase separation is even stronger and a possible percolation of bound magnetic polarons is discussed. Another aspect addressed in this work is the effect of ion-irradiation induced disorder on the electronic properties of Ga$_{1-x}$Mn$_{x}$As films and, in particular, whether any electronic inhomogeneities can be observed in this case. Finally, we put our findings into the context of the ongoing debate on the electronic structure and the development of spontaneous magnetization in these materials.

cond-mat.mtrl-sci↗

arXiv2017

On Identification of Distribution Grids

Large-scale integration of distributed energy resources into residential distribution feeders necessitates careful control of their operation through power flow analysis. While the knowledge of the distribution system model is crucial for this type of analysis, it is often unavailable or outdated. The recent introduction of synchrophasor technology in low-voltage distribution grids has created an unprecedented opportunity to learn this model from high-precision, time-synchronized measurements of voltage and current phasors at various locations. This paper focuses on joint estimation of model parameters (admittance values) and operational structure of a poly-phase distribution network from the available telemetry data via the lasso, a method for regression shrinkage and selection. We propose tractable convex programs capable of tackling the low rank structure of the distribution system and develop an online algorithm for early detection and localization of critical events that induce a change in the admittance matrix. The efficacy of these techniques is corroborated through power flow studies on four three-phase radial distribution systems serving real household demands.

cs.LG↗

arXiv2017

Ultrafast Synchronization via Local Observation

Rapid expansions of their size and frequent changes of their topology make it difficult to observe and analyze complex networks. We explore the properties of the Hankel matrix and propose an algorithm for calculating the final synchronization state that uses a local observation of a single node for a time period significantly shorter than the synchronization process. We find that synchronization can be achieved more quickly than the routine rhythm. This finding refines our understanding of the abundant ultrafast synchronization phenomena observed in nature, and it enables the efficient design of self-aligned robots.

physics.soc-ph↗

arXiv2017

Towards diluted magnetism in TaAs

Magnetism in Weyl semimetals is desired to investigate the interaction between the magnetic moments and Weyl fermions, e.g. to explore anomalous quantum Hall phenomena. Here we demonstrate that proton irradiation is an effective tool to induce ferromagnetism in the Weyl semimetal TaAs. The intrinsic magnetism is observed with a transition temperature above room temperature. The magnetic moments from d states are found to be localized around Ta atoms. Further, the first-principles calculations indicate that the d states localized on the nearest-neighbor Ta atoms of As vacancy sites are responsible for the observed magnetic moments and the long-ranged magnetic order. The results show the feasibility of inducing ferromagnetism in Weyl semimetals so that they may facilitate the applications of this material in spintronics.

cond-mat.mtrl-sci↗

arXiv2017

Interplay between localization and magnetism in (Ga,Mn)As and (In,Mn)As

Ion implantation of Mn combined with pulsed laser melting is employed to obtain two representative compounds of dilute ferromagnetic semiconductors (DFSs): Ga1-xMnxAs and In1-xMnxAs. In contrast to films deposited by the widely used molecular beam epitaxy, neither Mn interstitials nor As antisites are present in samples prepared by the method employed here. Under these conditions the influence of localization on the hole-mediated ferromagnetism is examined in two DFSs with a differing strength of p-d coupling. On the insulating side of the transition, ferromagnetic signatures persist to higher temperatures in In1-xMnxAs compared to Ga1-xMnxAs with the same Mn concentration x. This substantiates theoretical suggestions that stronger p-d coupling results in an enhanced contribution to localization, which reduces hole-mediated ferromagnetism. Furthermore, the findings support strongly the heterogeneous model of electronic states at the localization boundary and point to the crucial role of weakly localized holes in mediating efficient spin-spin interactions even on the insulator side of the metal-insulator transition.

cond-mat.mtrl-sci↗

arXiv2017

Words or Characters? Fine-grained Gating for Reading Comprehension

Previous work combines word-level and character-level representations using concatenation or scalar weighting, which is suboptimal for high-level tasks like reading comprehension. We present a fine-grained gating mechanism to dynamically combine word-level and character-level representations based on properties of the words. We also extend the idea of fine-grained gating to modeling the interaction between questions and paragraphs for reading comprehension. Experiments show that our approach can improve the performance on reading comprehension tasks, achieving new state-of-the-art results on the Children's Book Test dataset. To demonstrate the generality of our gating mechanism, we also show improved results on a social media tag prediction task.

cs.CL↗

arXiv2017

On the Powerball Method for Optimization

We propose a new method to accelerate the convergence of optimization algorithms. This method simply adds a power coefficient $γ\in[0,1)$ to the gradient during optimization. We call this the Powerball method and analyze the convergence rate for the Powerball method for strongly convex functions. While theoretically the Powerball method is guaranteed to have a linear convergence rate in the same order of the gradient method, we show that empirically it significantly outperforms the gradient descent and Newton's method, especially during the initial iterations. We demonstrate that the Powerball method provides a $10$-fold speedup of the convergence of both gradient descent and L-BFGS on multiple real datasets.

eess.SY↗

arXiv2017

Dehydrogenation through the pressure-induced polymerization processes of phosphine

PH3 is studied to understand the superconducting transition and responsible stoichiometry under high pressure by means of Raman, IR, and x-ray diffraction (XRD) measurements, and theoretical calculations. It is found PH3 is stable up to about 8 GPa and then starts to dehydrogenate through two dimerization processes at room temperature as pressure up to 25 GPa. Two resulting phosphorus hydrides, P2H4 and P4H6, are verified experimentally and can be recovered to ambient pressure. On further compression above 35 GPa, P4H6 directly decomposes into elemental phosphorus. The superconductivity transition temperatures of P4H6 at 100 and 200 GPa have been predicted to be 13 and 67 K in agreement with reported results, suggesting it might responsible for the superconductivity at higher pressures. Our results clearly show that P2H4 and P4H6 are only stable P-H compounds between PH3 and elemental phosphorus, shedding light on the superconducting mechanism.

cond-mat.mtrl-sci↗