arXiv subjects

Ke Zhang

Publications and source records attributed to Ke Zhang.

At least 325 records · Page 18Linked to original sources

arXiv2019

DeepAuto: A Hierarchical Deep Learning Framework for Real-Time Prediction in Cellular Networks

Accurate real-time forecasting of key performance indicators (KPIs) is an essential requirement for various LTE/5G radio access network (RAN) automation. However, an accurate prediction can be very challenging in large-scale cellular environments due to complex spatio-temporal dynamics, network configuration changes and unavailability of real-time network data. In this work, we introduce a reusable analytics framework that enables real-time KPI prediction using a hierarchical deep learning architecture. Our prediction approach, namely DeepAuto, stacks multiple long short-term memory (LSTM) networks horizontally to capture instantaneous, periodic and seasonal patterns in KPI time-series. It further merge with feed-forward networks to learn the impact of network configurations and other external factors. We validate the approach by predicting two important KPIs, including cell load and radio channel quality, using large-scale real network streaming measurement data from the operator. For cell load prediction, DeepAuto model showed up to 15% improvement in Root Mean Square Error (RMSE) compared to naive method of using recent measurements for short-term horizon and up to 32% improvement for longer-term prediction.

eess.SP↗

arXiv2019

Nanoconfined, dynamic electrolyte gating and memory effects in multilayered graphene-based membranes

Multilayered graphene-based nanoporous membranes with electrolyte incorporated between individual sheets is a unique nano-heterostructure system in which nanoconfined electrons in graphene and ions confined in between sheets are intimately coupled throughout the entire membrane. In contrast to the general notion that the electrolyte gating is unlikely to appear in multilayered graphene stacks, it is demonstrated in this work that the electrolyte gating effect in monolayer graphene can be transferred to its corresponding multilayered porous membranes. This gating effect presented on each individual graphene sheets through electrolyte confined in nanopores provides a real-time, electrical approach for probing the complex dynamics of nanoconfined electrical double layer. This has enabled the observation of the ionic memory effect in supercapacitors and produces new insights into the charging dynamics of supercapacitors. Such discoveries may stimulate the design of novel nanoionic devices.

cond-mat.mes-hall↗

arXiv2019

Gapless surface Dirac cone in antiferromagnetic topological insulator MnBi$_2$Te$_4$

The recent discovered antiferromagnetic topological insulators in Mn-Bi-Te family with intrinsic magnetic ordering have rapidly drawn broad interest since its cleaved surface state is believed to be gapped, hosting the unprecedented axion states with half-integer quantum Hall effect. Here, however, we show unambiguously by using high-resolution angle-resolved photoemission spectroscopy that a gapless Dirac cone at the (0001) surface of MnBi$_2$Te$_4$ exists between the bulk band gap. Such unexpected surface state remains unchanged across the bulk Néel temperature, and is even robust against severe surface degradation, indicating additional topological protection. Through symmetry analysis and $\textit{ab}$-$\textit{initio}$ calculations we consider different types of surface reconstruction of the magnetic moments as possible origins giving rise to such linear dispersion. Our results reveal that the intrinsic magnetic topological insulator hosts a rich platform to realize various topological phases such as topological crystalline insulator and time-reversal-preserved topological insulator, by tuning the magnetic configurations.

cond-mat.mtrl-sci↗

arXiv2019

Controllable Length Control Neural Encoder-Decoder via Reinforcement Learning

Controlling output length in neural language generation is valuable in many scenarios, especially for the tasks that have length constraints. A model with stronger length control capacity can produce sentences with more specific length, however, it usually sacrifices semantic accuracy of the generated sentences. Here, we denote a concept of Controllable Length Control (CLC) for the trade-off between length control capacity and semantic accuracy of the language generation model. More specifically, CLC is to alter length control capacity of the model so as to generate sentence with corresponding quality. This is meaningful in real applications when length control capacity and outputs quality are requested with different priorities, or to overcome unstability of length control during model training. In this paper, we propose two reinforcement learning (RL) methods to adjust the trade-off between length control capacity and semantic accuracy of length control models. Results show that our RL methods improve scores across a wide range of target lengths and achieve the goal of CLC. Additionally, two models LenMC and LenLInit modified on previous length-control models are proposed to obtain better performance in summarization task while still maintain the ability to control length.

cs.CL↗

arXiv2019

Efficient variants of the CMRH method for solving a sequence of multi-shifted non-Hermitian linear systems simultaneously

Multi-shifted linear systems with non-Hermitian coefficient matrices arise in numerical solutions of time-dependent partial/fractional differential equations (PDEs/FDEs), in control theory, PageRank problems, and other research fields. We derive efficient variants of the restarted Changing Minimal Residual method based on the cost-effective Hessenberg procedure (CMRH) for this problem class. Then, we introduce a flexible variant of the algorithm that allows to use variable preconditioning at each iteration to further accelerate the convergence of shifted CMRH. We analyse the performance of the new class of methods in the numerical solution of PDEs and FDEs, also against other multi-shifted Krylov subspace methods.

math.NA↗

arXiv2019

Fine-Grained Age Estimation in the wild with Attention LSTM Networks

Age estimation from a single face image has been an essential task in the field of human-computer interaction and computer vision, which has a wide range of practical application values. Accuracy of age estimation of face images in the wild is relatively low for existing methods, because they only take into account the global features, while neglecting the fine-grained features of age-sensitive areas. We propose a novel method based on our attention long short-term memory (AL) network for fine-grained age estimation in the wild, inspired by the fine-grained categories and the visual attention mechanism. This method combines the residual networks (ResNets) or the residual network of residual network (RoR) models with LSTM units to construct AL-ResNets or AL-RoR networks to extract local features of age-sensitive regions, which effectively improves the age estimation accuracy. First, a ResNets or a RoR model pretrained on ImageNet dataset is selected as the basic model, which is then fine-tuned on the IMDB-WIKI-101 dataset for age estimation. Then, we fine-tune the ResNets or the RoR on the target age datasets to extract the global features of face images. To extract the local features of age-sensitive regions, the LSTM unit is then presented to obtain the coordinates of the agesensitive region automatically. Finally, the age group classification is conducted directly on the Adience dataset, and age-regression experiments are performed by the Deep EXpectation algorithm (DEX) on MORPH Album 2, FG-NET and 15/16LAP datasets. By combining the global and the local features, we obtain our final prediction results. Experimental results illustrate the effectiveness and robustness of the proposed AL-ResNets or AL-RoR for age estimation in the wild, where it achieves better state-of-the-art performance than all other convolutional neural network.

cs.CV↗

arXiv2019

Systematic Variations of CO Gas Abundance with Radius in Gas-rich Protoplanetary Disks

CO is the most widely used gas tracer of protoplanetary disks. Its abundance is usually assumed to be an interstellar ratio throughout the warm molecular layer of the disk. But recent observations of low CO gas abundance in many protoplanetary disks challenge our understanding of physical and chemical evolutions in disks. Here we investigate the CO abundance structures in four well-studied disks and compare their structures with predictions of chemical processing of CO and transport of CO ice-coated dust grains in disks. We use spatially resolved CO isotopologue line observations and detailed thermo-chemical models to derive CO abundance structures. We find that the CO abundance varies with radius by an order of magnitude in these disks. We show that although chemical processes can efficiently reduce the total column of CO gas within 1 Myr under an ISM level of cosmic-ray ionization rate, the depletion mostly occurs at the deep region of a disk. Without sufficient vertical mixing, the surface layer is not depleted enough to reproduce weak CO emissions observed. The radial profiles of CO depletion in three disks are qualitatively consistent with predictions of pebble formation, settling, and drifting in disks. But the dust evolution alone cannot fully explain the high depletion observed in some disks. These results suggest that dust evolution may play a significant role in transporting volatile materials and a coupled chemical-dynamical study is necessary to understand what raw materials are available for planet formation at different distances from the central star.

astro-ph.EP↗

arXiv2019

Competing Ratio Loss for Discriminative Multi-class Image Classification

The development of deep convolutional neural network architecture is critical to the improvement of image classification task performance. A lot of studies of image classification based on deep convolutional neural network focus on the network structure to improve the image classification performance. Contrary to these studies, we focus on the loss function. Cross-entropy Loss (CEL) is widely used for training a multi-class classification deep convolutional neural network. While CEL has been successfully implemented in image classification tasks, it only focuses on the posterior probability of correct class when the labels of training images are one-hot. It cannot be discriminated against the classes not belong to correct class (wrong classes) directly. In order to solve the problem of CEL, we propose Competing Ratio Loss (CRL), which calculates the posterior probability ratio between the correct class and competing wrong classes to better discriminate the correct class from competing wrong classes, increasing the difference between the negative log likelihood of the correct class and the negative log likelihood of competing wrong classes, widening the difference between the probability of the correct class and the probabilities of wrong classes. To demonstrate the effectiveness of our loss function, we perform some sets of experiments on different types of image classification datasets, including CIFAR, SVHN, CUB200- 2011, Adience and ImageNet datasets. The experimental results show the effectiveness and robustness of our loss function on different deep convolutional neural network architectures and different image classification tasks, such as fine-grained image classification, hard face age estimation and large-scale image classification.

cs.CV↗

arXiv2019

Hybrid Function Sparse Representation towards Image Super Resolution

Sparse representation with training-based dictionary has been shown successful on super resolution(SR) but still have some limitations. Based on the idea of making the magnification of function curve without losing its fidelity, we proposed a function based dictionary on sparse representation for super resolution, called hybrid function sparse representation (HFSR). The dictionary we designed is directly generated by preset hybrid functions without additional training, which can be scaled to any size as is required due to its scalable property. We mixed approximated Heaviside function (AHF), sine function and DCT function as the dictionary. Multi-scale refinement is then proposed to utilize the scalable property of the dictionary to improve the results. In addition, a reconstruct strategy is adopted to deal with the overlaps. The experiments on Set14 SR dataset show that our method has an excellent performance particularly with regards to images containing rich details and contexts compared with non-learning based state-of-the art methods.

cs.CV↗

arXiv2019

Unlocking CO Depletion in Protoplanetary Disks II. Primordial C/H Predictions Inside the CO Snowline

CO is thought to be the main reservoir of volatile carbon in protoplanetary disks, and thus the primary initial source of carbon in the atmospheres of forming giant planets. However, recent observations of protoplanetary disks point towards low volatile carbon abundances in many systems, including at radii interior to the CO snowline. One potential explanation is that gas phase carbon is chemically reprocessed into less volatile species, which are frozen on dust grain surfaces as ice. This mechanism has the potential to change the primordial C/H ratio in the gas. However, current observations primarily probe the upper layers of the disk. It is not clear if the low volatile carbon abundances extend to the midplane, where planets form. We have run a grid of 198 chemical models, exploring how the chemical reprocessing of CO depends on disk mass, dust grain size distribution, temperature, cosmic ray and X-ray ionization rate, and initial water abundance. Building on our previous work focusing on the warm molecular layer, here we analyze the results for our grid of models in the disk midplane at 12 au. We find that either an ISM level cosmic ray ionization rate or the presence of UV photons due to a low dust surface density are needed to chemically reduce the midplane CO gas abundance by at least an order of magnitude within 1 Myr. In the majority of our models CO does not undergo substantial reprocessing by in situ chemistry and there is little change in the gas phase C/H and C/O ratios over the lifetime of the typical disk. However, in the small sub-set of disks where the disk midplane is subject to a source of ionization or photolysis, the gas phase C/O ratio increases by up to nearly 9 orders of magnitude due to conversion of CO into volatile hydrocarbons.

astro-ph.EP↗

arXiv2019

The Sea Exploration Problem: Data-driven Orienteering on a Continuous Surface

This paper describes a problem arising in sea exploration, where the aim is to schedule the expedition of a ship for collecting information about the resources on the seafloor. The aim is to collect data by probing on a set of carefully chosen locations, so that the information available is optimally enriched. This problem has similarities with the orienteering problem, where the aim is to plan a time-limited trip for visiting a set of vertices, collecting a prize at each of them, in such a way that the total value collected is maximum. In our problem, the score at each vertex is associated with an estimation of the level of the resource on the given surface, which is done by regression using Gaussian processes. Hence, there is a correlation among scores on the selected vertices; this is a first difference with respect to the standard orienteering problem. The second difference is the location of each vertex, which in our problem is a freely chosen point on a given surface.

cs.AI↗

arXiv2019

Realizing the Unique Potential of ALMA to Probe the Gas Reservoir of Planet Formation

Understanding the origin of the astonishing diversity of exoplanets is a key question for the coming decades. ALMA has revolutionized our view of the dust emission from protoplanetary disks, demonstrating the prevalence of ring and spiral structures that are likely sculpted by young planets in formation. To detect kinematic signatures of these protoplanets and to probe the chemistry of their gas accretion reservoir will require the imaging of molecular spectral line emission at high angular and spectral resolution. However, the current sensitivity of ALMA limits these important spectral studies to only the nearest protoplanetary disks. Although some promising results are emerging, including the identification of the snowlines of a few key molecules and the first attempt at detecting a protoplanet's spiral wake, it is not yet possible to search for these important signatures in a population of disks in diverse environments and ages. Harnessing the tremendous power of (sub)mm observations to pinpoint and characterize the chemistry of planets in formation will require a major increase of ALMA's spectral sensitivity (5-10x), increase in instantaneous bandwidth (2x) at high spectral resolution, and improved angular resolution (2x) in the 2030 era.

astro-ph.SR↗

arXiv2019

The Disk Gas Mass and the Far-IR Revolution

The gaseous mass of protoplanetary disks is a fundamental quantity in planet formation. The presence of gas is necessary to assemble planetesimals, it determines timescales of giant planet birth, and it is an unknown factor for a wide range of properties of planet formation, from chemical abundances (X/H) to the mass efficiency of planet formation. The gas mass obtained from traditional tracers, such as dust thermal continuum and CO isotopologues, are now known to have significant (1 - 2 orders of magnitude) discrepancies. Emission from the isotopologue of H2, hydrogen deuteride (HD), offers an alternative measurement of the disk gas mass. Of all of the regions of the spectrum, the far-infrared stands out in that orders of magnitude gains in sensitivity can be gleaned by cooling a large aperture telescope to 8 K. Such a facility can open up a vast new area of the spectrum to exploration. One of the primary benefits of this far-infrared revolution would be the ability to survey hundreds of planet-forming disks in HD emission to derive their gaseous masses. For the first time, we will have statistics on the gas mass as a function of evolution, tracing birth to dispersal as a function of stellar spectral type. These measurements have broad implications for our understanding of the time scale during which gas is available to form giant planets, the dynamical evolution of the seeds of terrestrial worlds, and the resulting chemical composition of pre-planetary embryos carrying the elements needed for life. Measurements of the ground-state line of HD requires a space-based observatory operating in the far-infrared at 112 microns.

astro-ph.EP↗

arXiv2019

The trail of water and the delivery of volatiles to habitable planets

Water is fundamental to our understanding of the evolution of planetary systems and the delivery of volatiles to the surfaces of potentially habitable planets. Yet, we currently have essentially no facilities capable of observing this key species comprehensively. With this white paper, we argue that we need a relatively large, cold space-based observatory equipped with a high-resolution spectrometer, in the mid- through far-infrared wavelength range (25-600~$μ$m) in order to answer basic questions about planet formation, such as where the Earth got its water, how giant planets and planetesimals grow, and whether water is generally available to planets forming in the habitable zone of their host stars.

astro-ph.EP↗

arXiv2019

The essential coexistence phenomenon in Hamiltonian dynamics

We construct an example of a Hamiltonian flow $f^t$ on a $4$-dimensional smooth manifold $\mathcal{M}$ which after being restricted to an energy surface $\mathcal{M}_e$ demonstrates essential coexistence of regular and chaotic dynamics that is there is an open and dense $f^t$-invariant subset $U\subset\mathcal{M}_e$ such that the restriction $f^t|U$ has non-zero Lyapunov exponents in all directions (except the direction of the flow) and is a Bernoulli flow while on the boundary $\partial U$, which has positive volume all Lyapunov exponents of the system are zero.

math.DS↗

arXiv2019

A high resolution mid-infrared survey of water emission from protoplanetary disks

We present the largest survey of spectrally resolved mid-infrared water emission to date, with spectra for 11 disks obtained with the Michelle and TEXES spectrographs on Gemini North. Water emission is detected in 6 of 8 disks around classical T Tauri stars. Water emission is not detected in the transitional disks SR 24 N and SR 24 S, in spite of SR 24 S having pre-transitional disk properties like DoAr 44, which does show water emission (Salyk et al. 2015). With R~100,000, the TEXES water spectra have the highest spectral resolution possible at this time, and allow for detailed lineshape analysis. We find that the mid-IR water emission lines are similar to the "narrow component" in CO rovibrational emission (Banzatti & Pontoppidan 2015), consistent with disk radii of a few AU. The emission lines are either single peaked, or consistent with a double peak. Single-peaked emission lines cannot be produced with a Keplerian disk model, and may suggest that water participates in the disk winds proposed to explain single-peaked CO emission lines (Bast et al. 2011, Pontoppidan et al. 2011). Double-peaked emission lines can be used to determine the radius at which the line emission luminosity drops off. For HL Tau, the lower limit on this measured dropoff radius is consistent with the 13 AU dark ring (ALMA partnership et al. 2015). We also report variable line/continuum ratios from the disks around DR Tau and RW Aur, which we attribute to continuum changes and line flux changes, respectively. The reduction in RW Aur line flux corresponds with an observed dimming at visible wavelengths (Rodriguez et al. 2013).

astro-ph.SR↗

arXiv2018

Science with an ngVLA: Tracing the Water Snowline in Protoplanetary disks with the ngVLA

The water snowline in protoplanetary disks is one of the most pivotal locations for planet formation: it sets a critical boundary of chemical composition in the disk and likely serves as a favorable site of planetesimal growth and planet formation. However, the water snowline at the mid-plane cannot be directly traced by water line observations as the dust emission around the water snowline is highly optically thick at these line frequencies. Alternatively, the water snowline maybe traced through the sharp changes in physical properties (surface density and dust size distribution) at this transition or through the snowlines of other molecules that sublimate co-spatially with water. In this chapter, we discuss the ngVLA's capability of tracing the mid-plane water snowline through mapping sharp transitions in the continuum spectral index caused by changes of dust size distributions and by using the NH3 snowline as a proxy. We found that with the current design of the ngVLA ability, a sharp transition in spectral index across the water snowline can be readily traced in relatively bright disks. A detection of the NH3 snowline is challenging under the current design, but may still be possible if the NH3-to-water abundance in the disk is close to that of the ISM ices. In summary, the proposed ngVLA is the only facility can provide sufficient sensitivity and spatial resolution to trace the evolution of water snowline at the disk mid-plane and to reveal its role in accelerating the planetesimal formation and in building the architecture of planetary systems.

astro-ph.EP↗

arXiv2018

Diffusion limit for a slow-fast standard map

Consider the map $(x, y) \mapsto (x + ε^{-α} \sin (2πx) + ε^{-1-α}z, z + ε\sin(2πx))$, which is conjugate to the Chirikov standard map with a large parameter. The parameter value $α= 1$ is related to "scattering by resonance" phenomena. For suitable $α$, we obtain a central limit theorem for the slow variable $z$ for a (Lebesgue) random initial condition. The result is proved by conjugating to the Chirikov standard map and utilizing the formalism of standard pairs. Our techniques also yield for the Chirikov standard map a related limit theorem and a "finite-time" decay of correlations result.

math.DS↗