arXiv subjects

Ke Zhang

Publications and source records attributed to Ke Zhang.

At least 289 records · Page 16Linked to original sources

arXiv2021

Compacting Deep Neural Networks for Internet of Things: Methods and Applications

Deep Neural Networks (DNNs) have shown great success in completing complex tasks. However, DNNs inevitably bring high computational cost and storage consumption due to the complexity of hierarchical structures, thereby hindering their wide deployment in Internet-of-Things (IoT) devices, which have limited computational capability and storage capacity. Therefore, it is a necessity to investigate the technologies to compact DNNs. Despite tremendous advances in compacting DNNs, few surveys summarize compacting-DNNs technologies, especially for IoT applications. Hence, this paper presents a comprehensive study on compacting-DNNs technologies. We categorize compacting-DNNs technologies into three major types: 1) network model compression, 2) Knowledge Distillation (KD), 3) modification of network structures. We also elaborate on the diversity of these approaches and make side-by-side comparisons. Moreover, we discuss the applications of compacted DNNs in various IoT applications and outline future directions.

cs.LG↗

arXiv2021

Destruction of refractory carbon grains drives the final stage of proto-planetary disk chemistry

Here we aim to explore the origin of the strong C2H lines to reimagine the chemistry of protoplanetary disks. There are a few key aspects that drive our analysis. First, C2H is detected in young and old systems, hinting at a long-lived chemistry. Second, as a radical, C2H is rapidly destroyed, within <1000 yr. These two statements hint that the chemistry responsible for C2H emission must be predominantly in the gas-phase and must be in equilibrium. Combining new and published chemical models we find that elevating the total volatile (gas and ice) C/O ratio is the only natural way to create a long lived, high C2H abundance. Most of the \ce{C2H} resides in gas with a Fuv/n-gas ~ 10^-7 G0 cm^3. To elevate the volatile C/O ratio, additional carbon has to be released into the gas to enable an equilibrium chemistry under oxygen-poor conditions. Photo-ablation of carbon-rich grains seems the most straightforward way to elevate the C/O ratio above 1.5, powering a long-lived equilibrium cycle. The regions at which the conditions are optimal for the presence of high C/O ratio and elevated C2H abundances in the gas disk set by the Fuv/n-gas ~ 10^-7 G0 cm^3 condition lie just outside the pebble disk as well as possibly in disk gaps. This process can thus also explain the (hints of) structure seen in C2H observations.

astro-ph.EP↗

arXiv2021

Probing fast oscillating scalar dark matter with atoms and molecules

Light scalar Dark Matter with scalar couplings to matter is expected within several scenarios to induce variations in the fundamental constants of nature. Such variations can be searched for, among other ways, via atomic spectroscopy. Sensitive atomic observables arise primarily due to possible changes in the fine-structure constant or the electron mass. Most of the searches to date have focused on slow variations of the constants (i.e. modulation frequencies $<$ 1 Hz). In a recent experiment \mbox{[Phys. Rev. Lett. 123, 141102 (2019)]} called WReSL (Weekend Relaxion-Search Laboratory), we reported on a direct search for rapid variations in the radio-frequency band. Such a search is particularly motivated within a class of relaxion Dark Matter models. We discuss the WReSL experiment, report on progress towards improved measurements of rapid fundamental constant variations, and discuss the planned extension of the work to molecules, in which rapid variations of the nuclear mass can be sensitively searched for.

physics.atom-ph↗

arXiv2021

Efficient erbium-doped thin-film lithium niobate waveguide amplifiers

Lithium niobate on insulator (LNOI) is an emerging photonic platform with great promises for future optical communications, nonlinear optics and microwave photonics. An important integrated photonic building block, active waveguide amplifiers, however, is still missing in the LNOI platform. Here we report an efficient and compact waveguide amplifier based on erbium-doped LNOI waveguides, realized by a sequence of erbium-doped crystal growth, ion slicing and lithography-based waveguide fabrication. Using a compact 5-mm-long waveguide, we demonstrate on-chip net gain of > 5 dB for 1530-nm signal light with a relatively low pump power of 21 mW at 980 nm. The efficient LNOI waveguide amplifiers could become an important fundamental element in future lithium niobate photonic integrated circuits.

physics.optics↗

arXiv2020

Observation of spin-momentum-layer locking in centrosymmetric BiOI

Spin polarization effects in nonmagnetic materials are generally believed as an outcome of spin-orbit coupling provided that the global inversion symmetry is lacking, also known as 'spin-momentum locking'. The recently discovered hidden spin polarization indicates that specific atomic site asymmetry could also induce measurable spin polarization, leading to a paradigm shift to centrosymmetric crystals for potential spintronic applications. Here, combining spin- and angle-resolved photoemission spectroscopy and theoretical calculations, we report distinct spin-layer locking phenomena surrounding different high-symmetry momenta in a centrosymmetric, layered material BiOI. The measured spin is highly polarized along the Brillouin zone boundary, while is almost vanishing around the zone center due to its nonsymmorphic crystal structure. Our work not only demonstrates the existence of hidden spin polarization, but also uncovers the microscopic mechanism of the way spin, momentum and layer locking to each other, shedding lights on the design metrics for future spintronic devices.

cond-mat.mtrl-sci↗

arXiv2020

The TW Hya Rosetta Stone Project III: Resolving the Gaseous Thermal Profile of the Disk

The thermal structure of protoplanetary disks is a fundamental characteristic of the system that has wide reaching effects on disk evolution and planet formation. In this study, we constrain the 2D thermal structure of the protoplanetary disk TW Hya structure utilizing images of seven CO lines. This includes new ALMA observations of 12CO J=2-1 and C18O J=2-1 as well as archival ALMA observations of 12CO J=3-2, 13CO J=3-2, 6-5, C18O J= 3-2, 6-5. Additionally, we reproduce a Herschel observation of the HD J=1-0 line flux, the spectral energy distribution, and utilize a recent quantification of CO radial depletion in TW Hya. These observations were modeled using the thermochemical code RAC2D, and our best fit model reproduces all spatially resolved CO surface brightness profiles. The resulting thermal profile finds a disk mass of 0.025 Msun and a thin upper layer of gas depleted of small dust with a thickness of approx 1.2% of the corresponding radius. Using our final thermal structure, we find that CO alone is not a viable mass tracer as its abundance is degenerate with the total H2 surface density. Different mass models can readily match the spatially resolved CO line profiles with disparate abundance assumptions. Mass determination requires additional knowledge and, in this work, HD provides the additional constraint to derive the gas mass and supports the inference of CO depletion in the TW Hya disk. Our final thermal structure confirms the use of HD as a powerful probe of protoplanetary disk mass. Additionally, the method laid out in this paper is an employable strategy for extraction of disk temperatures and masses in the future.

astro-ph.EP↗

arXiv2020

Deep Reinforcement Learning and Permissioned Blockchain for Content Caching in Vehicular Edge Computing and Networks

Vehicular Edge Computing (VEC) is a promising paradigm to enable huge amount of data and multimedia content to be cached in proximity to vehicles. However, high mobility of vehicles and dynamic wireless channel condition make it challenge to design an optimal content caching policy. Further, with much sensitive personal information, vehicles may be not willing to caching their contents to an untrusted caching provider. Deep Reinforcement Learning (DRL) is an emerging technique to solve the problem with high-dimensional and time-varying features. Permission blockchain is able to establish a secure and decentralized peer-to-peer transaction environment. In this paper, we integrate DRL and permissioned blockchain into vehicular networks for intelligent and secure content caching. We first propose a blockchain empowered distributed content caching framework where vehicles perform content caching and base stations maintain the permissioned blockchain. Then, we exploit the advanced DRL approach to design an optimal content caching scheme with taking mobility into account. Finally, we propose a new block verifier selection method, Proof-of-Utility (PoU), to accelerate block verification process. Security analysis shows that our proposed blockchain empowered content caching can achieve security and privacy protection. Numerical results based on a real dataset from Uber indicate that the DRL-inspired content caching scheme significantly outperforms two benchmark policies.

cs.CR↗

arXiv2020

Deep Reinforcement Learning for Stochastic Computation Offloading in Digital Twin Networks

The rapid development of Industrial Internet of Things (IIoT) requires industrial production towards digitalization to improve network efficiency. Digital Twin is a promising technology to empower the digital transformation of IIoT by creating virtual models of physical objects. However, the provision of network efficiency in IIoT is very challenging due to resource-constrained devices, stochastic tasks, and resources heterogeneity. Distributed resources in IIoT networks can be efficiently exploited through computation offloading to reduce energy consumption while enhancing data processing efficiency. In this paper, we first propose a new paradigm Digital Twin Networks (DTN) to build network topology and the stochastic task arrival model in IIoT systems. Then, we formulate the stochastic computation offloading and resource allocation problem to minimize the long-term energy efficiency. As the formulated problem is a stochastic programming problem, we leverage Lyapunov optimization technique to transform the original problem into a deterministic per-time slot problem. Finally, we present Asynchronous Actor-Critic (AAC) algorithm to find the optimal stochastic computation offloading policy. Illustrative results demonstrate that our proposed scheme is able to significantly outperforms the benchmarks.

cs.LG↗

arXiv2020

Edge Intelligence for Energy-efficient Computation Offloading and Resource Allocation in 5G Beyond

5G beyond is an end-edge-cloud orchestrated network that can exploit heterogeneous capabilities of the end devices, edge servers, and the cloud and thus has the potential to enable computation-intensive and delay-sensitive applications via computation offloading. However, in multi user wireless networks, diverse application requirements and the possibility of various radio access modes for communication among devices make it challenging to design an optimal computation offloading scheme. In addition, having access to complete network information that includes variables such as wireless channel state, and available bandwidth and computation resources, is a major issue. Deep Reinforcement Learning (DRL) is an emerging technique to address such an issue with limited and less accurate network information. In this paper, we utilize DRL to design an optimal computation offloading and resource allocation strategy for minimizing system energy consumption. We first present a multi-user end-edge-cloud orchestrated network where all devices and base stations have computation capabilities. Then, we formulate the joint computation offloading and resource allocation problem as a Markov Decision Process (MDP) and propose a new DRL algorithm to minimize system energy consumption. Numerical results based on a real-world dataset demonstrate that the proposed DRL-based algorithm significantly outperforms the benchmark policies in terms of system energy consumption. Extensive simulations show that learning rate, discount factor, and number of devices have considerable influence on the performance of the proposed algorithm.

cs.LG↗

arXiv2020

Recognition and standardization of cardiac MRI orientation via multi-tasking learning and deep neural networks

In this paper, we study the problem of imaging orientation in cardiac MRI, and propose a framework to categorize the orientation for recognition and standardization via deep neural networks. The method uses a new multi-tasking strategy, where both the tasks of cardiac segmentation and orientation recognition are simultaneously achieved. For multiple sequences and modalities of MRI, we propose a transfer learning strategy, which adapts our proposed model from a single modality to multiple modalities. We embed the orientation recognition network in a Cardiac MRI Orientation Adjust Tool, i.e., CMRadjustNet. We implemented two versions of CMRadjustNet, including a user-interface (UI) software, and a command-line tool. The former version supports MRI image visualization, orientation prediction, adjustment, and storage operations; and the latter version enables the batch operations. The source code, neural network models and tools have been released and open via https://zmiclab.github.io/projects.html.

eess.IV↗

arXiv2020

Anatomy Prior Based U-net for Pathology Segmentation with Attention

Pathological area segmentation in cardiac magnetic resonance (MR) images plays a vital role in the clinical diagnosis of cardiovascular diseases. Because of the irregular shape and small area, pathological segmentation has always been a challenging task. We propose an anatomy prior based framework, which combines the U-net segmentation network with the attention technique. Leveraging the fact that the pathology is inclusive, we propose a neighborhood penalty strategy to gauge the inclusion relationship between the myocardium and the myocardial infarction and no-reflow areas. This neighborhood penalty strategy can be applied to any two labels with inclusive relationships (such as the whole infarction and myocardium, etc.) to form a neighboring loss. The proposed framework is evaluated on the EMIDEC dataset. Results show that our framework is effective in pathological area segmentation.

eess.IV↗

arXiv2020

Low-latency Federated Learning and Blockchain for Edge Association in Digital Twin empowered 6G Networks

Emerging technologies such as digital twins and 6th Generation mobile networks (6G) have accelerated the realization of edge intelligence in Industrial Internet of Things (IIoT). The integration of digital twin and 6G bridges the physical system with digital space and enables robust instant wireless connectivity. With increasing concerns on data privacy, federated learning has been regarded as a promising solution for deploying distributed data processing and learning in wireless networks. However, unreliable communication channels, limited resources, and lack of trust among users, hinder the effective application of federated learning in IIoT. In this paper, we introduce the Digital Twin Wireless Networks (DTWN) by incorporating digital twins into wireless networks, to migrate real-time data processing and computation to the edge plane. Then, we propose a blockchain empowered federated learning framework running in the DTWN for collaborative computing, which improves the reliability and security of the system, and enhances data privacy. Moreover, to balance the learning accuracy and time cost of the proposed scheme, we formulate an optimization problem for edge association by jointly considering digital twin association, training data batch size, and bandwidth allocation. We exploit multi-agent reinforcement learning to find an optimal solution to the problem. Numerical results on real-world dataset show that the proposed scheme yields improved efficiency and reduced cost compared to benchmark learning method.

cs.LG↗

arXiv2020

Multi-Vehicle Routing Problems with Soft Time Windows: A Multi-Agent Reinforcement Learning Approach

Multi-vehicle routing problem with soft time windows (MVRPSTW) is an indispensable constituent in urban logistics distribution systems. Over the past decade, numerous methods for MVRPSTW have been proposed, but most are based on heuristic rules that require a large amount of computation time. With the current rapid increase of logistics demands, traditional methods incur the dilemma between computational efficiency and solution quality. To efficiently solve the problem, we propose a novel reinforcement learning algorithm called the Multi-Agent Attention Model that can solve routing problem instantly benefit from lengthy offline training. Specifically, the vehicle routing problem is regarded as a vehicle tour generation process, and an encoder-decoder framework with attention layers is proposed to generate tours of multiple vehicles iteratively. Furthermore, a multi-agent reinforcement learning method with an unsupervised auxiliary network is developed for the model training. By evaluated on four synthetic networks with different scales, the results demonstrate that the proposed method consistently outperforms Google OR-Tools and traditional methods with little computation time. In addition, we validate the robustness of the well-trained model by varying the number of customers and the capacities of vehicles.

cs.AI↗

arXiv2020

Chemical Evolution in a Protoplanetary Disk within Planet Carved Gaps and Dust Rings

Recent surveys of protoplanetary disks show that substructure in dust thermal continuum emission maps is common in protoplanetary disks. These substructures, most prominently rings and gaps, shape and change the chemical and physical conditions of the disk, along with the dust size distributions. In this work, we use a thermochemical code to focus on the chemical evolution that is occurring within the gas-depleted gap and the dust-rich ring often observed behind it. The composition of these spatial locations are of great import, as the gas and ice-coated grains will end up being part of the atmospheres of gas giants and/or the seeds of rocky planets. Our models show that the dust temperature at the midplane of the gap increases, enough to produce local sublimation of key volatiles and pushing the molecular layer closer to the midplane, while it decreases in the dust-rich ring, causing a higher volatile deposition onto the dust grain surfaces. Further, the ring itself presents a freeze-out trap for volatiles in local flows powered by forming planets, becoming a site of localized volatile enhancement. Within the gas depleted gap, the line emission depends on several different parameters, such as: the depth of the gap in surface density, the location of the dust substructure, and the abundance of common gas tracers, such as CO. In order to break this uncertainty between abundance and surface density, other methods such as disk kinematics, become necessary to constrain the disk structure and its chemical evolution.

astro-ph.EP↗

arXiv2020

Visualizing the Kinematics of Planet Formation

A stunning range of substructures in the dust of protoplanetary disks is routinely observed across a range of wavelengths. These gaps, rings and spirals are highly indicative of a population of unseen planets, hinting at the possibility of current observational facilities being able to capture planet-formation in action. Over the last decade, our understanding of the influence of a young planet on the dynamical structure of its parental disk has progressed significantly, revealing a host of potentially observable features which would betray the presence of a deeply embedded planet. In concert, recent observations have shown that subtle perturbations in the kinematic structure of protoplanetary disks are found in multiple sources, potentially the characteristic disturbances associated with embedded planets. In this work, we review the theoretical background of planet-disk interactions, focusing on the kinematical features, and the current methodologies used to observe these interactions in spatially and spectrally resolved observations. We discuss the potential pit falls of such kinematical detections of planets, providing best-practices for imaging and analysing interferometric data, along with a set of criteria to use as a benchmark for any claimed detection of embedded planets. We finish with a discussion on the current state of simulations in regard to planet-disk interactions, highlighting areas of particular interest and future directions which will provide the most significant impact in our search for embedded planets. This work is the culmination of the 'Visualizing the Kinematics of Planet Formation' workshop, held in October 2019 at the Center for Computational Astrophysics at the Flatiron Institute in New York City.

astro-ph.EP↗

arXiv2020

MU-GAN: Facial Attribute Editing based on Multi-attention Mechanism

Facial attribute editing has mainly two objectives: 1) translating image from a source domain to a target one, and 2) only changing the facial regions related to a target attribute and preserving the attribute-excluding details. In this work, we propose a Multi-attention U-Net-based Generative Adversarial Network (MU-GAN). First, we replace a classic convolutional encoder-decoder with a symmetric U-Net-like structure in a generator, and then apply an additive attention mechanism to build attention-based U-Net connections for adaptively transferring encoder representations to complement a decoder with attribute-excluding detail and enhance attribute editing ability. Second, a self-attention mechanism is incorporated into convolutional layers for modeling long-range and multi-level dependencies across image regions. experimental results indicate that our method is capable of balancing attribute editing ability and details preservation ability, and can decouple the correlation among attributes. It outperforms the state-of-the-art methods in terms of attribute manipulation accuracy and image quality.

cs.CV↗

arXiv2020

Half-Magnetic Topological Insulator

Topological magnets are a new family of quantum materials providing great potential to realize emergent phenomena, such as quantum anomalous Hall effect and axion-insulator state. Here we present our discovery that stoichiometric ferromagnet MnBi8Te13 with natural heterostructure MnBi2Te4-(Bi2Te3)3 is an unprecedented half-magnetic topological insulator, with the magnetization existing at the MnBi2Te4 surface but not at the opposite surface terminated by triple Bi2Te3 layers. Our angle-resolved photoemission spectroscopy measurements unveil a massive Dirac gap at the MnBi2Te4 surface, and gapless Dirac cone on the other side. Remarkably, the Dirac gap (~28 meV) at MnBi2Te4 surface decreases monotonically with increasing temperature and closes right at the Curie temperature, thereby representing the first smoking-gun spectroscopic evidence of magnetization-induced topological surface gap among all known magnetic topological materials. We further demonstrate theoretically that the half-magnetic topological insulator is desirable to realize the half-quantized surface anomalous Hall effect, which serves as a direct proof of the general concept of axion electrodynamics in condensed matter systems.

cond-mat.mtrl-sci↗

arXiv2020

CO Depletion in Protoplanetary Disks: A Unified Picture Combining Physical Sequestration and Chemical Processing

The gas-phase CO abundance (relative to hydrogen) in protoplanetary disks decreases by up to 2 orders of magnitude from its ISM value ${\sim}10^{-4}$, even after accounting for freeze-out and photo-dissociation. Previous studies have shown that while local chemical processing of CO and the sequestration of CO ice on solids in the midplane can both contribute, neither of these processes appears capable of consistently reaching the observed depletion factors on the relevant timescale of $1{-}3\mathrm{~Myr}$. In this study, we model these processes simultaneously by including a compact chemical network (centered on carbon and oxygen) to 2D ($r+z$) simulations of the outer ($r>20\mathrm{~au}$) disk regions that include turbulent diffusion, pebble formation, and pebble dynamics. In general, we find that the CO/H$_2$ abundance is a complex function of time and location. Focusing on CO in the warm molecular layer, we find that only the most complete model (with chemistry and pebble evolution included) can reach depletion factors consistent with observations. In the absence of pressure traps, highly-efficient planetesimal formation, or high cosmic ray ionization rates, this model also predicts a resurgence of CO vapor interior to the CO snowline. We show the impact of physical and chemical processes on the elemental (C/O) and (C/H) ratios (in the gas and ice phases), discuss the use of CO as a disk mass tracer, and, finally, connect our predicted pebble ice compositions to those of pristine planetesimals as found in the Cold Classical Kuiper Belt and debris disks.

astro-ph.SR↗