arXiv subjects

Ning Liu

Publications and source records attributed to Ning Liu.

At least 217 records · Page 12Linked to original sources

arXiv2019

A Stochastic-Computing based Deep Learning Framework using Adiabatic Quantum-Flux-Parametron SuperconductingTechnology

The Adiabatic Quantum-Flux-Parametron (AQFP) superconducting technology has been recently developed, which achieves the highest energy efficiency among superconducting logic families, potentially huge gain compared with state-of-the-art CMOS. In 2016, the successful fabrication and testing of AQFP-based circuits with the scale of 83,000 JJs have demonstrated the scalability and potential of implementing large-scale systems using AQFP. As a result, it will be promising for AQFP in high-performance computing and deep space applications, with Deep Neural Network (DNN) inference acceleration as an important example. Besides ultra-high energy efficiency, AQFP exhibits two unique characteristics: the deep pipelining nature since each AQFP logic gate is connected with an AC clock signal, which increases the difficulty to avoid RAW hazards; the second is the unique opportunity of true random number generation (RNG) using a single AQFP buffer, far more efficient than RNG in CMOS. We point out that these two characteristics make AQFP especially compatible with the \emph{stochastic computing} (SC) technique, which uses a time-independent bit sequence for value representation, and is compatible with the deep pipelining nature. Further, the application of SC has been investigated in DNNs in prior work, and the suitability has been illustrated as SC is more compatible with approximate computations. This work is the first to develop an SC-based DNN acceleration framework using AQFP technology.

cs.NE↗

arXiv2019

An Algorithm for Transmitting VR Video Based on Adaptive Modulation

Virtual reality (VR) is making waves around the world recently. However, traditional video streaming is not suitable for VR video because of the huge size and view switch requirements of VR videos. Since the view of each user is limited, it is unnecessary to send the whole 360-degree scene at high quality which can be a heavy burden for the transmission system. Assuming filed-of-view (FoV) of each user can be predicted with high probability, we can divide the video screen into partitions and send those partitions which will appear in FoV at high quality. Hence, we propose an novel strategy for VR video streaming. First, we define a quality-of-experience metric to measure the viewing experience of users and define a channel model to reflect the fluctuation of the wireless channel. Next, we formulate the optimization problem and find its feasible solution by convex optimization. In order to improve bandwidth efficiency, we also add adaptive modulation to this part. Finally, we compare our algorithm with other VR streaming algorithm in the simulation. It turns out that our algorithm outperforms other algorithms.

cs.NI↗

arXiv2019

Anomalous Dome-like Superconductivity in RE2(Cu1-xNix)5As3O2(RE=La, Pr, Nd)

Significant manifestation of interplay of superconductivity and charge density wave, spin density wave or magnetism is dome-like variation in superconducting critical temperature (Tc) for cuprate, iron-based and heavy Fermion superconductors. Overall behavior is that the ordered temperature is gradually suppressed and the Tc is enhanced under external control parameters. Many phenomena like pesudogap, quantum critical point and strange metal emerge in the different doping range. Exploring dome-shaped Tc in new superconductors is of importance to detect emergent effects. Here, we report that the observation of superconductivity in new layered Cu-based compound RE2Cu5As3O2 (RE=La, Pr, Nd), in which the Tc exhibits dome-like variation with maximum Tc of 2.5 K, 1.2 K and 1.0 K as substituting Cu by large amount of Ni ions. The transitions of T* in former two compounds can be suppressed by either Ni doping or rare earth replacement. Simultaneously, the structural parameters like As-As bond length and c/a ratio exhibit unusual variations as Ni-doping level goes through the optimal value. The robustness of superconductivity, up to 60% of Ni doping, reveals the unexpected impurity effect on inducing and enhancing superconductivity in this novel layered materials

cond-mat.supr-con↗

arXiv2019

T-odd top partner pair production in the dilepton final states at the LHC in the littlest Higgs Model with T-parity

In the littlest Higgs Model with T-parity, we discuss the pair production of the T-odd top partner ($T_{-}$) which decays almost 100\% into the top quark and the lightest T-odd particle ($A_{H}$). Considering the current constraints, we investigate the observability of the T-odd top partner pair production through the process $pp\rightarrow T_{-}\bar{T}_{-}\rightarrow t\bar{t}A_{H}A_{H}$ in final states with two leptons at 14 TeV LHC. We analyze the signal significance and find that the lower limit on the T-odd top partner mass are about 1.04 TeV, 1.14 TeV, 1.23 TeV at $2σ$ confidence level at 14TeV LHC with the integrated luminosity of 30 fb$^{-1}$, 100fb$^{-1}$, 300fb$^{-1}$. This lower limit can be raised to about 1.34(1.44) TeV if we use 1000(3000) fb$^{-1}$ of integrated luminosity.

hep-ph↗

arXiv2019

Two-dimensional honeycomb borophene oxide: A promising anode material offering super high capacity for Li/Na-ion batteries

Rational design of novel two-dimensional (2D) electrode materials with high capacity is crucial for the further development of Li-ion and Na-ion batteries. Herein, based on first-principles calculations, we systemically investigate Li and Na storage behaviors in the recently discovered 2D topological nodal-loop metal - the honeycomb borophene oxide (h-B2O). We show that h-B2O is an almost ideal anode material. It has good conductivity before and after Li/Na adsorption, fast ion diffusion with diffusion barrier less than 0.5 eV, low open-circuit voltage (less than 1 V), and small lattice change (less than 6.2%) during intercalation. Most remarkably, its theoretical storage capacity is extremely high, reaching up to 2137 mAh/g for Li and 1425 mAh/g for Na. Its Li storage capacity is more than six times higher than graphite (372 mAh/g), and is actually the highest among all 2D materials discovered to date. Our results strongly suggest that 2D h-B2O is an exceedingly promising anode material for both Li- and Na-ion batteries with super high capacity.

cond-mat.mtrl-sci↗

arXiv2019

Single production of vector-like top partner decaying to $Wb$ in the leptonic channel at $ep$ colliders in the LHT model

In the littlest Higgs model with T-parity(LHT), we study the single production of vector-like top partner with the subsequent decay $T_{+}\to Wb$ in the leptonic channel at the $ep$ colliders. Focus on the LHeC ($ \sqrt{s} $ = 1.98 TeV) and FCC-eh ($ \sqrt{s} $ = 5.29 TeV), we investigate the observability of the single top partner production with the unpolarized and polarized electron beams, respectively. As a result, the statistical significance can be enhanced by the polarized electron beams. Under the current constraints, the search for $T_{+}$ in the $Wb$ channel at the LHeC cannot provide a stronger limit on the top partner mass. By contrast, the search for the $T_{+}$ in this channel at the FCC-eh with polarized $ e^- $ beams can exclude the top partner mass up to 1350 GeV, 1500 GeV and 1565 GeV with integrated luminosities of 100 fb$^{-1}$, 1000 fb$^{-1}$ and 3000 fb$^{-1}$ at the 2$σ$ level, which is an improvement with respect to the current indirect searches and the LHC direct searches. Furthermore, we also give an extrapolation to the high-luminosity LHC with $\sqrt{s}=14$ TeV and $ L=3000~\rm{fb}^{-1} $. Our results show that the LHT model is still a natural solution to the shortcomings of the electroweak and scalar sector although it has been constrained severely.

hep-ph↗

arXiv2019

ADCrowdNet: An Attention-injective Deformable Convolutional Network for Crowd Understanding

We propose an attention-injective deformable convolutional network called ADCrowdNet for crowd understanding that can address the accuracy degradation problem of highly congested noisy scenes. ADCrowdNet contains two concatenated networks. An attention-aware network called Attention Map Generator (AMG) first detects crowd regions in images and computes the congestion degree of these regions. Based on detected crowd regions and congestion priors, a multi-scale deformable network called Density Map Estimator (DME) then generates high-quality density maps. With the attention-aware training scheme and multi-scale deformable convolutional scheme, the proposed ADCrowdNet achieves the capability of being more effective to capture the crowd features and more resistant to various noises. We have evaluated our method on four popular crowd counting datasets (ShanghaiTech, UCF_CC_50, WorldEXPO'10, and UCSD) and an extra vehicle counting dataset TRANCOS, and our approach beats existing state-of-the-art approaches on all of these datasets.

cs.CV↗

arXiv2019

StructADMM: A Systematic, High-Efficiency Framework of Structured Weight Pruning for DNNs

Weight pruning methods of DNNs have been demonstrated to achieve a good model pruning rate without loss of accuracy, thereby alleviating the significant computation/storage requirements of large-scale DNNs. Structured weight pruning methods have been proposed to overcome the limitation of irregular network structure and demonstrated actual GPU acceleration. However, in prior work the pruning rate (degree of sparsity) and GPU acceleration are limited (to less than 50%) when accuracy needs to be maintained. In this work,we overcome these limitations by proposing a unified, systematic framework of structured weight pruning for DNNs. It is a framework that can be used to induce different types of structured sparsity, such as filter-wise, channel-wise, and shape-wise sparsity, as well non-structured sparsity. The proposed framework incorporates stochastic gradient descent with ADMM, and can be understood as a dynamic regularization method in which the regularization target is analytically updated in each iteration. Without loss of accuracy on the AlexNet model, we achieve 2.58X and 3.65X average measured speedup on two GPUs, clearly outperforming the prior work. The average speedups reach 3.15X and 8.52X when allowing a moderate ac-curacy loss of 2%. In this case the model compression for convolutional layers is 15.0X, corresponding to 11.93X measured CPU speedup. Our experiments on ResNet model and on other data sets like UCF101 and CIFAR-10 demonstrate the consistently higher performance of our framework.

cs.NE↗

arXiv2019

Polarization of top quark in vector-like quark decay

Vector-like quarks (VLQs) are attractive extensions to the Standard Model. They mix with the SM quarks and can lead to rich phenomenology. Determination of VLQ's interaction structure with the SM is then an important issue, which can be inferred from the decay products of VLQs, such as top quark. We calculate the spin-analyzing powers for charged leptons from top quark in VLQ decay for various VLQ scenarios. We find that the top polarization effect will be helpful to distinguish different natures of the VLQ couplings with the SM.

hep-ph↗

arXiv2019

Delay Outage Probability of Multi-relay Selection for Mobile Relay Edge Computing Systems

In this paper, we deal with the problem of relay selection in mobile edge computing networks, where a source node transmits a computation-intensive task to a destination via the aid of multiple relay nodes. It differs from the traditional relay nodes in the way that each relay node is equipped with an edge computing server, thus each relay node can execute the received task and forwards the computed result to the destination. Accordingly, we define a delay outage probability to evaluate the impact of the relay computing ability on the link communications, and then propose a latency-best relay selection (LBRS) scheme that not only consider the communication capability, but also consider the computing ability. The performance of the proposed relay selection scheme with the traditional communication-only relay selection (CORS) and computing-only selection (CPORS) schemes in terms of the delay outage probability and the diversity order is analyzed, compared with other relay selection schemes. We show that the proposed LBRS scheme reduces to the traditional CORS scheme under the high signal-to-noise ratio (SNR) region. We further reveal that the diversity orders of both the proposed LBRS and CORS schemes are dependent on the computing ability of relay nodes.

cs.IT↗

arXiv2018

Three-dimensional honeycomb carbon: Junction line distortion and novel emergent fermions

Carbon enjoys a vast number of allotropic forms, each possessing unique properties determined by the lattice structures and bonding characters. Here, based on first-principles calculations, we propose a new three-dimensional carbon allotrope--hC28. We show that hC28 possesses exceptional energetic, dynamical, thermal, and mechanical stability. It is energetically more stable than most other synthesized or proposed carbon allotropes. The material has a relatively small bulk modulus, but is thermally stable at temperatures as high as 2000 K. The structural, mechanical, x-ray diffraction, and electronic properties are systematically investigated. Particularly, we show that its low-energy band structure hosts multiple unconventional emergent fermions, including the quadratic-contact-point fermions, the birefringent Dirac fermions, and the triple-point fermions. We construct effective models to characterize each kind of fermions. Our work not only discovers a new carbon allotropic form, it also reveals remarkable mechanical and electronic properties for this new material, which may pave the way towards both fundamental studies as well as practical applications.

physics.comp-ph↗

arXiv2018

Structured Weight Matrices-Based Hardware Accelerators in Deep Neural Networks: FPGAs and ASICs

Both industry and academia have extensively investigated hardware accelerations. In this work, to address the increasing demands in computational capability and memory requirement, we propose structured weight matrices (SWM)-based compression techniques for both \emph{field programmable gate array} (FPGA) and \emph{application-specific integrated circuit} (ASIC) implementations. In algorithm part, SWM-based framework adopts block-circulant matrices to achieve a fine-grained tradeoff between accuracy and compression ratio. The SWM-based technique can reduce computational complexity from O($n^2$) to O($n\log n$) and storage complexity from O($n^2$) to O($n$) for each layer and both training and inference phases. For FPGA implementations on deep convolutional neural networks (DCNNs), we achieve at least 152X and 72X improvement in performance and energy efficiency, respectively using the SWM-based framework, compared with the baseline of IBM TrueNorth processor under same accuracy constraints using the data set of MNIST, SVHN, and CIFAR-10. For FPGA implementations on long short term memory (LSTM) networks, the proposed SWM-based LSTM can achieve up to 21X enhancement in performance and 33.5X gains in energy efficiency compared with the baseline accelerator. For ASIC implementations, the SWM-based ASIC design exhibits impressive advantages in terms of power, throughput, and energy efficiency. Experimental results indicate that this method is greatly suitable for applying DNNs onto both FPGAs and mobile/IoT devices.

cs.DC↗

arXiv2018

VIBNN: Hardware Acceleration of Bayesian Neural Networks

Bayesian Neural Networks (BNNs) have been proposed to address the problem of model uncertainty in training and inference. By introducing weights associated with conditioned probability distributions, BNNs are capable of resolving the overfitting issue commonly seen in conventional neural networks and allow for small-data training, through the variational inference process. Frequent usage of Gaussian random variables in this process requires a properly optimized Gaussian Random Number Generator (GRNG). The high hardware cost of conventional GRNG makes the hardware implementation of BNNs challenging. In this paper, we propose VIBNN, an FPGA-based hardware accelerator design for variational inference on BNNs. We explore the design space for massive amount of Gaussian variable sampling tasks in BNNs. Specifically, we introduce two high performance Gaussian (pseudo) random number generators: the RAM-based Linear Feedback Gaussian Random Number Generator (RLF-GRNG), which is inspired by the properties of binomial distribution and linear feedback logics; and the Bayesian Neural Network-oriented Wallace Gaussian Random Number Generator. To achieve high scalability and efficient memory access, we propose a deep pipelined accelerator architecture with fast execution and good hardware utilization. Experimental results demonstrate that the proposed VIBNN implementations on an FPGA can achieve throughput of 321,543.4 Images/s and energy efficiency upto 52,694.8 Images/J while maintaining similar accuracy as its software counterpart.

cs.LG↗

arXiv2018

Deep Reinforcement Learning for Dynamic Treatment Regimes on Medical Registry Data

This paper presents the first deep reinforcement learning (DRL) framework to estimate the optimal Dynamic Treatment Regimes from observational medical data. This framework is more flexible and adaptive for high dimensional action and state spaces than existing reinforcement learning methods to model real-life complexity in heterogeneous disease progression and treatment choices, with the goal of providing doctor and patients the data-driven personalized decision recommendations. The proposed DRL framework comprises (i) a supervised learning step to predict the most possible expert actions, and (ii) a deep reinforcement learning step to estimate the long-term value function of Dynamic Treatment Regimes. Both steps depend on deep neural networks. As a key motivational example, we have implemented the proposed framework on a data set from the Center for International Bone Marrow Transplant Research (CIBMTR) registry database, focusing on the sequence of prevention and treatments for acute and chronic graft versus host disease after transplantation. In the experimental results, we have demonstrated promising accuracy in predicting human experts' decisions, as well as the high expected reward function in the DRL-based dynamic treatment regimes.

cs.AI↗

arXiv2017

FFT-Based Deep Learning Deployment in Embedded Systems

Deep learning has delivered its powerfulness in many application domains, especially in image and speech recognition. As the backbone of deep learning, deep neural networks (DNNs) consist of multiple layers of various types with hundreds to thousands of neurons. Embedded platforms are now becoming essential for deep learning deployment due to their portability, versatility, and energy efficiency. The large model size of DNNs, while providing excellent accuracy, also burdens the embedded platforms with intensive computation and storage. Researchers have investigated on reducing DNN model size with negligible accuracy loss. This work proposes a Fast Fourier Transform (FFT)-based DNN training and inference model suitable for embedded platforms with reduced asymptotic complexity of both computation and storage, making our approach distinguished from existing approaches. We develop the training and inference algorithms based on FFT as the computing kernel and deploy the FFT-based inference model on embedded platforms achieving extraordinary processing speed.

cs.LG↗

arXiv2017

An indirect probe of higgsino world at the CEPC

The higgsino world is one of the popular natural SUSY scenarios, in which $|μ| \ll M_{gauginos} \ll M_{scalars}$. As such, searching for light degenerate higgsinos becomes an essential task of testing naturalness in supersymmetry (SUSY). In this work, we study the indirect effects of light higgsinos in the process $e^+e^- \to W^+W^-$ at future Higgs factories, such as Circular Electron Positron Collider (CEPC) in China. we find that the higgsino mass parameter $μ\lesssim 210$ GeV favored by naturalness can be covered if the accuracy of the measurement of $W^+W^-$ production can reach 0.1\% at 240 GeV CEPC.

hep-ph↗

arXiv2017

CirCNN: Accelerating and Compressing Deep Neural Networks Using Block-CirculantWeight Matrices

Large-scale deep neural networks (DNNs) are both compute and memory intensive. As the size of DNNs continues to grow, it is critical to improve the energy efficiency and performance while maintaining accuracy. For DNNs, the model size is an important factor affecting performance, scalability and energy efficiency. Weight pruning achieves good compression ratios but suffers from three drawbacks: 1) the irregular network structure after pruning; 2) the increased training complexity; and 3) the lack of rigorous guarantee of compression ratio and inference accuracy. To overcome these limitations, this paper proposes CirCNN, a principled approach to represent weights and process neural networks using block-circulant matrices. CirCNN utilizes the Fast Fourier Transform (FFT)-based fast multiplication, simultaneously reducing the computational complexity (both in inference and training) from O(n2) to O(nlogn) and the storage complexity from O(n2) to O(n), with negligible accuracy loss. Compared to other approaches, CirCNN is distinct due to its mathematical rigor: it can converge to the same effectiveness as DNNs without compression. The CirCNN architecture, a universal DNN inference engine that can be implemented on various hardware/software platforms with configurable network architecture. To demonstrate the performance and energy efficiency, we test CirCNN in FPGA, ASIC and embedded processors. Our results show that CirCNN architecture achieves very high energy efficiency and performance with a small hardware footprint. Based on the FPGA implementation and ASIC synthesis results, CirCNN achieves 6-102X energy efficiency improvements compared with the best state-of-the-art results.

cs.CV↗

arXiv2017

A Hierarchical Framework of Cloud Resource Allocation and Power Management Using Deep Reinforcement Learning

Automatic decision-making approaches, such as reinforcement learning (RL), have been applied to (partially) solve the resource allocation problem adaptively in the cloud computing system. However, a complete cloud resource allocation framework exhibits high dimensions in state and action spaces, which prohibit the usefulness of traditional RL techniques. In addition, high power consumption has become one of the critical concerns in design and control of cloud computing systems, which degrades system reliability and increases cooling cost. An effective dynamic power management (DPM) policy should minimize power consumption while maintaining performance degradation within an acceptable level. Thus, a joint virtual machine (VM) resource allocation and power management framework is critical to the overall cloud computing system. Moreover, novel solution framework is necessary to address the even higher dimensions in state and action spaces. In this paper, we propose a novel hierarchical framework for solving the overall resource allocation and power management problem in cloud computing systems. The proposed hierarchical framework comprises a global tier for VM resource allocation to the servers and a local tier for distributed power management of local servers. The emerging deep reinforcement learning (DRL) technique, which can deal with complicated control problems with large state space, is adopted to solve the global tier problem. Furthermore, an autoencoder and a novel weight sharing structure are adopted to handle the high-dimensional state space and accelerate the convergence speed. On the other hand, the local tier of distributed server power managements comprises an LSTM based workload predictor and a model-free RL based power manager, operating in a distributed manner.

cs.DC↗