arXiv subjects

Hao Yu

Publications and source records attributed to Hao Yu.

At least 271 records · Page 15Linked to original sources

arXiv2018

Using Complex Network Theory for Temporal Locality in Network Traffic Flows

Monitoring the interaction behaviors of network traffic flows and detecting unwanted Internet applications and anomalous flows have become a challenging problem, since many applications obfuscate their network traffic flow using unregistered port numbers or payload encryption. In this paper, the temporal locality complex network model--TLCN is proposed as a way to monitor, analyze and visualize network traffic flows. TLCNs model the interaction behaviors of large-scale network traffic flows, where the nodes and the edges can be defined to represent different flow levels and flow interactions separately. Then, the statistical characteristics and dynamic behaviors of the TLCNs are studied to represent TLCN's structure representing ability to the flow interactions. According to the analysis of TLCN statistical characteristics with different Internet applications, we found that the weak interaction flows prefer to form the small-world TLCN and the strong interaction flows prefer to the scale-free TLCN. In the studies of anomaly behaviors of TLCNs, the network structure of attacked TLCNs can have a remarkable feature for three attack patterns, and the evolution of TLCNs exhibits a good consistency between TLCN structure and attack process. With the introduction of TLCNs, we are able to harness a wealth of tools and graph modeling techniques from a diverse set of disciplines.

cs.NI↗

arXiv2018

DEEPEYE: A Compact and Accurate Video Comprehension at Terminal Devices Compressed with Quantization and Tensorization

As it requires a huge number of parameters when exposed to high dimensional inputs in video detection and classification, there is a grand challenge to develop a compact yet accurate video comprehension at terminal devices. Current works focus on optimizations of video detection and classification in a separated fashion. In this paper, we introduce a video comprehension (object detection and action recognition) system for terminal devices, namely DEEPEYE. Based on You Only Look Once (YOLO), we have developed an 8-bit quantization method when training YOLO; and also developed a tensorized-compression method of Recurrent Neural Network (RNN) composed of features extracted from YOLO. The developed quantization and tensorization can significantly compress the original network model yet with maintained accuracy. Using the challenging video datasets: MOMENTS and UCF11 as benchmarks, the results show that the proposed DEEPEYE achieves 3.994x model compression rate with only 0.47% mAP decreased; and 15,047x parameter reduction and 2.87x speed-up with 16.58% accuracy improvement.

cs.CV↗

arXiv2018

Fast K-Means Clustering with Anderson Acceleration

We propose a novel method to accelerate Lloyd's algorithm for K-Means clustering. Unlike previous acceleration approaches that reduce computational cost per iterations or improve initialization, our approach is focused on reducing the number of iterations required for convergence. This is achieved by treating the assignment step and the update step of Lloyd's algorithm as a fixed-point iteration, and applying Anderson acceleration, a well-established technique for accelerating fixed-point solvers. Classical Anderson acceleration utilizes m previous iterates to find an accelerated iterate, and its performance on K-Means clustering can be sensitive to choice of m and the distribution of samples. We propose a new strategy to dynamically adjust the value of m, which achieves robust and consistent speedups across different problem instances. Our method complements existing acceleration techniques, and can be combined with them to achieve state-of-the-art performance. We perform extensive experiments to evaluate the performance of the proposed method, where it outperforms other algorithms in 106 out of 120 test cases, and the mean decrease ratio of computational time is more than 33%.

cs.LG↗

arXiv2018

Loss Rank Mining: A General Hard Example Mining Method for Real-time Detectors

Modern object detectors usually suffer from low accuracy issues, as foregrounds always drown in tons of backgrounds and become hard examples during training. Compared with those proposal-based ones, real-time detectors are in far more serious trouble since they renounce the use of region-proposing stage which is used to filter a majority of backgrounds for achieving real-time rates. Though foregrounds as hard examples are in urgent need of being mined from tons of backgrounds, a considerable number of state-of-the-art real-time detectors, like YOLO series, have yet to profit from existing hard example mining methods, as using these methods need detectors fit series of prerequisites. In this paper, we propose a general hard example mining method named Loss Rank Mining (LRM) to fill the gap. LRM is a general method for real-time detectors, as it utilizes the final feature map which exists in all real-time detectors to mine hard examples. By using LRM, some elements representing easy examples in final feature map are filtered and detectors are forced to concentrate on hard examples during training. Extensive experiments validate the effectiveness of our method. With our method, the improvements of YOLOv2 detector on auto-driving related dataset KITTI and more general dataset PASCAL VOC are over 5% and 2% mAP, respectively. In addition, LRM is the first hard example mining strategy which could fit YOLOv2 perfectly and make it better applied in series of real scenarios where both real-time rates and accurate detection are strongly demanded.

cs.CV↗

arXiv2018

Dynamics of macrospin under periodic field and spin transfer torque

The dynamics of a macrospin model for a single magnetic domain is investigated in two cases: (i) under the action of a periodic magnetic field and (ii) under the external field plus a spin transfer torque from spin-polarized current or spin current. It has been proved that (i) without spin transfer torque, the trajectory of magnetization (hysteresis) is always a closed curve oscillating between two stable points and following the same path each time; (ii) under the action of a constant field (or with small periodic perturbation) and a constant spin transfer torque, the spin would always turn to the stable solution finally, which is independent of the parameters or initial conditions; (iii) considering a periodic oscillating field plus a spin transfer torque, the system can also have a stable solution around a limit circle, but on some special points on the attractor, the system may be sensitive to the initial conditions and have no limit circle.

cond-mat.mes-hall↗

arXiv2018

Merging and Evolution: Improving Convolutional Neural Networks for Mobile Applications

Compact neural networks are inclined to exploit "sparsely-connected" convolutions such as depthwise convolution and group convolution for employment in mobile applications. Compared with standard "fully-connected" convolutions, these convolutions are more computationally economical. However, "sparsely-connected" convolutions block the inter-group information exchange, which induces severe performance degradation. To address this issue, we present two novel operations named merging and evolution to leverage the inter-group information. Our key idea is encoding the inter-group information with a narrow feature map, then combining the generated features with the original network for better representation. Taking advantage of the proposed operations, we then introduce the Merging-and-Evolution (ME) module, an architectural unit specifically designed for compact networks. Finally, we propose a family of compact neural networks called MENet based on ME modules. Extensive experiments on ILSVRC 2012 dataset and PASCAL VOC 2007 dataset demonstrate that MENet consistently outperforms other state-of-the-art compact networks under different computational budgets. For instance, under the computational budget of 140 MFLOPs, MENet surpasses ShuffleNet by 1% and MobileNet by 1.95% on ILSVRC 2012 top-1 accuracy, while by 2.3% and 4.1% on PASCAL VOC 2007 mAP, respectively.

cs.CV↗

arXiv2018

Thermoelectric properties of polycrystalline palladium sulfide

A suite measurements of the electrical, thermal, and vibrational properties are conducted on palladium sulfide (PdS) in order to investigate its thermoelectric performance. The tetragonal structure with the space group $P$42/$m$ for PdS is determined from X-ray diffraction measurement. The unique temperature dependence of mobility suggests that acoustic phonons and ion impurity scattering are two dominant scattering mechanisms within the compound. The obtained power factor of $27$ $μ$Wcm$^{-1}$K$^{-2}$ at 800 K is the largest value in the remaining transition-metal sulfides studied so far. The maximum value of the dimensionless figure of merit is 0.33 at 800 K. The observed phonon softening with temperature indicates that the reduction of the lattice thermal conductivity is mainly controlled by the enhanced lattice anharmonicity. These results indicate that the binary bulk PdS has promising potential to have good thermoelectrical performance.

cond-mat.mtrl-sci↗

arXiv2018

Origin of efficient thermoelectric performance in half-Heusler FeNb$_{0.8}$Ti$_{0.2}$Sb

A half-Heusler material FeNb$_{0.8}$Ti$_{0.2}$Sb has been identified as a promising thermoelectric material due to its excellent thermoelectric performance at high temperatures. The origins of the efficient thermoelectric performance are investigated through a series of low-temperature (2 - 400 K) measurements. The high data coherence of the low and high temperatures is observed. An optimal and nearly temperature-independent carrier concentration is identified, which is ideal for the power factor. The obtained single type of hole carrier is also beneficial to the large Seebeck coefficient. The electronic thermal conductivity is found to be comparable to the lattice thermal conductivity and becomes the dominant component above 200 K. These findings again indicate that electron scattering plays a key role in the electrical and thermal transport properties. The dimensionless figure of merit is thus mainly governed by the electronic properties. These effects obtained at low temperatures with the avoidance of possible thermal fluctuations together offer the physical origin for the excellent thermoelectric performance in this material.

cond-mat.mtrl-sci↗

arXiv2018

Pressure-induced superconductivity in palladium sulfide

An extended study on PdS is carried out with the measurements of the resistivity, Hall coefficient, Raman scattering, and X-ray diffraction at high pressures up to 42.3 GPa. With increasing pressure, superconductivity is observed accompanying with a structural phase transition at around 19.5 GPa. The coexistence of semiconducting and metallic phases observed at normal state is examined by the Raman scattering and X-ray diffraction between 19.5 and 29.5 GPa. After that, only the metallic normal state maintains with an almost constant superconducting transition temperature. The similar evolution between the superconducting transition temperature and carrier concentration with pressure supports the phonon-mediated superconductivity in this material. These results highlight the important role of pressure played in inducing superconductivity from these narrow band-gap semiconductors.

cond-mat.mtrl-sci↗

arXiv2018

Scalar particle production in a simple Horndeski theory

The scalar particle production through a scalar field non-minimally coupled with geometry is investigated in the context of a spatially homogeneous and isotropic universe. In this paper, in order to study the evolution of particle production over time in the case of analytical solutions, we focus on a simple Horndeski theory. We first suppose that the universe is dominated by a scalar field and derive the energy conservation condition. Then from the thermodynamic point of view, the macroscopic non-conservation of the scalar field energy-momentum tensor can be explained as an irreversible production of the scalar particles. Based on the explanation, we obtain a scalar particle production rate and the corresponding entropy. Finally, since the universe, in general, could be regarded as a closed system satisfying the laws of thermodynamics, we naturally impose some thermodynamic constraints on it. The thermodynamic properties of the universe can provide additional constraints on the simple Horndeski theory.

gr-qc↗

arXiv2017

Online Learning in Weakly Coupled Markov Decision Processes: A Convergence Time Study

We consider multiple parallel Markov decision processes (MDPs) coupled by global constraints, where the time varying objective and constraint functions can only be observed after the decision is made. Special attention is given to how well the decision maker can perform in $T$ slots, starting from any state, compared to the best feasible randomized stationary policy in hindsight. We develop a new distributed online algorithm where each MDP makes its own decision each slot after observing a multiplier computed from past information. While the scenario is significantly more challenging than the classical online learning context, the algorithm is shown to have a tight $O(\sqrt{T})$ regret and constraint violations simultaneously. To obtain such a bound, we combine several new ingredients including ergodicity and mixing time bound in weakly coupled MDPs, a new regret analysis for online constrained optimization, a drift analysis for queue processes, and a perturbation analysis based on Farkas' Lemma.

math.OC↗

arXiv2017

S-OHEM: Stratified Online Hard Example Mining for Object Detection

One of the major challenges in object detection is to propose detectors with highly accurate localization of objects. The online sampling of high-loss region proposals (hard examples) uses the multitask loss with equal weight settings across all loss types (e.g, classification and localization, rigid and non-rigid categories) and ignores the influence of different loss distributions throughout the training process, which we find essential to the training efficacy. In this paper, we present the Stratified Online Hard Example Mining (S-OHEM) algorithm for training higher efficiency and accuracy detectors. S-OHEM exploits OHEM with stratified sampling, a widely-adopted sampling technique, to choose the training examples according to this influence during hard example mining, and thus enhance the performance of object detectors. We show through systematic experiments that S-OHEM yields an average precision (AP) improvement of 0.5% on rigid categories of PASCAL VOC 2007 for both the IoU threshold of 0.6 and 0.7. For KITTI 2012, both results of the same metric are 1.6%. Regarding the mean average precision (mAP), a relative increase of 0.3% and 0.5% (1% and 0.5%) is observed for VOC07 (KITTI12) using the same set of IoU threshold. Also, S-OHEM is easy to integrate with existing region-based detectors and is capable of acting with post-recognition level regressors.

cs.CV↗

arXiv2017

Online Convex Optimization with Stochastic Constraints

This paper considers online convex optimization (OCO) with stochastic constraints, which generalizes Zinkevich's OCO over a known simple fixed set by introducing multiple stochastic functional constraints that are i.i.d. generated at each round and are disclosed to the decision maker only after the decision is made. This formulation arises naturally when decisions are restricted by stochastic environments or deterministic environments with noisy observations. It also includes many important problems as special cases, such as OCO with long term constraints, stochastic constrained convex optimization, and deterministic constrained convex optimization. To solve this problem, this paper proposes a new algorithm that achieves $O(\sqrt{T})$ expected regret and constraint violations and $O(\sqrt{T}\log(T))$ high probability regret and constraint violations. Experiments on a real-world data center scheduling problem further verify the performance of the new algorithm.

math.OC↗

arXiv2017

On the Convergence Time of Dual Subgradient Methods for Strongly Convex Programs

This paper studies the convergence time of dual gradient methods for general (possibly non-differentiable) strongly convex programs. For general convex programs, the convergence time of dual subgradient/gradient methods with simple running averages (running averages started from iteration $0$) is known to be $O(\frac{1}{ε^{2}})$. This paper shows that the convergence time for general strongly convex programs is $O(\frac{1}ε)$. This paper also considers a variation of the average scheme, called the sliding running averages, and shows that if the dual function of the strongly convex program is locally quadratic (Note that the locally quadratic property is implied by the locally strongly concave property.) then the convergence time of the dual gradient method with sliding running averages is $O(\log(\frac{1}ε))$. The convergence time analysis is further verified by numerical experiments.

math.OC↗

arXiv2017

A Simple Parallel Algorithm with an $O(1/t)$ Convergence Rate for General Convex Programs

This paper considers convex programs with a general (possibly non-differentiable) convex objective function and Lipschitz continuous convex inequality constraint functions. A simple algorithm is developed and achieves an $O(1/t)$ convergence rate. Similar to the classical dual subgradient algorithm and the ADMM algorithm, the new algorithm has a parallel implementation when the objective and constraint functions are separable. However, the new algorithm has a faster $O(1/t)$ convergence rate compared with the best known $O(1/\sqrt{t})$ convergence rate for the dual subgradient algorithm with primal averaging. Further, it can solve convex programs with nonlinear constraints, which cannot be handled by the ADMM algorithm. The new algorithm is applied to a multipath network utility maximization problem and yields a decentralized flow control algorithm with the fast $O(1/t)$ convergence rate.

math.OC↗

arXiv2017

A Primal-Dual Parallel Method with $O(1/ε)$ Convergence for Constrained Composite Convex Programs

This paper considers large scale constrained convex (possibly composite and non-separable) programs, which are usually difficult to solve by interior point methods or other Newton-type methods due to the non-smoothness or the prohibitive computation and storage complexity for Hessians and matrix inversions. Instead, they are often solved by first order gradient based methods or decomposition based methods. The conventional primal-dual subgradient method, also known as the Arrow-Hurwicz-Uzawa subgradient method, is a low complexity algorithm with an $O(1/ε^2)$ convergence time. Recently, a new Lagrangian dual type algorithm with a faster $O(1/ε)$ convergence time is proposed in Yu and Neely (2017). However, if the objective or constraint functions are not separable, each iteration of the Lagrangian dual type method in Yu and Neely (2017) requires to solve a unconstrained convex program, which can have huge complexity. This paper proposes a new primal-dual type algorithm with $O(1/ε)$ convergence for general constrained convex programs. Each iteration of the new algorithm can be implemented in parallel with low complexity even when the original problem is composite and non-separable.

math.OC↗

arXiv2017

A GPU-Outperforming FPGA Accelerator Architecture for Binary Convolutional Neural Networks

FPGA-based hardware accelerators for convolutional neural networks (CNNs) have obtained great attentions due to their higher energy efficiency than GPUs. However, it is challenging for FPGA-based solutions to achieve a higher throughput than GPU counterparts. In this paper, we demonstrate that FPGA acceleration can be a superior solution in terms of both throughput and energy efficiency when a CNN is trained with binary constraints on weights and activations. Specifically, we propose an optimized FPGA accelerator architecture tailored for bitwise convolution and normalization that features massive spatial parallelism with deep pipelines stages. A key advantage of the FPGA accelerator is that its performance is insensitive to data batch size, while the performance of GPU acceleration varies largely depending on the batch size of the data. Experiment results show that the proposed accelerator architecture for binary CNNs running on a Virtex-7 FPGA is 8.3x faster and 75x more energy-efficient than a Titan X GPU for processing online individual requests in small batch sizes. For processing static data in large batch sizes, the proposed solution is on a par with a Titan X GPU in terms of throughput while delivering 9.5x higher energy efficiency.

cs.DC↗

arXiv2017

Dynamic Transmit Covariance Design in MIMO Fading Systems With Unknown Channel Distributions and Inaccurate Channel State Information

This paper considers dynamic transmit covariance design in point-to-point MIMO fading systems with unknown channel state distributions and inaccurate channel state information subject to both long term and short term power constraints. First, the case of instantaneous but possibly inaccurate channel state information at the transmitter (CSIT) is treated. By extending the drift-plus-penalty technique, a dynamic transmit covariance policy is developed and is shown to approach optimality with an $O(δ)$ gap, where $δ$ is the inaccuracy measure of CSIT, regardless of the channel state distribution and without requiring knowledge of this distribution. Next, the case of delayed and inaccurate channel state information is considered. The optimal transmit covariance solution that maximizes the ergodic capacity is fundamentally different in this case, and a different online algorithm based on convex projections is developed. The proposed algorithm for this delayed-CSIT case also has an $O(δ)$ optimality gap, where $δ$ is again the inaccuracy measure of CSIT.

cs.IT↗