arXiv subjects

Gang Wang

Publications and source records attributed to Gang Wang.

At least 343 records · Page 19Linked to original sources

arXiv2020

SparTerm: Learning Term-based Sparse Representation for Fast Text Retrieval

Term-based sparse representations dominate the first-stage text retrieval in industrial applications, due to its advantage in efficiency, interpretability, and exact term matching. In this paper, we study the problem of transferring the deep knowledge of the pre-trained language model (PLM) to Term-based Sparse representations, aiming to improve the representation capacity of bag-of-words(BoW) method for semantic-level matching, while still keeping its advantages. Specifically, we propose a novel framework SparTerm to directly learn sparse text representations in the full vocabulary space. The proposed SparTerm comprises an importance predictor to predict the importance for each term in the vocabulary, and a gating controller to control the term activation. These two modules cooperatively ensure the sparsity and flexibility of the final text representation, which unifies the term-weighting and expansion in the same framework. Evaluated on MSMARCO dataset, SparTerm significantly outperforms traditional sparse methods and achieves state of the art ranking performance among all the PLM-based sparse models.

cs.IR↗

arXiv2020

PDLight: A Deep Reinforcement Learning Traffic Light Control Algorithm with Pressure and Dynamic Light Duration

Existing ineffective and inflexible traffic light control at urban intersections can often lead to congestion in traffic flows and cause numerous problems, such as long delay and waste of energy. How to find the optimal signal timing strategy is a significant challenge in urban traffic management. In this paper, we propose PDlight, a deep reinforcement learning (DRL) traffic light control algorithm with a novel reward as PRCOL (Pressure with Remaining Capacity of Outgoing Lane). Serving as an improvement over the pressure used in traffic control algorithms, PRCOL considers not only the number of vehicles on the incoming lane but also the remaining capacity of the outgoing lane. Simulation results using both synthetic and real-world data-sets show that the proposed PDlight yields lower average travel time compared with several state-of-the-art algorithms, PressLight and Colight, under both fixed and dynamic green light duration.

cs.LG↗

arXiv2020

A Traffic Light Dynamic Control Algorithm with Deep Reinforcement Learning Based on GNN Prediction

Today's intelligent traffic light control system is based on the current road traffic conditions for traffic regulation. However, these approaches cannot exploit the future traffic information in advance. In this paper, we propose GPlight, a deep reinforcement learning (DRL) algorithm integrated with graph neural network (GNN) , to relieve the traffic congestion for multi-intersection intelligent traffic control system. In GPlight, the graph neural network (GNN) is first used to predict the future short-term traffic flow at the intersections. Then, the results of traffic flow prediction are used in traffic light control, and the agent combines the predicted results with the observed current traffic conditions to dynamically control the phase and duration of the traffic lights at the intersection. Experiments on both synthetic and two real-world data-sets of Hangzhou and New-York verify the effectiveness and rationality of the GPlight algorithm.

cs.LG↗

arXiv2020

On the clustering properties of produced particles in high-energy $pp$ collisions

Minijets provide useful information on parton interactions in the low transverse-momentum (low-$p_T$) region. Because minijets produce clusters, we study the clustering properties of produced particles in high-energy $pp$ collisions as a first step to identify minijets. We develop an algorithm to find clusters by using the k-means clustering method, in conjunction with a k-number (cluster number) selection principle in the space of pseudorapidity and azimuthal angles. We test the clustering algorithm using events generated by PYTHIA 8.1, for $pp$ collision at $\sqrt{s}=200$ GeV. We find that clustering of low-$p_T$ hadrons occurs in high multiplicity events. However similar clustering properties are also present for particles produced randomly in a finite pseudorapidity and azimuthal angle space. To distinguish the dynamics from random generations of events, it is necessary to examine the correlation between particles and between clusters. We find that the correlations between clusters may provide a useful tool to distinguish the underlying dynamics of the reaction mechanism.

hep-ph↗

arXiv2020

A Multistep Lyapunov Approach for Finite-Time Analysis of Biased Stochastic Approximation

Motivated by the widespread use of temporal-difference (TD-) and Q-learning algorithms in reinforcement learning, this paper studies a class of biased stochastic approximation (SA) procedures under a mild "ergodic-like" assumption on the underlying stochastic noise sequence. Building upon a carefully designed multistep Lyapunov function that looks ahead to several future updates to accommodate the stochastic perturbations (for control of the gradient bias), we prove a general result on the convergence of the iterates, and use it to derive non-asymptotic bounds on the mean-square error in the case of constant stepsizes. This novel looking-ahead viewpoint renders finite-time analysis of biased SA algorithms under a large family of stochastic perturbations possible. For direct comparison with existing contributions, we also demonstrate these bounds by applying them to TD- and Q-learning with linear function approximation, under the practical Markov chain observation model. The resultant finite-time error bound for both the TD- as well as the Q-learning algorithms is the first of its kind, in the sense that it holds i) for the unmodified versions (i.e., without making any modifications to the parameter updates) using even nonlinear function approximators; as well as for Markov chains ii) under general mixing conditions and iii) starting from any initial distribution, at least one of which has to be violated for existing results to be applicable.

stat.ML↗

arXiv2020

Urban Sensing based on Mobile Phone Data: Approaches, Applications and Challenges

Data volume grows explosively with the proliferation of powerful smartphones and innovative mobile applications. The ability to accurately and extensively monitor and analyze these data is necessary. Much concern in mobile data analysis is related to human beings and their behaviours. Due to the potential value that lies behind these massive data, there have been different proposed approaches for understanding corresponding patterns. To that end, monitoring people's interactions, whether counting them at fixed locations or tracking them by generating origin-destination matrices is crucial. The former can be used to determine the utilization of assets like roads and city attractions. The latter is valuable when planning transport infrastructure. Such insights allow a government to predict the adoption of new roads, new public transport routes, modification of existing infrastructure, and detection of congestion zones, resulting in more efficient designs and improvement. Smartphone data exploration can help research in various fields, e.g., urban planning, transportation, health care, and business marketing. It can also help organizations in decision making, policy implementation, monitoring and evaluation at all levels. This work aims to review the methods and techniques that have been implemented to discover knowledge from mobile phone data. We classify these existing methods and present a taxonomy of the related work by discussing their pros and cons.

cs.CY↗

arXiv2020

Effect of Spatial and Temporal Traffic Statistics on the Performance of Wireless Networks

The traffic in wireless networks has become diverse and fluctuating both spatially and temporally due to the emergence of new wireless applications and the complexity of scenarios. The purpose of this paper is to quantitatively analyze the impact of the wireless traffic, which fluctuates both spatially and temporally, on the performance of the wireless networks. Specially, we propose to combine the tools from stochastic geometry and queueing theory to model the spatial and temporal fluctuation of traffic, which to our best knowledge has seldom been evaluated analytically. We derive the spatial and temporal statistics, the total arrival rate, the stability of queues and the delay of users by considering two different spatial properties of traffic, i.e., the uniformly and non-uniformly distributed cases. The numerical results indicate that although the fluctuation of traffic (reflected by the variance of total arrival rate) when the users are clustered is much fiercer than that when the users are uniformly distributed, the unstable probability is smaller. Our work provides a useful reference for the design of wireless networks when the complex spatio-temporal fluctuation of the traffic is considered.

cs.NI↗

arXiv2020

DA4AD: End-to-End Deep Attention-based Visual Localization for Autonomous Driving

We present a visual localization framework based on novel deep attention aware features for autonomous driving that achieves centimeter level localization accuracy. Conventional approaches to the visual localization problem rely on handcrafted features or human-made objects on the road. They are known to be either prone to unstable matching caused by severe appearance or lighting changes, or too scarce to deliver constant and robust localization results in challenging scenarios. In this work, we seek to exploit the deep attention mechanism to search for salient, distinctive and stable features that are good for long-term matching in the scene through a novel end-to-end deep neural network. Furthermore, our learned feature descriptors are demonstrated to be competent to establish robust matches and therefore successfully estimate the optimal camera poses with high precision. We comprehensively validate the effectiveness of our method using a freshly collected dataset with high-quality ground truth trajectories and hardware synchronization between sensors. Results demonstrate that our method achieves a competitive localization accuracy when compared to the LiDAR-based localization solutions under various challenging circumstances, leading to a potential low-cost localization solution for autonomous driving.

cs.CV↗

arXiv2020

Efficient phonon cascades in hot photoluminescence of WSe$_2$ monolayers

Energy relaxation of photo-excited charge carriers is of significant fundamental interest and crucial for the performance of monolayer (1L) transition metal dichaclogenides (TMDs) in optoelectronics. We measure light scattering and emission in 1L-WSe$_2$ close to the laser excitation energy (down to~$\sim$0.6meV). We detect a series of periodic maxima in the hot photoluminescence intensity, stemming from energy states higher than the A-exciton state, in addition to sharp, non-periodic Raman lines related to the phonon modes. We find a period $\sim$15meV for peaks both below (Stokes) and above (anti-Stokes) the laser excitation energy. We detect 7 maxima from 78K to room temperature in the Stokes signal and 5 in the anti-Stokes, of increasing intensity with temperature. We assign these to phonon cascades, whereby carriers undergo phonon-induced transitions between real states in the free-carrier gap with a probability of radiative recombination at each step. We infer that intermediate states in the conduction band at the $Λ$-valley of the Brillouin zone participate in the cascade process of 1L-WSe$_2$. The observations explain the primary stages of carrier relaxation, not accessible so far in time-resolved experiments. This is important for optoelectronic applications, such as photodetectors and lasers, because these determine the recovery rate and, as a consequence, the devices' speed and efficiency.

cond-mat.mes-hall↗

arXiv2020

Numerical simulation of sky localization for LISA-TAIJI joint observation

LISA is considered to be launched alongside the Athena to probe the energetic astrophysical processes. LISA can determine the direction of sources for Athena's follow-up observation. As another space gravitational wave mission, TAIJI is expected to be launched in the 2030s. The LISA-TAIJI network would provide abundant merits for sources understanding. In this work, we simulate the joint LISA-TAIJI observations for gravitational waves from coalescing supermassive black hole binaries and monochromatic sources. By using the numerical mission orbits, we evaluate the performances of sky localization for various time-delay interferometry channels. For 30 days observation until coalescence, the LISA-TAIJI network in optimal operation can localize all simulated binary sources, $(10^7,\ 3.3 \times 10^6)\ M_\odot$, $(10^6,\ 3.3 \times 10^5)\ M_\odot$ and $(10^5,\ 3.3 \times 10^4)\ M_\odot$ at redshift $z=2$, in 0.4 deg$^2$ (field of view of Wide Field Imager on Athena). The angular resolution can be improved by more than 10 times comparing to LISA or TAIJI single detector at a given percentage of population. The improvements for monochromatic sources at 3 mHz and 10 mHz are relatively moderate in one-year observation. The precision of sky localization could be improved by around 1 to 3 times comparing to single LISA at a given percentage of sources. For a simulated 90 days observation for monochromatic waves, the LISA-TAIJI network still represents a considerable localization advantage which could be more than 10 times better.

gr-qc↗

arXiv2020

Learning while Respecting Privacy and Robustness to Distributional Uncertainties and Adversarial Data

Data used to train machine learning models can be adversarial--maliciously constructed by adversaries to fool the model. Challenge also arises by privacy, confidentiality, or due to legal constraints when data are geographically gathered and stored across multiple learners, some of which may hold even an "anonymized" or unreliable dataset. In this context, the distributionally robust optimization framework is considered for training a parametric model, both in centralized and federated learning settings. The objective is to endow the trained model with robustness against adversarially manipulated input data, or, distributional uncertainties, such as mismatches between training and testing data distributions, or among datasets stored at different workers. To this aim, the data distribution is assumed unknown, and lies within a Wasserstein ball centered around the empirical data distribution. This robust learning task entails an infinite-dimensional optimization problem, which is challenging. Leveraging a strong duality result, a surrogate is obtained, for which three stochastic primal-dual algorithms are developed: i) stochastic proximal gradient descent with an $ε$-accurate oracle, which invokes an oracle to solve the convex sub-problems; ii) stochastic proximal gradient descent-ascent, which approximates the solution of the convex sub-problems via a single gradient ascent step; and, iii) a distributionally robust federated learning algorithm, which solves the sub-problems locally at different workers where data are stored. Compared to the empirical risk minimization and federated learning methods, the proposed algorithms offer robustness with little computation overhead. Numerical tests using image datasets showcase the merits of the proposed algorithms under several existing adversarial attacks and distributional uncertainties.

cs.LG↗

arXiv2020

Ultralong carrier lifetime of topological edge states in a-Bi4Br4

The rising of quantum spin Hall insulators (QSHI) in two-dimensional (2D) systems has been attracting significant interest in current research, for which the 1D helical edge states, a hallmark of QSHI, are widely expected to be a promising platform for next-generation optoelectronics. However, the dynamics of the 1D edge states has not yet been experimentally addressed. Here, we report the observation of optical response of the topological helical edge states in a-Bi4Br4, using the infrared-pump infrared-probe microscopic spectroscopy. Remarkably, we observe that the carrier lifetime of the helical edge states reaches nanosecond-scale at room temperature, which is about 2 - 3 orders longer than that of most 2D topological surface states and is even comparable with that of the well developed optoelectronics semiconductors used in modern industry. The ultralong carrier lifetime of the topological edge states may be attributed to their helical and 1D nature. Our findings not only provide an ideal material for further investigations of the carrier dynamics of 1D helical edge states but also pave the way for its application in optoelectronics.

cond-mat.mtrl-sci↗

arXiv2020

Distributed Consensus of Nonlinear Multi-Agent Systems With Mismatched Uncertainties and Unknown High-Frequency Gains (Extended Version)

This brief addresses the distributed consensus problem of nonlinear multi-agent systems under a general directed communication topology. Each agent is governed by higher-order dynamics with mismatched uncertainties, multiple completely unknown high-frequency gains, and external disturbances. The main contribution of this brief is to present a new distributed consensus algorithm, enabling the control input of each agent to require minimal information from its neighboring agents, that is, only their output information. To this end, a dynamic system is explicitly constructed for each agent to generate a reference output. Theoretical and simulation verifications of the proposed algorithm are rigorously studied to ensure that asymptotic consensus can be achieved and that all closed-loop signals remain bounded.

eess.SY↗

arXiv2020

Gauss-Newton Unrolled Neural Networks and Data-driven Priors for Regularized PSSE with Robustness

Distributed renewable generation, elastic loads, and purposeful manipulation of meter readings challenge the monitoring and control of today's power systems (PS). In this context, to maintain a comprehensive view of the system in real time, fast and robust state estimation (SE) methods are urgently needed. Conventional PSSE solvers typically entail minimizing a nonlinear and nonconvex least-squares by e.g., the workhorse Gauss-Newton method. Those iterative solvers however, are sensitive to initialization and may get stuck in local minima. To overcome these hurdles and inspired by recent image denoising techniques, this paper advocates a learnable regularization term for PSSE that uses a deep neural network (DNN) prior. For the resultant regularized PSSE problem, a "Gauss-Newton-like" alternating minimization solver is first developed. To accommodate real-time monitoring, a novel end-to-end DNN is constructed by unrolling the proposed alternating minimization solver. Interestingly, the power network topology can be easily incorporated into the DNN by designing a graph neural network (GNN) based prior. To further endow the physics-based DNN with robustness against bad data, an adversarial DNN training method is discussed. Numerical tests using real load data on the IEEE $118$-bus benchmark system showcase the improved estimation and robustness performance of the proposed scheme compared with several state-of-the-art alternatives.

eess.SY↗

arXiv2020

Reinforcement Learning for Caching with Space-Time Popularity Dynamics

With the tremendous growth of data traffic over wired and wireless networks along with the increasing number of rich-media applications, caching is envisioned to play a critical role in next-generation networks. To intelligently prefetch and store contents, a cache node should be able to learn what and when to cache. Considering the geographical and temporal content popularity dynamics, the limited available storage at cache nodes, as well as the interactive in uence of caching decisions in networked caching settings, developing effective caching policies is practically challenging. In response to these challenges, this chapter presents a versatile reinforcement learning based approach for near-optimal caching policy design, in both single-node and network caching settings under dynamic space-time popularities. The herein presented policies are complemented using a set of numerical tests, which showcase the merits of the presented approach relative to several standard caching policies.

cs.IT↗

arXiv2020

Confinement of long-lived interlayer excitons in WS$_2$/WSe$_2$ heterostructures

Interlayer excitons in layered materials constitute a novel platform to study many-body phenomena arising from long-range interactions between quantum particles. The ability to localise individual interlayer excitons in potential energy traps is a key step towards simulating Hubbard physics in artificial lattices. Here, we demonstrate spatial localisation of long-lived interlayer excitons in a strongly confining trap array using a WS$_{2}$/WSe$_{2}$ heterostructure on a nanopatterned substrate. We detect long-lived interlayer excitons with lifetime approaching 0.2 ms and show that their confinement results in a reduced lifetime in the microsecond range and stronger emission rate with sustained optical selection rules. The combination of a permanent dipole moment, spatial confinement and long lifetime places interlayer excitons in a regime that satisfies one of the requirements for observing long-range dynamics in an optically resolvable trap lattice.

cond-mat.mes-hall↗

arXiv2020

Adaptive Distributed Laser Charging for Efficient Wireless Power Transfer

Distributed laser charging (DLC) is a wireless power transfer technology for mobile electronics. Similar to traditional wireless charging systems, the DLC system can only provide constant power to charge a battery. However, Li-ion battery needs dynamic input current and voltage, thus power, in order to optimize battery charging performance. Therefore, neither power transmission efficiency nor battery charging performance can be optimized by the DLC system. We at first propose an adaptive DLC (ADLC) system to optimize wireless power transfer efficiency and battery charging performance. Then, we analyze ADLC's power conversion to depict the adaptation mechanism. Finally, we evaluate the ADLC's power conversion performance by simulation, which illustrates its efficiency improvement by saving at least 60.4% of energy, comparing with the fixed-power charging system.

eess.SY↗

arXiv2020

MobiGyges: A mobile hidden volume for preventing data loss, improving storage utilization, and avoiding device reboot

Sensitive data protection is essential for mobile users. Plausibly Deniable Encryption (PDE) systems provide an effective manner to protect sensitive data by hiding them on the device. However, existing PDE systems can lose data due to overriding the hidden volume, waste physical storage because of the reserved area used for avoiding data loss, and require device reboot when using the hidden volume. This paper presents MobiGyges, a hidden volume-based mobile PDE system, to fill the gap. MobiGyges addresses the problem of data loss by restricting each storage block used only by one volume, and it improves storage utilization by eliminating the reserved area. MobiGyges can also avoid device reboot by mounting the hidden volume dynamically on-demand with the Dynamic Mounting service. Moreover, we identify two novel PDE oriented attacks, the capacity comparison attack and the fill-to-full attack. MobiGyges can defend them by jointly leveraging the Shrunk U-disk method and multi-level deniability. We implement the MobiGyges proof-of-concept system on a real mobile phone Google Nexus 6P with LineageOS 13. Experimental results show that MobiGyges prevents data loss, avoids device reboot, improves storage utilization by over 30% with acceptable performance overhead compared with current works.

cs.CR↗