Searcharxiv⌕ Search

arXiv subjects

Wei Wei

Publications and source records attributed to Wei Wei.

At least 505 records · Page 28Linked to original sources
arXiv2017

Structural Embedding of Syntactic Trees for Machine Comprehension

Deep neural networks for machine comprehension typically utilizes only word or character embeddings without explicitly taking advantage of structured linguistic information such as constituency trees and dependency trees. In this paper, we propose structural embedding of syntactic trees (SEST), an algorithm framework to utilize structured information and encode them into vector representations that can boost the performance of algorithms for the machine comprehension. We evaluate our approach using a state-of-the-art neural attention model on the SQuAD dataset. Experimental results demonstrate that our model can accurately identify the syntactic boundaries of the sentences and extract answers that are syntactically coherent over the baseline methods.

cs.CL↗
arXiv2017

Learning to Identify Ambiguous and Misleading News Headlines

Accuracy is one of the basic principles of journalism. However, it is increasingly hard to manage due to the diversity of news media. Some editors of online news tend to use catchy headlines which trick readers into clicking. These headlines are either ambiguous or misleading, degrading the reading experience of the audience. Thus, identifying inaccurate news headlines is a task worth studying. Previous work names these headlines "clickbaits" and mainly focus on the features extracted from the headlines, which limits the performance since the consistency between headlines and news bodies is underappreciated. In this paper, we clearly redefine the problem and identify ambiguous and misleading headlines separately. We utilize class sequential rules to exploit structure information when detecting ambiguous headlines. For the identification of misleading headlines, we extract features based on the congruence between headlines and bodies. To make use of the large unlabeled data set, we apply a co-training method and gain an increase in performance. The experiment results show the effectiveness of our methods. Then we use our classifiers to detect inaccurate headlines crawled from different sources and conduct a data analysis.

cs.CL↗
arXiv2017

Efficient Online Inference for Infinite Evolutionary Cluster models with Applications to Latent Social Event Discovery

The Recurrent Chinese Restaurant Process (RCRP) is a powerful statistical method for modeling evolving clusters in large scale social media data. With the RCRP, one can allow both the number of clusters and the cluster parameters in a model to change over time. However, application of the RCRP has largely been limited due to the non-conjugacy between the cluster evolutionary priors and the Multinomial likelihood. This non-conjugacy makes inference di cult and restricts the scalability of models which use the RCRP, leading to the RCRP being applied only in simple problems, such as those that can be approximated by a single Gaussian emission. In this paper, we provide a novel solution for the non-conjugacy issues for the RCRP and an example of how to leverage our solution for one speci c problem - the social event discovery problem. By utilizing Sequential Monte Carlo methods in inference, our approach can be massively paralleled and is highly scalable, to the extent it can work on tens of millions of documents. We are able to generate high quality topical and location distributions of the clusters that can be directly interpreted as real social events, and our experimental results suggest that the approaches proposed achieve much better predictive performance than techniques reported in prior work. We also demonstrate how the techniques we develop can be used in a much more general ways toward similar problems.

cs.AI↗
arXiv2017

Beyond Low Rank: A Data-Adaptive Tensor Completion Method

Low rank tensor representation underpins much of recent progress in tensor completion. In real applications, however, this approach is confronted with two challenging problems, namely (1) tensor rank determination; (2) handling real tensor data which only approximately fulfils the low-rank requirement. To address these two issues, we develop a data-adaptive tensor completion model which explicitly represents both the low-rank and non-low-rank structures in a latent tensor. Representing the non-low-rank structure separately from the low-rank one allows priors which capture the important distinctions between the two, thus enabling more accurate modelling, and ultimately, completion. Through defining a new tensor rank, we develop a sparsity induced prior for the low-rank structure, with which the tensor rank can be automatically determined. The prior for the non-low-rank structure is established based on a mixture of Gaussians which is shown to be flexible enough, and powerful enough, to inform the completion process for a variety of real tensor data. With these two priors, we develop a Bayesian minimum mean squared error estimate (MMSE) framework for inference which provides the posterior mean of missing entries as well as their uncertainty. Compared with the state-of-the-art methods in various applications, the proposed model produces more accurate completion results.

cs.CV↗
arXiv2017

When Unsupervised Domain Adaptation Meets Tensor Representations

Domain adaption (DA) allows machine learning methods trained on data sampled from one distribution to be applied to data sampled from another. It is thus of great practical importance to the application of such methods. Despite the fact that tensor representations are widely used in Computer Vision to capture multi-linear relationships that affect the data, most existing DA methods are applicable to vectors only. This renders them incapable of reflecting and preserving important structure in many problems. We thus propose here a learning-based method to adapt the source and target tensor representations directly, without vectorization. In particular, a set of alignment matrices is introduced to align the tensor representations from both domains into the invariant tensor subspace. These alignment matrices and the tensor subspace are modeled as a joint optimization problem and can be learned adaptively from the data using the proposed alternative minimization scheme. Extensive experiments show that our approach is capable of preserving the discriminative power of the source domain, of resisting the effects of label noise, and works effectively for small sample sizes, and even one-shot DA. We show that our method outperforms the state-of-the-art on the task of cross-domain visual recognition in both efficacy and efficiency, and particularly that it outperforms all comparators when applied to DA of the convolutional activations of deep convolutional networks.

cs.CV↗
arXiv2017

Optimum stabilization of self-mode-locked quantum dash lasers using dual optical feedback with improved tolerance against phase delay mismatch

We experimentally investigate the RF linewidth and timing jitter over a wide range of delay tuning in a self-mode-locked two-section quantum dash lasers emitting at ~ 1.55 micron and operating at ~ 21 GHz repetition rate subject to single and dual optical feedback into gain section. Various feedback conditions are investigated and optimum levels determined for narrowest linewidth and reduced timing jitter for both single and dual loop configurations. We demonstrate that dual loop feedback, with the shorter feedback cavity tuned to be fully resonant, followed by fine tuning of the phase of the longer feedback cavity, gives stable narrow RF spectra across the widest delay range, unlike single loop feedback. In addition, for dual loop configurations, under fully resonant conditions, integrated timing jitter is reduced from 3.9 ps to 295 fs [10 kHz-100 MHz], the RF linewidth narrows from 100 kHz to < 1 kHz, with more than 30 dB fundamental side-mode suppression. We show that dual loop optical feedback with separate fine tuning of both external cavities is far superior to single loop feedback, with increased system tolerance against phase delay mismatch, making it a robust and cost-effective technique for developing practical, reliable and low-noise mode-locked lasers, optoelectronic oscillators and pulsed photonic circuits.

physics.optics↗
arXiv2016

China building energy consumption: definitions and measures from an operational perspective

There is an increasing awareness of the significance of Chinese building energy consumption(BEC). However, something worth discussing is that estimate the building energy consumption adopting the definition of life cycle or operation. In the existing studies with various evaluation methods, the issue about the amount of energy consumed by China buildings has not been understood. In order to settle the disputes over the calculation of BEC, this paper establish an appropriate accounting method of building energy to present BEC situation in China and lay the foundation for building energy efficiency. Adopting the conception of building operational energy consumption, we find that the energy consumption of buildings just accounts for 15% - 16% of the final total energy consumption in China; by contrast, the previous calculations usually have double accounting through top-down approach if central heat-supply of buildings was given into additional consideration.

econ.GN↗
arXiv2016

Distributed Optimal Gas-Power Flow Using Convex Optimization and ADMM

This paper proposes a convex optimization based distributed algorithm to solve multi-period optimal gas-power flow (OGPF) in coupled energy distribution systems. At the gas distribution system side, the non-convex Weymouth gas flow equations is convexified as quadratic constraints. The optimal gas flow (OGF) subproblem is solved by an iterative second-order cone programming procedure, whose efficiency is two orders of magnitudes higher than traditional nonlinear methods. A convex quadratic program based initiation scheme is suggested, which helps to find a high-quality starting point. At the power distribution system side, convex relaxation is performed on the non-convex branch flow equations, and the optimal power flow (OPF) subproblem gives rise to a second order cone program. Tightness is guaranteed by the radial topology. In the proposed distributed algorithm, OGF and OPF are solved independently, and coordinated by the alternating direction multiplier method (ADMM). Numerical results corroborate significant enhancements on computational robustness and efficiency compared with existing centralized OGPF methods.

math.OC↗
arXiv2016

Risk Assessment of Multi-timescale Cascading Outages based on Markovian Tree Search

In the risk assessment of cascading outages, the rationality of simulation and efficiency of computation are both of great significance. To overcome the drawback of sampling-based methods that huge computation resources are required and the shortcoming of initial contingency selection practices that the dependencies in sequences of outages are omitted, this paper proposes a novel risk assessment approach by searching on Markovian Tree. The Markovian tree model is reformulated from the quasi-dynamic multi-timescale simulation model proposed recently to ensure reasonable modeling and simulation of cascading outages. Then a tree search scheme is established to avoid duplicated simulations on same cascade paths, significantly saving computation time. To accelerate the convergence of risk assessment, a risk estimation index is proposed to guide the search for states with major contributions to the risk, and the risk assessment is realized based on the risk estimation index with a forward tree search and backward update algorithm. The effectiveness of the proposed method is illustrated on a 4-node power system, and its convergence profile as well as efficiency is demonstrated on the RTS-96 test system.

eess.SY↗
arXiv2016

Chiral phase transition in QED$_3$ at finite temperature

Chiral phase transition in (2+1)-dimensional quantum electrodynamics (QED$_3$) at finite temperature is investigated in the framework of truncated Dyson-Schwinger equations (DSEs). We go beyond the widely used instantaneous approximation and adopt a method that retains the full frequency dependence of the fermion self-energy. We also take further step to include the effects of wave-function renormalizations and introduce a minimal dressing of the bare vertex. Finally, with the more complete solutions of the truncated DSEs, we revisit the study of chiral phase transition in finite-temperature QED$_3$.

hep-ph↗
arXiv2016

Extended LTLvis Motion Planning interface (Extended Technical Report)

This paper introduces an extended version of the Linear Temporal Logic (LTL) graphical interface. It is a sketch based interface built on the Android platform which makes the LTL control interface more straightforward and friendly to nonexpert users. By predefining a set of areas of interest, this interface can quickly and efficiently create plans that satisfy extended plan goals in LTL. The interface can also allow users to customize the paths for this plan by sketching a set of reference trajectories. Given the custom paths by the user, the LTL specification and the environment, the interface generates a plan balancing the customized paths and the LTL specifications. We also show experimental results with the implemented interface.

cs.RO↗
arXiv2016

Strategic Bidding and Equilibria in Coupled Gas and Electricity Markets

The wide adoption of gas fired units and power-to-gas technology brings remarkable interdependency between natural gas and electricity infrastructures. This paper studies the equilibria of coupled gas and electricity energy markets driven by the strategic bidding behavior: each producer endeavours to maximizes its own profit subjecting to the market clearing outcome. The market equilibria is formulated as an equilibrium problem with equilibrium constraints (EPEC). A special diagonalization algorithm (DA) is devised, in which the unilateral equilibria of the gas or electricity market is found in the inner loop given the rival's strategy; the interaction of the two markets are tackled in the outer loop. Case studies on two test systems validates the proposed methodology.

math.OC↗
arXiv2016

Stability and steady state analysis of distributed cooperative droop controlled DC microgrids

Distributed cooperative droop control consisting of the primary decentralized droop control and the {secondary} distributed correction control is studied in this paper, which aims to achieve an exact current sharing between generators, worked in the voltage control mode, of DC microgrids. For the DC microgrids with the distributed cooperative droop control, the dynamic stability has not been well investigated although its steady performance has been widely reported. This paper focuses on the stability problem and shows it is equivalent to the semistability problem of a class of second-order matrix systems. Some further sufficient conditions as well followed. The steady state is analyzed deeply for some special cases. A DC microgrid of three nodes is simulated on the Matlab/Simulink platform to illustrate the efficacy of analytic results.

math.OC↗
arXiv2016

Robust Defense Strategy for Gas-Electric Systems Against Malicious Attacks

This paper proposes a methodology to identify and protect vulnerable components of connected gas and electric infrastructures from malicious attacks, and to guarantee a resilient operation by deploying valid corrective actions (while accounting for the interdependency of gas pipeline network and power transmission network). The proposed mathematical formulation reduces to a tri-level optimization problem, where the lower level is a multiperiod economic dispatch of the gas-electric system, the middle level distinguishes the most threatening attack on the coupled physical infrastructures, and the upper level provides optimal preventive decisions to reinforce the vulnerable components and increase the system resilience. By reformulating the lower level problem as a mixed integer linear programming (MILP), a nested column-and-constraint generation (C\&CG) algorithm is developed to solve the min-max-min model. Case studies on two test systems demonstrate the effectiveness and efficiency of the proposed methodology.

math.OC↗
arXiv2016

Robust Risk-Constrained Unit Commitment with Large-scale Wind Generation: An Adjustable Uncertainty Set Approach

This paper addresses two vital issues which are barely discussed in the literature on robust unit commitment (RUC): 1) how much the potential operational loss could be if the realization of uncertainty is beyond the prescribed uncertainty set; 2) how large the prescribed uncertainty set should be when it is used for RUC decision making. In this regard, a robust risk-constrained unit commitment (RRUC) formulation is proposed to cope with large-scale volatile and uncertain wind generation. Differing from existing RUC formulations, the wind generation uncertainty set in RRUC is adjustable via choosing diverse levels of operational risk. By optimizing the uncertainty set, RRUC can allocate operational flexibility of power systems over spatial and temporal domains optimally, reducing operational cost in a risk-constrained manner. Moreover, since impact of wind generation realization out of the prescribed uncertainty set on operational risk is taken into account, RRUC outperforms RUC in the case of rare events. Three algorithms based on column and constraint generation (C&CG) are derived to solve the RRUC. As the proposed algorithms are quite general, they can also apply to other RUC models to improve their computational efficiency. Simulations on a modified IEEE 118-bus system demonstrate the effectiveness and efficiency of the proposed methodology

math.OC↗
arXiv2015

Admissible Region of Large-Scale Uncertain Wind Generation Considering Small-Signal Stability of Power System

The increasing integration of wind generation has brought great challenges to small-signal stability analysis of bulk power systems, since the volatility and uncertainty nature of wind generation may considerably affect equilibriums of the systems. In this regard, this paper develops a conceptual framework to depict the influence of uncertain wind power injections (WPIs) on small-signal stability of bulk power systems. To do this, a new concept, the admissible region of uncertain wind generation considering small-signal stability (SSAR) is introduced to geometrically measure how much uncertain wind generation can be accommodated by a bulk power system without breaking its small-signal stability. As a generalization of the traditional concept of the small-signal stability region (SSSR), SSAR is derived by extending the SSSR to a higher-dimensional injection space that incorporates both the conventional nodal generation injections and the WPIs, and then mapping it onto the lower-dimensional WPI space. Case studies on the modified New England 39-bus system with multiple wind farms illustrate the SSAR concept and its potential applications.

math.OC↗
arXiv2015

Robust Unit Commitment Considering Strategic Wind Generation Curtailment

Wind generation is traditionally treated as a non-dispatchable resource and is fully absorbed unless there are security issues. To tackle the operational reliability issues caused by the volatile and non-dispatchable wind generation, many dispatch frameworks have been proposed, including robust unit commitment (RUC) considering wind variation. One of the drawbacks that commonly exist in those dispatch frameworks is increased demand on flexibility resources and associated costs. To improve wind dispatchability and reduce flexibility resource costs, in this paper, we propose a novel RUC model considering strategic wind generation curtailment (WGC). Strategic WGC can reduce wind uncertainty and variability and increase the visibility of wind generation capacity. As a result, the ramping requirement for wind generation will be reduced and ramp-up capability of wind generation can be increased, leading to reduced day-ahead operational cost with guaranteed operational reliability requirement of power systems. The economic benefits also include profits gained by wind farm by providing ramping-up capacities other auxiliary services. We also propose a solution algorithm based on the column and constraint generation (C&CG). Simulations on the IEEE 39-bus system and two larger test systems demonstrate the effectiveness of the proposed RUC model and efficiency of the proposed computational methodology.

math.OC↗