Rui Liu, Wei Wei, Weiguang Mao, Maria Chikina
Attention models have been intensively studied to improve NLP tasks such as machine comprehension via both question-aware passage attention model and self-matching attention model. Our research proposes phase conductor (PhaseCond) for attention models in two meaningful ways. First, PhaseCond, an architecture of multi-layered attention models, consists of multiple phases each implementing a stack of attention layers producing passage representations and a stack of inner or outer fusion layers regulating the information flow. Second, we extend and improve the dot-product attention function for PhaseCond by simultaneously encoding multiple question and passage embedding layers from different perspectives. We demonstrate the effectiveness of our proposed model PhaseCond on the SQuAD dataset, showing that our model significantly outperforms both state-of-the-art single-layered and multiple-layered attention models. We deepen our results with new findings via both detailed qualitative analysis and visualized examples showing the dynamic changes through multi-layered attention models.
Rui Liu, Tianyi Wu, Barzan Mozafari
There has been substantial research on sub-linear time approximate algorithms for Maximum Inner Product Search (MIPS). To achieve fast query time, state-of-the-art techniques require significant preprocessing, which can be a burden when the number of subsequent queries is not sufficiently large to amortize the cost. Furthermore, existing methods do not have the ability to directly control the suboptimality of their approximate results with theoretical guarantees. In this paper, we propose the first approximate algorithm for MIPS that does not require any preprocessing, and allows users to control and bound the suboptimality of the results. We cast MIPS as a Best Arm Identification problem, and introduce a new bandit setting that can fully exploit the special structure of MIPS. Our approach outperforms state-of-the-art methods on both synthetic and real-world datasets.
Rui Liu, Zheng Lin, Weiping Wang
Recently, generative methods have been widely used in keyphrase prediction, thanks to their capability to produce both present keyphrases that appear in the source text and absent keyphrases that do not match any source text. However, the absent keyphrases are generated at the cost of the performance on present keyphrase prediction, since previous works mainly use generative models that rely on the copying mechanism and select words step by step. Besides, the extractive model that directly extracts a text span is more suitable for predicting the present keyphrase. Considering the different characteristics of extractive and generative methods, we propose to divide the keyphrase prediction into two subtasks, i.e., present keyphrase extraction (PKE) and absent keyphrase generation (AKG), to fully exploit their respective advantages. On this basis, a joint inference framework is proposed to make the most of BERT in two subtasks. For PKE, we tackle this task as a sequence labeling problem with the pre-trained language model BERT. For AKG, we introduce a Transformer-based architecture, which fully integrates the present keyphrase knowledge learned from PKE by the fine-tuned BERT. The experimental results show that our approach can achieve state-of-the-art results on both tasks on benchmark datasets.
Rui Liu, Xichan Zhu, Xuan Zhao, Jian Ma
A gradual takeover strategy is proposed, in which the dynamic driving privilege assignment in real-time and the driving privilege gradual handover are realized. Firstly, the driving privilege assignment based on the risk level is achieved. The naturalistic driving data is applied to study the driver behavior during danger. TTC (time to collision) is defined as an obvious risk measure, whereas the time before the host vehicle has to brake assuming that the target vehicle is braking is defined as the potential risk measure, i.e. the time margin (TM). A risk assessment algorithm is proposed based on the obvious risk and potential risk. Secondly, the driving privilege gradual handover is realized. The non-cooperative MPC (model predictive control) is employed to resolve the conflicts between the driver and active safety system. The naturalistic driving data are applied to verify the effectiveness of the risk assessment algorithm, and the risk assessment algorithm performs better than TTC in the ROC (receiver operating characteristic). It is identified that the Nash equilibrium of the non-cooperative MPC can be achieved by using a non-iterative method. The driving privilege gradual handover is realized by using the confidence matrixes updating. The simulation verification shows that the gradual takeover strategy can achieve the driving privilege gradual handover between the driver and active safety system.
Yuwei He, Rui Liu, Lijuan Liu, Jun Chen, Wensi Wang, Yuming Wang
Jul 11, 2020·astro-ph.SR·PDF Recently solar flares exhibiting a double J-shaped ribbons in the lower solar atmosphere have been paid increasing attention in the context of extending the two-dimensional standard flare model to three dimensions, as motivated by the spatial correlation between photospheric current channels and flare ribbons. Here we study the electric currents through the photospheric area swept by flare ribbons (termed synthesized ribbon area or SRA), with a sample of 71 two-ribbon flares, of which 36 are J-shaped. Electric currents flowing through one ribbon are highly correlated with those through the other, therefore belonging to the same current system. The non-neutrality factor of this current system is independent of the flare magnitude, implying that both direct and return currents participate in flares. J-shaped flares are distinct from non-J-shaped flares in the following aspects: 1) Electric current densities within J-shaped SRA are significantly smaller than those within non-J-shaped SRA, but J-shaped SRA and its associated magnetic flux are also significantly larger. 2) Electric currents through SRA are positively correlated with the flare magnitude, but J-shaped flares show stronger correlation than non-J-shaped flares. 3) The majority (75\%) of J-shaped flares are eruptive, while the majority (86\%) of non-J-shaped flares are confined; accordingly, hosting active regions of J-shaped flares are more likely to be sigmoidal than non-J-shaped flares. Thus, J-shaped flares constitute a distinct subset of two-ribbon flares, probably the representative of eruptive ones. Further, we found that combining SRA and its associated magnetic flux has the potential to differentiate eruptive from confined flares.
Jialin Chen, Yingna Su, Rui Liu, Bernhard Kliem, Qingmin Zhang, Haisheng Ji, Tie Liu
Nov 25, 2021·astro-ph.SR·PDF We investigate the failed partial eruption of a filament system in NOAA AR 12104 on 2014 July 5, using multiwavelength EUV, magnetogram, and H$α$ observations, as well as magnetic field modeling. The filament system consists of two almost co-spatial segments with different end points, both resembling a C shape. Following an ejection and a precursor flare related to flux cancellation, only the upper segment rises and then displays a prominent twisted structure, while rolling over toward its footpoints. The lower segment remains undisturbed, indicating that the system possesses a double-decker structure. The erupted segment ends up with a reverse-C shape, with material draining toward its footpoints, while losing its twist. Using the flux rope insertion method, we construct a model of the source region that qualitatively reproduces key elements of the observed evolution. At the eruption onset, the model consists of a flux rope atop a flux bundle with negligible twist, which is consistent with the observational interpretation that the filament possesses a double-decker structure. The flux rope reaches the critical height of the torus instability during its initial relaxation, while the lower flux bundle remains in stable equilibrium. The eruption terminates when the flux rope reaches a dome-shaped quasi-separatrix layer that is reminiscent of a magnetic fan surface, although no magnetic null is found. The flux rope is destroyed by reconnection with the confining overlying flux above the dome, transferring its twist in the process.
Qifei Yu, Zhexin Shen, Yijiang Pang, Rui Liu
A mixed aerial and ground robot team, which includes both unmanned ground vehicles (UGVs) and unmanned aerial vehicles (UAVs), is widely used for disaster rescue, social security, precision agriculture, and military missions. However, team capability and corresponding configuration vary since robots have different motion speeds, perceiving ranges, reaching areas, and resilient capabilities to the dynamic environment. Due to heterogeneous robots inside a team and the resilient capabilities of robots, it is challenging to perform a task with an optimal balance between reasonable task allocations and maximum utilization of robot capability. To address this challenge for effective mixed ground and aerial teaming, this paper developed a novel teaming method, proficiency aware multi-agent deep reinforcement learning (Mix-RL), to guide ground and aerial cooperation by considering the best alignments between robot capabilities, task requirements, and environment conditions. Mix-RL largely exploits robot capabilities while being aware of the adaption of robot capabilities to task requirements and environment conditions. Mix-RL's effectiveness in guiding mixed teaming was validated with the task "social security for criminal vehicle tracking".
Rui Liu, Siddharth Bhatia, Bryan Hooi
An edge stream is a common form of presentation of dynamic networks. It can evolve with time, with new types of nodes or edges being continuously added. Existing methods for anomaly detection rely on edge occurrence counts or compare pattern snippets found in historical records. In this work, we propose Isconna, which focuses on both the frequency and the pattern of edge records. The burst detection component targets anomalies between individual timestamps, while the pattern detection component highlights anomalies across segments of timestamps. These two components together produce three intermediate scores, which are aggregated into the final anomaly score. Isconna does not actively explore or maintain pattern snippets; it instead measures the consecutive presence and absence of edge records. Isconna is an online algorithm, it does not keep the original information of edge records; only statistical values are maintained in a few count-min sketches (CMS). Isconna's space complexity $O(rc)$ is determined by two user-specific parameters, the size of CMSs. In worst case, Isconna's time complexity can be up to $O(rc)$, but it can be amortized in practice. Experiments show that Isconna outperforms five state-of-the-art frequency- and/or pattern-based baselines on six real-world datasets with up to 20 million edge records.
Hanya Pan, Rui Liu, Tingyu Gou, Bernhard Kliem, Yingna Su, Jun Chen, Yuming Wang
Solar filaments often erupt partially. Although how they split remains elusive, the splitting process has the potential of revealing the filament structure and eruption mechanism. Here we investigate the pre-eruption splitting of an apparently single filament and its subsequent partial eruption on 2012 September 27. The evolution is characterized by three stages with distinct dynamics. During the quasi-static stage, the splitting proceeds gradually for about 1.5 hrs, with the upper branch rising at a few kilometers per second and displaying swirling motions about its axis. During the precursor stage that lasts for about 10 min, the upper branch rises at tens of kilometers per second, with a pair of conjugated dimming regions starting to develop at its footpoints; with the swirling motions turning chaotic, the axis of the upper branch whips southward, which drives an arc-shaped EUV front propagating in the similar direction. During the eruption stage, the upper branch erupts with the onset of a C3.7-class two-ribbon flare, while the lower branch remains stable. Judging from the well separated footpoints of the upper branch from those of the lower one, we suggest that the pre-eruption filament processes a double-decker structure composed of two distinct flux bundles, whose formation is associated with gradual magnetic flux cancellations and converging photospheric flows around the polarity inversion line.
Muhan Na, Rui Liu, Feilong, Guanglai Gao
Cyrillic and Traditional Mongolian are the two main members of the Mongolian writing system. The Cyrillic-Traditional Mongolian Bidirectional Conversion (CTMBC) task includes two conversion processes, including Cyrillic Mongolian to Traditional Mongolian (C2T) and Traditional Mongolian to Cyrillic Mongolian conversions (T2C). Previous researchers adopted the traditional joint sequence model, since the CTMBC task is a natural Sequence-to-Sequence (Seq2Seq) modeling problem. Recent studies have shown that Recurrent Neural Network (RNN) and Self-attention (or Transformer) based encoder-decoder models have shown significant improvement in machine translation tasks between some major languages, such as Mandarin, English, French, etc. However, an open problem remains as to whether the CTMBC quality can be improved by utilizing the RNN and Transformer models. To answer this question, this paper investigates the utility of these two powerful techniques for CTMBC task combined with agglutinative characteristics of Mongolian language. We build the encoder-decoder based CTMBC model based on RNN and Transformer respectively and compare the different network configurations deeply. The experimental results show that both RNN and Transformer models outperform the traditional joint sequence model, where the Transformer achieves the best performance. Compared with the joint sequence baseline, the word error rate (WER) of the Transformer for C2T and T2C decreased by 5.72\% and 5.06\% respectively.
Kailin Liang, Bin Liu, Yifan Hu, Rui Liu, Feilong Bao, Guanglai Gao
Text-to-Speech (TTS) synthesis for low-resource languages is an attractive research issue in academia and industry nowadays. Mongolian is the official language of the Inner Mongolia Autonomous Region and a representative low-resource language spoken by over 10 million people worldwide. However, there is a relative lack of open-source datasets for Mongolian TTS. Therefore, we make public an open-source multi-speaker Mongolian TTS dataset, named MnTTS2, for the benefit of related researchers. In this work, we prepare the transcription from various topics and invite three professional Mongolian announcers to form a three-speaker TTS dataset, in which each announcer records 10 hours of speeches in Mongolian, resulting 30 hours in total. Furthermore, we build the baseline system based on the state-of-the-art FastSpeech2 model and HiFi-GAN vocoder. The experimental results suggest that the constructed MnTTS2 dataset is sufficient to build robust multi-speaker TTS models for real-world applications. The MnTTS2 dataset, training recipe, and pretrained models are released at: \url{https://github.com/ssmlkl/MnTTS2}
Yifan Hu, Pengkai Yin, Rui Liu, Feilong Bao, Guanglai Gao
This paper introduces a high-quality open-source text-to-speech (TTS) synthesis dataset for Mongolian, a low-resource language spoken by over 10 million people worldwide. The dataset, named MnTTS, consists of about 8 hours of transcribed audio recordings spoken by a 22-year-old professional female Mongolian announcer. It is the first publicly available dataset developed to promote Mongolian TTS applications in both academia and industry. In this paper, we share our experience by describing the dataset development procedures and faced challenges. To demonstrate the reliability of our dataset, we built a powerful non-autoregressive baseline system based on FastSpeech2 model and HiFi-GAN vocoder, and evaluated it using the subjective mean opinion score (MOS) and real time factor (RTF) metrics. Evaluation results show that the powerful baseline system trained on our dataset achieves MOS above 4 and RTF about $3.30\times10^{-1}$, which makes it applicable for practical use. The dataset, training recipe, and pretrained TTS models are freely available \footnote{\label{github}\url{https://github.com/walker-hyf/MnTTS}}.
Rui Liu, Yuanyuan Chen, Anran Li, Yi Ding, Han Yu, Cuntai Guan
Insufficient data is a long-standing challenge for Brain-Computer Interface (BCI) to build a high-performance deep learning model. Though numerous research groups and institutes collect a multitude of EEG datasets for the same BCI task, sharing EEG data from multiple sites is still challenging due to the heterogeneity of devices. The significance of this challenge cannot be overstated, given the critical role of data diversity in fostering model robustness. However, existing works rarely discuss this issue, predominantly centering their attention on model training within a single dataset, often in the context of inter-subject or inter-session settings. In this work, we propose a hierarchical personalized Federated Learning EEG decoding (FLEEG) framework to surmount this challenge. This innovative framework heralds a new learning paradigm for BCI, enabling datasets with disparate data formats to collaborate in the model training process. Each client is assigned a specific dataset and trains a hierarchical personalized model to manage diverse data formats and facilitate information exchange. Meanwhile, the server coordinates the training procedure to harness knowledge gleaned from all datasets, thus elevating overall performance. The framework has been evaluated in Motor Imagery (MI) classification with nine EEG datasets collected by different devices but implementing the same MI task. Results demonstrate that the proposed frame can boost classification performance up to 16.7% by enabling knowledge sharing between multiple datasets, especially for smaller datasets. Visualization results also indicate that the proposed framework can empower the local models to put a stable focus on task-related areas, yielding better performance. To the best of our knowledge, this is the first end-to-end solution to address this important challenge.
Rui Liu, Jianping Pan
Air pollution has become a global concern for many years. Vehicular crowdsensing systems make it possible to monitor air quality at a fine granularity. To better utilize the sensory data with varying credibility, truth discovery frameworks are introduced. However, in urban cities, there is a significant difference in traffic volumes of streets or blocks, which leads to a data sparsity problem for truth discovery. Protecting the privacy of participant vehicles is also a crucial task. We first present a data masking-based privacy-preserving truth discovery framework, which incorporates spatial and temporal correlations to solve the sparsity problem. To further improve the truth discovery performance of the presented framework, an enhanced version is proposed with anonymous communication and data perturbation. Both frameworks are more lightweight than the existing cryptography-based methods. We also evaluate the work with simulations and fully discuss the performance and possible extensions.
Tingyu Gou, Rui Liu, Astrid M. Veronig, Bin Zhuang, Ting Li, Wensi Wang, Mengjiao Xu, Yuming Wang
Solar coronal mass ejections are the most energetic events in the Solar System. In their standard formation model, a magnetic flux rope builds up into a coronal mass ejection through magnetic reconnection that continually converts overlying, untwisted magnetic flux into twisted flux enveloping the pre-existing rope. However, only a minority of coronal mass ejections carry a coherent magnetic flux rope as their core structure, which casts doubt on the universality of this orderly wrapping process. Here we provide observational evidence of a different formation and eruption mechanism of a magnetic flux rope from an S-shaped thread, where its magnetic flux is fully replaced via flare reconnections. One of the footpoints of the sigmoidal feature slipped and expanded during the formation, and then moved to a completely new place, associated with the highly dynamical evolution of flare ribbons and a twofold increase in magnetic flux through the footpoint, during the eruption. Such a configuration is not predicted by standard formation models or numerical simulations and highlights the three-dimensional nature of magnetic reconnections between the flux rope and the surrounding magnetic field.
Rui Liu, Viacheslav S. Titov, Tingyu Gou, Yuming Wang, Kai Liu, Haimin Wang
May 27, 2014·astro-ph.SR·PDF We report the observation of an X-class long-duration flare which is clearly confined. It appears as a compact-loop flare in the traditional EUV passbands (171 and 195 Å), but in the passbands sensitive to flare plasmas (94 and 131 Å), it exhibits a cusp-shaped structure above an arcade of loops like other long-duration events. Inspecting images in a running difference approach, we find that the seemingly diffuse, quasi-static cusp-shaped structure consists of multiple nested loops that repeatedly rise upward and disappear approaching the cusp edge. Over the gradual phase, we detect numerous episodes of loop rising, each lasting minutes. A differential emission measure analysis reveals that the temperature is highest at the top of the arcade and becomes cooler at higher altitudes within the cusp-shaped structure, contrary to typical long-duration flares. With a nonlinear force-free model, our analysis shows that the event mainly involves two adjacent sheared arcades separated by a T-type hyperbolic flux tube (HFT). One of the arcades harbors a magnetic flux rope, which is identified with a filament that survives the flare owing to the strong confining field. We conclude that a new emergence of magnetic flux in the other arcade triggers the flare, while the preexisting HFT and flux rope dictate the structure and dynamics of the flare loops and ribbons during the long-lasting decay phase, and that a quasi-separatrix layer high above the HFT could account for the cusp-shaped structure.
Dongni Tan, Xujian Huang, Rui Liu
We introduce a new class of Banach spaces, called generalized-lush spaces (GL-spaces for short), which contains almost-CL-spaces, separable lush spaces (specially, separable $C$-rich subspaces of $C(K)$), and even the two-dimensional space with hexagonal norm. We obtain that the space $C(K,E)$ of the vector-valued continuous functions is a GL-space whenever $E$ is, and show that the GL-space is stable under $c_0$-, $l_1$- and $l_\infty$-sums. As an application, we prove that the Mazur-Ulam property holds for a larger class of Banach spaces, called local-GL-spaces, including all lush spaces and GL-spaces. Furthermore, we generalize the stability properties of GL-spaces to local-GL-spaces. From this, we can obtain many examples of Banach spaces having the Mazur-Ulam property.
Rui Liu, Mingcheng Yang, Ke Chen, Meiying Hou, Kiwing To
Using an event-driven molecular dynamics simulation, we show that simple monodisperse granular beads confined in coupled columns may oscillate as a new type of granular clock. To trigger this oscillation, the system needs to be driven against gravity into a density-inverted state, with a high-density clustering phase supported from below by a gas-like low-density phase (Leidenfrost effect) in each column. Our analysis reveals that the density-inverted structure and the relaxation dynamics between the phases can amplify any small asymmetry between the columns, and lead to a giant oscillation. The oscillation occurs only for an intermediate range of the coupling strength, and the corresponding phase diagram can be universally described with a characteristic height of the density-inverted structure. A minimal two-phase model is proposed and linear stability analysis shows that the triggering mechanism of the oscillation can be explained as a switchable two-parameter Hopf bifurcation. Numerical solutions of the model also reproduce similar oscillatory dynamics to the simulation results.
Rui Liu, Chang Liu, Shuo Wang, Na Deng, Haimin Wang
Sigmoids are one of the most important precursor structures for solar eruptions. In this Letter, we study a sigmoid eruption on 2010 August 1 with EUV data obtained by the Atmospheric Imaging Assembly (AIA) on board the Solar Dynamic Observatory (SDO). In AIA 94 Å (Fe XVIII; 6 MK), topological reconfiguration due to tether-cutting reconnection is unambiguously observed for the first time, i.e., two opposite J-shaped loops reconnect to form a continuous S-shaped loop, whose central portion is dipped and aligned along the magnetic polarity inversion line (PIL), and a compact loop crossing the PIL. A causal relationship between photospheric flows and coronal tether-cutting reconnections is evidenced by the detection of persistent converging flows toward the PIL using line-of-sight magnetograms obtained by the Helioseismic and Magnetic Imager (HMI) on board SDO. The S-shaped loop remains in quasi-equilibrium in the lower corona for about 50 minutes, with the central dipped portion rising slowly at ~10 km s-1. The speed then increases to ~60 km s-1 about 10 minutes prior to the onset of a GOES-class C3.2 flare, as the S-shaped loop speeds up its transformation into an arch-shaped loop, which eventually leads to a loop-like coronal mass ejection (CME). The AIA observations combined with H? filtergrams as well as hard X-ray (HXR) imaging and spectroscopy are consistent with most flare loops being formed by reconnection of the stretched legs of less-sheared J-shaped loops that envelopes the rising flux rope, in agreement with the standard tether-cutting scenario.
Deguang Han, David R. Larson, Bei Liu, Rui Liu
We develop elements of a general dilation theory for operator-valued measures and bounded linear maps between operator algebras that are not necessarily completely-bounded. We prove our main results by extending and generalizing some known results from the theory of frames and framings.