arXiv subjects

Qing Li

Publications and source records attributed to Qing Li.

At least 523 records · Page 29Linked to original sources

arXiv2020

MetaMix: Improved Meta-Learning with Interpolation-based Consistency Regularization

Model-Agnostic Meta-Learning (MAML) and its variants are popular few-shot classification methods. They train an initializer across a variety of sampled learning tasks (also known as episodes) such that the initialized model can adapt quickly to new tasks. However, current MAML-based algorithms have limitations in forming generalizable decision boundaries. In this paper, we propose an approach called MetaMix. It generates virtual feature-target pairs within each episode to regularize the backbone models. MetaMix can be integrated with any of the MAML-based algorithms and learn the decision boundaries generalizing better to new tasks. Experiments on the mini-ImageNet, CUB, and FC100 datasets show that MetaMix improves the performance of MAML-based algorithms and achieves state-of-the-art result when integrated with Meta-Transfer Learning.

cs.CV↗

arXiv2020

An Investigation of Few-Shot Learning in Spoken Term Classification

In this paper, we investigate the feasibility of applying few-shot learning algorithms to a speech task. We formulate a user-defined scenario of spoken term classification as a few-shot learning problem. In most few-shot learning studies, it is assumed that all the N classes are new in a N-way problem. We suggest that this assumption can be relaxed and define a N+M-way problem where N and M are the number of new classes and fixed classes respectively. We propose a modification to the Model-Agnostic Meta-Learning (MAML) algorithm to solve the problem. Experiments on the Google Speech Commands dataset show that our approach outperforms the conventional supervised learning approach and the original MAML.

cs.CL↗

arXiv2020

Unconventional superconductivity induced by suppressing an iron-selenium based Mott insulator CsFe4-xSe4

There are several FeSe based superconductors, including the bulk FeSe, monolayer FeSe thin film, intercalated KxFe2-ySe2 and Li1-xFexOHFeSe, etc. Their normal states all show metallic behavior. The key player here is the FeSe layer which exhibits the highest superconducting transition temperature in the form of monolayer thin film. Recently a new FeSe based compound, CsFe4-xSe4 with the space group of Bmmm was found. Interestingly the system shows a strong insulator-like behavior although it shares the same FeSe planes as other relatives. Density functional theory calculations indicate that it should be a metal, in sharp contrast with the experimental observations. Here we report the emergence of unconventional superconductivity by applying pressure to suppress this insulator-like behavior. At ambient pressure, the insulator-like behavior cannot be modeled as a band insulator, but can be described by the variable-range-hopping model for correlated systems. Furthermore, the specific heat down to 400 mK has been measured and a significant residual coefficient gamma_0=C/T|T->0 is observed, which contrasts the insulator-like state and suggests some quantum freedom of spin dynamics. By applying pressure the insulator-like behavior is gradually suppressed and the system becomes a metal, finally superconductivity is achieved at about 5.1 K. The superconducting transition strongly depends on magnetic field and applied current, indicating a fragile superfluid density. Our results suggest that the superconductivity is established by diluted Cooper pairs on top of a strong correlation background in CsFe4-xSe4.

cond-mat.supr-con↗

arXiv2020

Imaging viscous flow of the Dirac fluid in graphene

The electron-hole plasma in charge-neutral graphene is predicted to realize a quantum critical system whose transport features a universal hydrodynamic description, even at room temperature. This quantum critical "Dirac fluid" is expected to have a shear viscosity close to a minimum bound, with an inter-particle scattering rate saturating at the Planckian time $\hbar/(k_B T)$. While electrical transport measurements at finite carrier density are consistent with hydrodynamic electron flow in graphene, a "smoking gun" of viscous behavior remains elusive. In this work, we directly image viscous Dirac fluid flow in graphene at room temperature via measurement of the associated stray magnetic field. Nanoscale magnetic imaging is performed using quantum spin magnetometers realized with nitrogen vacancy (NV) centers in diamond. Scanning single-spin and wide-field magnetometry reveals a parabolic Poiseuille profile for electron flow in a graphene channel near the charge neutrality point, establishing the viscous transport of the Dirac fluid. This measurement is in contrast to the conventional uniform flow profile imaged in an Ohmic conductor. Via combined imaging-transport measurements, we obtain viscosity and scattering rates, and observe that these quantities are comparable to the universal values expected at quantum criticality. This finding establishes a nearly-ideal electron fluid in neutral graphene at room temperature. Our results pave the way to study hydrodynamic transport in quantum critical fluids relevant to strongly-correlated electrons in high-$T_c$ superconductors. This work also highlights the capability of quantum spin magnetometers to probe correlated-electronic phenomena at the nanoscale.

cond-mat.mes-hall↗

arXiv2020

Closed Loop Neural-Symbolic Learning via Integrating Neural Perception, Grammar Parsing, and Symbolic Reasoning

The goal of neural-symbolic computation is to integrate the connectionist and symbolist paradigms. Prior methods learn the neural-symbolic models using reinforcement learning (RL) approaches, which ignore the error propagation in the symbolic reasoning module and thus converge slowly with sparse rewards. In this paper, we address these issues and close the loop of neural-symbolic learning by (1) introducing the \textbf{grammar} model as a \textit{symbolic prior} to bridge neural perception and symbolic reasoning, and (2) proposing a novel \textbf{back-search} algorithm which mimics the top-down human-like learning procedure to propagate the error through the symbolic reasoning module efficiently. We further interpret the proposed learning framework as maximum likelihood estimation using Markov chain Monte Carlo sampling and the back-search algorithm as a Metropolis-Hastings sampler. The experiments are conducted on two weakly-supervised neural-symbolic tasks: (1) handwritten formula recognition on the newly introduced HWF dataset; (2) visual question answering on the CLEVR dataset. The results show that our approach significantly outperforms the RL methods in terms of performance, converging speed, and data efficiency. Our code and data are released at \url{https://liqing-ustc.github.io/NGS}.

stat.ML↗

arXiv2020

A Competence-aware Curriculum for Visual Concepts Learning via Question Answering

Humans can progressively learn visual concepts from easy to hard questions. To mimic this efficient learning ability, we propose a competence-aware curriculum for visual concept learning in a question-answering manner. Specifically, we design a neural-symbolic concept learner for learning the visual concepts and a multi-dimensional Item Response Theory (mIRT) model for guiding the learning process with an adaptive curriculum. The mIRT effectively estimates the concept difficulty and the model competence at each learning step from accumulated model responses. The estimated concept difficulty and model competence are further utilized to select the most profitable training samples. Experimental results on CLEVR show that with a competence-aware curriculum, the proposed method achieves state-of-the-art performances with superior data efficiency and convergence speed. Specifically, the proposed model only uses 40% of training data and converges three times faster compared with other state-of-the-art methods.

cs.CV↗

arXiv2020

MiniNet: An extremely lightweight convolutional neural network for real-time unsupervised monocular depth estimation

Predicting depth from a single image is an attractive research topic since it provides one more dimension of information to enable machines to better perceive the world. Recently, deep learning has emerged as an effective approach to monocular depth estimation. As obtaining labeled data is costly, there is a recent trend to move from supervised learning to unsupervised learning to obtain monocular depth. However, most unsupervised learning methods capable of achieving high depth prediction accuracy will require a deep network architecture which will be too heavy and complex to run on embedded devices with limited storage and memory spaces. To address this issue, we propose a new powerful network with a recurrent module to achieve the capability of a deep network while at the same time maintaining an extremely lightweight size for real-time high performance unsupervised monocular depth prediction from video sequences. Besides, a novel efficient upsample block is proposed to fuse the features from the associated encoder layer and recover the spatial size of features with the small number of model parameters. We validate the effectiveness of our approach via extensive experiments on the KITTI dataset. Our new model can run at a speed of about 110 frames per second (fps) on a single GPU, 37 fps on a single CPU, and 2 fps on a Raspberry Pi 3. Moreover, it achieves higher depth accuracy with nearly 33 times fewer model parameters than state-of-the-art models. To the best of our knowledge, this work is the first extremely lightweight neural network trained on monocular video sequences for real-time unsupervised monocular depth estimation, which opens up the possibility of implementing deep learning-based real-time unsupervised monocular depth prediction on low-cost embedded devices.

cs.CV↗

arXiv2020

PoseGAN: A Pose-to-Image Translation Framework for Camera Localization

Camera localization is a fundamental requirement in robotics and computer vision. This paper introduces a pose-to-image translation framework to tackle the camera localization problem. We present PoseGANs, a conditional generative adversarial networks (cGANs) based framework for the implementation of pose-to-image translation. PoseGANs feature a number of innovations including a distance metric based conditional discriminator to conduct camera localization and a pose estimation technique for generated camera images as a stronger constraint to improve camera localization performance. Compared with learning-based regression methods such as PoseNet, PoseGANs can achieve better performance with model sizes that are 70% smaller. In addition, PoseGANs introduce the view synthesis technique to establish the correspondence between the 2D images and the scene, \textit{i.e.}, given a pose, PoseGANs are able to synthesize its corresponding camera images. Furthermore, we demonstrate that PoseGANs differ in principle from structure-based localization and learning-based regressions for camera localization, and show that PoseGANs exploit the geometric structures to accomplish the camera localization task, and is therefore more stable than and superior to learning-based regressions which rely on local texture features instead. In addition to camera localization and view synthesis, we also demonstrate that PoseGANs can be successfully used for other interesting applications such as moving object elimination and frame interpolation in video sequences.

cs.CV↗

arXiv2020

Two superconducting components with different symmetries in Nd1-xSrxNiO2 films

The pairing mechanism in cuprates remains as one of the most challenging issues in the field of condensed matter physics. The unique 3d9 electron orbital of the Cu2+ ionic states in cuprates is supposed to be the major player for the occurrence of superconductivity. Recently, superconductivity at about 9-15 K was discovered in infinite layer thin films of nickelate Nd1-xSrxNiO2 (x=0.1-0.2) which is believed to have the similar 3d9 orbital electrons. The key issue concerned here is about the superconducting gap function. Here we report the first set data of single particle tunneling measurements on the superconducting nickelate thin films. We find predominantly two types of tunneling spectra, one shows a V-shape feature which can be fitted very well by a d-wave gap function with gap maximum of about 3.9 meV, another one exhibits a full gap of about 2.35 meV. Some spectra demonstrate mixed contributions of these two components. Our results suggest that the newly found Ni-based superconductors play as close analogs to cuprates, and thus demonstrate the commonality of unconventional superconductivity.

cond-mat.supr-con↗

arXiv2020

Contrasting physical properties of the trilayer nickelates Nd$_4$Ni$_3$O$_{10}$ and Nd$_4$Ni$_3$O$_8$

We report the crystal structures and physical properties of trilayer nickelates Nd$_4$Ni$_3$O$_{10}$ and Nd$_4$Ni$_3$O$_8$. Measurements of magnetization and electrical resistivity display contrasting behaviors in the two compounds. Nd$_4$Ni$_3$O$_{10}$ shows a paramagnetic metallic behavior with a metal to metal phase transition($T^{\ast}$) at about 162 K, as revealed by both magnetic susceptibility and resistivity. Further magnetoresistance and Hall coefficient results show a negative magnetoresistance at low temperatures and the carrier type of Nd$_4$Ni$_3$O$_{10}$ is dominated by hole-type charge carriers. The significant enhancement of Hall coefficient and resistivity below $T^{\ast}$ suggest that effective charge carrier density decreases when cooling through the transition temperature. In contrast, Nd$_4$Ni$_3$O$_8$ shows an insulating behavior despite small value of resistivity at room temperature. The compound shows paramagnetic behavior, with the similar magnetic moments as in Nd$_4$Ni$_3$O$_{10}$ derived from the Curie-Weiss fitting. This may suggest that the magnetic moments in both systems are contributed by the Nd ions. By applying pressures up to about 49 GPa, the insulating behavior is still present and becomes even stronger. Our results suggest that the different Ni configurations ($Ni^{1+/2+}$ or $Ni^{2+/3+}$) and competition between localized and itinerant electrons may account for the contrasting behaviors in trilayer nickelates Nd$_4$Ni$_3$O$_{10}$ and Nd$_4$Ni$_3$O$_8$.

cond-mat.mtrl-sci↗

arXiv2020

How does boiling occur in lattice Boltzmann simulations?

In recent years, the lattice Boltzmann (LB) method has been widely employed to simulate boiling phenomena [A. Márkus and G. Házi, Phys. Rev. E 83, 046705 (2011); Biferale et al., Phys. Rev. Lett. 108, 104502 (2012); Li et al., Phys. Rev. E 96, 063303 (2017); Wu et al., Int. J. Heat Mass Transfer 126, 773 (2018)]. However, a very important issue still remains open, i.e., how does boiling occur in the LB simulations? For instance, the existing LB studies showed that the boiling on a hydrophobic surface begins at a lower wall superheat than that on a hydrophilic surface, which qualitatively agrees well with experimental studies, but no one has yet explained how this phenomenon appears in the LB simulations and what happened in the simulations after changing the wettability of the heating surface. In this paper, the LB boiling mechanism is revealed by analyzing boiling on a flat surface with mixed wettability and boiling on a structured surface with homogeneous wettability. Through a theoretical analysis, we demonstrate that, when the same wall superheat is applied, in the LB boiling simulations the fluid density near the heating surface decreases faster on a hydrophobic surface than that on a hydrophilic surface. Accordingly, a lower wall superheat can induce the phase transition from liquid to vapor on a hydrophobic surface than that on a hydrophilic surface. Furthermore, a similar theoretical analysis shows that the fluid density decreases fastest at concave corners in the case of a structured surface with homogeneous wettability, which explains why vapor bubbles are nucleated at concave corners in the LB simulations of boiling on structured surfaces.

physics.flu-dyn↗

arXiv2020

The prescribed time sliding mode control for attitude tracking of spacecraft

With the development of the space missions, there are extensive missions in the demand of the prescribed time convergence. However, it is still a difficult work to combine the prescribed time method with the sliding mode control due to the infinite gain of the prescribed time method while approaching the prescribed time and two periods of sliding mode control. In this paper, a new prescribed time sliding mode control method is proposed for general systems with matched disturbances, from the second-order system to the high-order system. A novel sliding mode variable with explicit time term is designed for achieving the prescribed time convergence. More importantly, as time approaches the prescribed time, the singularity of control input can be avoided. Finally, this paper presents a disturbance observer based prescribed time sliding mode control method for attitude tracking of spacecraft and the efficiency of this method has been verified through the numerical simulations.

eess.SY↗

arXiv2020

A new theory of fluid-solid coupling in a porous medium for application to the ultrasonic evaluation of tissue remodeling using bioelastomers

Bioelastomers have demonstrated tremendous value and potential in the field of tissue repair due to increasing health demands. Improved non-invasive methods are required for monitoring tissue development assisted by bioelastomers. In this paper, we present a novel theory of fluid-solid coupling in a porous medium for application to the ultrasonic evaluation of tissue remodeling using bioelastomers. The common assumption of equal solid and liquid displacements used in the conventional description of a fluid-saturated porous solid cannot be applied to soft media, such as bioelastomers. We revise the geoacoustic theory of Biot to allow for relative motion between a fluid and a solid in an aggregate and derive an expression for a characteristic fluid-solid coupling parameter. Unlike the conventional method, the propagation speed of shear waves observed by ultrasound shear wave elastography is considered a known quantity in the novel theory, and the calculated value of the coupling parameter is used to evaluate the status of tissue repair. The model is validated by analyzing selected cases. The conditions under which the model can be applied are identified. However, further development of the theory is required to extract dynamic parameters that can be used to monitor the entire tissue remodeling process. In this paper, a theoretical approach is developed that can be used to analyze the mechanics of tissue repair. The theory has potential applications in the field of acellular in situ tissue engineering for non-invasive monitoring of the complex mechanical remodeling process of tissue regeneration and bioelastomer degradation.

physics.med-ph↗

arXiv2020

RMB-DPOP: Refining MB-DPOP by Reducing Redundant Inferences

MB-DPOP is an important complete algorithm for solving Distributed Constraint Optimization Problems (DCOPs) by exploiting a cycle-cut idea to implement memory-bounded inference. However, each cluster root in the algorithm is responsible for enumerating all the instantiations of its cycle-cut nodes, which would cause redundant inferences when its branches do not have the same cycle-cut nodes. Additionally, a large number of cycle-cut nodes and the iterative nature of MB-DPOP further exacerbate the pathology. As a result, MB-DPOP could suffer from huge coordination overheads and cannot scale up well. Therefore, we present RMB-DPOP which incorporates several novel mechanisms to reduce redundant inferences and improve the scalability of MB-DPOP. First, using the independence among the cycle-cut nodes in different branches, we distribute the enumeration of instantiations into different branches whereby the number of nonconcurrent instantiations reduces significantly and each branch can perform memory bounded inference asynchronously. Then, taking the topology into the consideration, we propose an iterative allocation mechanism to choose the cycle-cut nodes that cover a maximum of active nodes in a cluster and break ties according to their relative positions in a pseudo-tree. Finally, a caching mechanism is proposed to further reduce unnecessary inferences when the historical results are compatible with the current instantiations. We theoretically show that with the same number of cycle-cut nodes RMB-DPOP requires as many messages as MB-DPOP in the worst case and the experimental results show our superiorities over the state-of-the-art.

cs.MA↗

arXiv2020

Incorporating Effective Global Information via Adaptive Gate Attention for Text Classification

The dominant text classification studies focus on training classifiers using textual instances only or introducing external knowledge (e.g., hand-craft features and domain expert knowledge). In contrast, some corpus-level statistical features, like word frequency and distribution, are not well exploited. Our work shows that such simple statistical information can enhance classification performance both efficiently and significantly compared with several baseline models. In this paper, we propose a classifier with gate mechanism named Adaptive Gate Attention model with Global Information (AGA+GI), in which the adaptive gate mechanism incorporates global statistical features into latent semantic features and the attention layer captures dependency relationship within the sentence. To alleviate the overfitting issue, we propose a novel Leaky Dropout mechanism to improve generalization ability and performance stability. Our experiments show that the proposed method can achieve better accuracy than CNN-based and RNN-based approaches without global information on several benchmarks.

cs.CL↗

arXiv2020

Self-awareness based resource allocation strategy for containment of epidemic spreading

Resource support between individuals is of particular importance in controlling or mitigating epidemic spreading, especially during pandemics. Whereas there remains the question of how we can protect ourselves from being infected while helping others by donating resources in fighting against the epidemic. To answer the question, we propose a novel resource allocation model by considering the awareness of self-protection of individuals. In the model, a tuning parameter is introduced to quantify the reaction strength of individuals when they are aware of the disease. And then, a coupled model of resource allocation and disease spreading is proposed to study the impact of self-awareness on resource allocation and, its impact on the dynamics of epidemic spreading. Through theoretical analysis and extensive Monte Carlo simulations, we find that in the stationary state, the system converges to two states: the whole healthy or the completely infected, which indicates an abrupt increase in the prevalence when there is a shortage of resources. More importantly, we find that too cautious and too selfless for the people during the outbreak of an epidemic are both not suitable for disease control. Through extensive simulations, we find the optimal point, at which there is a maximum value of the epidemic threshold, and an outbreak can be delayed to the greatest extent. At last, we study further the effects of network structure on the coupled dynamics. We find that the degree heterogeneity promotes the outbreak of disease, and the network structure does not alter the optimal phenomenon in behavior response.

physics.soc-ph↗

arXiv2020

LO-Net: Deep Real-time Lidar Odometry

We present a novel deep convolutional network pipeline, LO-Net, for real-time lidar odometry estimation. Unlike most existing lidar odometry (LO) estimations that go through individually designed feature selection, feature matching, and pose estimation pipeline, LO-Net can be trained in an end-to-end manner. With a new mask-weighted geometric constraint loss, LO-Net can effectively learn feature representation for LO estimation, and can implicitly exploit the sequential dependencies and dynamics in the data. We also design a scan-to-map module, which uses the geometric and semantic information learned in LO-Net, to improve the estimation accuracy. Experiments on benchmark datasets demonstrate that LO-Net outperforms existing learning based approaches and has similar accuracy with the state-of-the-art geometry-based approach, LOAM.

cs.CV↗

arXiv2020

A3: An Automatic Topology-Aware Malfunction Detection and Fixation System in Data Center Networks

Link failures and cable miswirings are not uncommon in building data center networks, which prevents the existing automatic address configuration methods from functioning correctly. However, accurately detecting such malfunctions is not an easy task because there could be no observable node degree changes. Fixing or correcting such malfunctions is even harder as almost no work can provide accurate fixation suggestions now. To solve the problems, we design and implement A3, an automatic topology-aware malfunction detection and fixation system. A3 innovatively formulates the problem of finding minimal fixation to the problem of computing minimum graph difference (NP-hard) and solves it in O(k^6) and O(k^3) for any less than k/2 and k/4 undirected link malfunctions for FatTree, respectively. Our evaluation demonstrates that for less than k/2 undirected link malfunctions, A3 is 100% accurate for malfunction detection and provides the minimum fixation result. For greater or equal to k/2 undirected link malfunctions, A3 still has accuracy of about 100% and provides the near optimal fixation result.

cs.NI↗