arXiv subjects

Yu Zhang

Publications and source records attributed to Yu Zhang.

At least 1,369 records · Page 76Linked to original sources

arXiv2020

Robust Design for Intelligent Reflecting Surfaces Assisted MISO Systems

In this work, we study the statistically robust beamforming design for an intelligent reflecting surfaces (IRS) assisted multiple-input single-output (MISO) wireless system under imperfect channel state information (CSI), where the channel estimation errors are assumed to be additive Gaussian. We aim at jointly optimizing the transmit/receive beamformers and IRS phase shifts to minimize the average mean squared error (MSE) at the user. In particular, to tackle the non-convex optimization problem, an efficient algorithm is developed by capitalizing on alternating optimization and majorization-minimization techniques. Simulation results show that the proposed scheme achieves robust MSE performance in the presence of CSI error, and substantially outperforms conventional non-robust methods.

cs.IT↗

arXiv2020

Model-guided Multi-path Knowledge Aggregation for Aerial Saliency Prediction

As an emerging vision platform, a drone can look from many abnormal viewpoints which brings many new challenges into the classic vision task of video saliency prediction. To investigate these challenges, this paper proposes a large-scale video dataset for aerial saliency prediction, which consists of ground-truth salient object regions of 1,000 aerial videos, annotated by 24 subjects. To the best of our knowledge, it is the first large-scale video dataset that focuses on visual saliency prediction on drones. Based on this dataset, we propose a Model-guided Multi-path Network (MM-Net) that serves as a baseline model for aerial video saliency prediction. Inspired by the annotation process in eye-tracking experiments, MM-Net adopts multiple information paths, each of which is initialized under the guidance of a classic saliency model. After that, the visual saliency knowledge encoded in the most representative paths is selected and aggregated to improve the capability of MM-Net in predicting spatial saliency in aerial scenarios. Finally, these spatial predictions are adaptively combined with the temporal saliency predictions via a spatiotemporal optimization algorithm. Experimental results show that MM-Net outperforms ten state-of-the-art models in predicting aerial video saliency.

cs.CV↗

arXiv2020

Top-K Influential Nodes in Social Networks: A Game Perspective

Influence maximization, the fundamental of viral marketing, aims to find top-$K$ seed nodes maximizing influence spread under certain spreading models. In this paper, we study influence maximization from a game perspective. We propose a Coordination Game model, in which every individual makes its decision based on the benefit of coordination with its network neighbors, to study information propagation. Our model serves as the generalization of some existing models, such as Majority Vote model and Linear Threshold model. Under the generalized model, we study the hardness of influence maximization and the approximation guarantee of the greedy algorithm. We also combine several strategies to accelerate the algorithm. Experimental results show that after the acceleration, our algorithm significantly outperforms other heuristics, and it is three orders of magnitude faster than the original greedy method.

cs.AI↗

arXiv2020

Where are the Dangerous Intersections for Pedestrians and Cyclists: A Colocation-Based Approach

Pedestrians and cyclists are vulnerable road users. They are at greater risk for being killed in a crash than other road users. The percentage of fatal crashes that involve a pedestrian or cyclist is higher than the overall percentage of total trips taken by both modes. Because of this risk, finding ways to minimize problematic street environments is critical. Understanding traffic safety spatial patterns and identifying dangerous locations with significantly high crash risks for pedestrians and cyclists is essential in order to design possible countermeasures to improve road safety. This research develops two indicators for examining spatial correlation patterns between elements of the built environment (intersections) and crashes (pedestrian- or cyclist-involved). The global colocation quotient detects the overall connection in an area while the local colocation quotient identifies the locations of high-risk intersections. To illustrate our approach, we applied the methods to inspect the colocation patterns between pedestrian- or cyclist-vehicle crashes and intersections in Houston, Texas and we identified among many intersections the ones that significantly attract crashes. We also scrutinized those intersections, discussed possible attributes leading to high colocation of crashes and proposed corresponding countermeasures.

physics.soc-ph↗

arXiv2020

Spectroscopic evidence for a spin and valley polarized metallic state in a non-magic-angle twisted bilayer graphene

In the magic-angle twisted bilayer graphene (MA-TBG), strong electron-electron (e-e) correlations caused by the band-flattening lead to many exotic quantum phases such as superconductivity, correlated insulator, ferromagnetism, and quantum anomalous Hall effects, when its low-energy van Hove singularities (VHSs) are partially filled. Here our high-resolution scanning tunneling microscope and spectroscopy measurements demonstrate that the e-e correlation in a non-magic-angle TBG with a twist angle θ = 1.49 still plays an important role in determining its electronic properties. Our most interesting observation on that sample is that when one of its VHS is partially filled, the one associated peak in the spectrum splits into four peaks. Our analysis based on the continuum model suggests that such a one-to-four split of the VHS originates from the formation of an interaction-driven spin-valley-polarized metallic state near the VHS, lifting both the spin and valley degeneracies. Our results for this non-magic-angle TBG reveal a new symmetry-breaking phase, which has not been identified in the MA-TBG or in other systems.

cond-mat.mtrl-sci↗

arXiv2020

Attention: to Better Stand on the Shoulders of Giants

Science of science (SciSci) is an emerging discipline wherein science is used to study the structure and evolution of science itself using large data sets. The increasing availability of digital data on scholarly outcomes offers unprecedented opportunities to explore SciSci. In the progress of science, the previously discovered knowledge principally inspires new scientific ideas, and citation is a reasonably good reflection of this cumulative nature of scientific research. The researches that choose potentially influential references will have a lead over the emerging publications. Although the peer review process is the mainly reliable way of predicting a paper's future impact, the ability to foresee the lasting impact based on citation records is increasingly essential in the scientific impact analysis in the era of big data. This paper develops an attention mechanism for the long-term scientific impact prediction and validates the method based on a real large-scale citation data set. The results break conventional thinking. Instead of accurately simulating the original power-law distribution, emphasizing the limited attention can better stand on the shoulders of giants.

cs.SI↗

arXiv2020

ContextNet: Improving Convolutional Neural Networks for Automatic Speech Recognition with Global Context

Convolutional neural networks (CNN) have shown promising results for end-to-end speech recognition, albeit still behind other state-of-the-art methods in performance. In this paper, we study how to bridge this gap and go beyond with a novel CNN-RNN-transducer architecture, which we call ContextNet. ContextNet features a fully convolutional encoder that incorporates global context information into convolution layers by adding squeeze-and-excitation modules. In addition, we propose a simple scaling method that scales the widths of ContextNet that achieves good trade-off between computation and accuracy. We demonstrate that on the widely used LibriSpeech benchmark, ContextNet achieves a word error rate (WER) of 2.1%/4.6% without external language model (LM), 1.9%/4.1% with LM and 2.9%/7.0% with only 10M parameters on the clean/noisy LibriSpeech test sets. This compares to the previous best published system of 2.0%/4.6% with LM and 3.9%/11.3% with 20M parameters. The superiority of the proposed ContextNet model is also verified on a much larger internal dataset.

eess.AS↗

arXiv2020

Conformer: Convolution-augmented Transformer for Speech Recognition

Recently Transformer and Convolution neural network (CNN) based models have shown promising results in Automatic Speech Recognition (ASR), outperforming Recurrent neural networks (RNNs). Transformer models are good at capturing content-based global interactions, while CNNs exploit local features effectively. In this work, we achieve the best of both worlds by studying how to combine convolution neural networks and transformers to model both local and global dependencies of an audio sequence in a parameter-efficient way. To this regard, we propose the convolution-augmented transformer for speech recognition, named Conformer. Conformer significantly outperforms the previous Transformer and CNN based models achieving state-of-the-art accuracies. On the widely used LibriSpeech benchmark, our model achieves WER of 2.1%/4.3% without using a language model and 1.9%/3.9% with an external language model on test/testother. We also observe competitive performance of 2.7%/6.3% with a small model of only 10M parameters.

eess.AS↗

arXiv2020

Scalability in Perception for Autonomous Driving: Waymo Open Dataset

The research community has increasing interest in autonomous driving research, despite the resource intensity of obtaining representative real world data. Existing self-driving datasets are limited in the scale and variation of the environments they capture, even though generalization within and between operating regions is crucial to the overall viability of the technology. In an effort to help align the research community's contributions with real-world self-driving problems, we introduce a new large scale, high quality, diverse dataset. Our new dataset consists of 1150 scenes that each span 20 seconds, consisting of well synchronized and calibrated high quality LiDAR and camera data captured across a range of urban and suburban geographies. It is 15x more diverse than the largest camera+LiDAR dataset available based on our proposed diversity metric. We exhaustively annotated this data with 2D (camera image) and 3D (LiDAR) bounding boxes, with consistent identifiers across frames. Finally, we provide strong baselines for 2D as well as 3D detection and tracking tasks. We further study the effects of dataset size and generalization across geographies on 3D detection methods. Find data, code and more up-to-date information at http://www.waymo.com/open.

cs.CV↗

arXiv2020

Deep Learning for Massive MIMO with 1-Bit ADCs: When More Antennas Need Fewer Pilots

This paper considers uplink massive MIMO systems with 1-bit analog-to-digital converters (ADCs) and develops a deep-learning based channel estimation framework. In this framework, the prior channel estimation observations and deep neural network models are leveraged to learn the non-trivial mapping from quantized received measurements to channels. For that, we derive the sufficient length and structure of the pilot sequence to guarantee the existence of this mapping function. This leads to the interesting, and \textit{counter-intuitive}, observation that when more antennas are employed by the massive MIMO base station, our proposed deep learning approach achieves better channel estimation performance, for the same pilot sequence length. Equivalently, for the same channel estimation performance, this means that when more antennas are employed, fewer pilots are required. This observation is also analytically proved for some special channel models. Simulation results confirm our observations and show that more antennas lead to better channel estimation both in terms of the normalized mean squared error and the achievable signal-to-noise ratio per antenna.

cs.IT↗

arXiv2020

Quantum liquid from strange frustration in the trimer magnet Ba4Ir3O10

Quantum spin systems such as magnetic insulators usually show classical magnetic order, but such classical states can give way to quantum liquids with exotic entanglement through two known mechanisms of frustration: geometric frustration in lattices with triangle motifs, and spin-orbit-coupling frustration in the exactly solvable quantum liquid of Kitaev's honeycomb lattice. Here we present the experimental observation of a new kind of frustrated quantum liquid arising in an unlikely place: the magnetic insulator Ba4Ir3O10 where Ir3O12 trimers form an unfrustrated square lattice. Experimentally we find a quantum liquid state persisting down to 0.2 K that is stabilized by strong antiferromagnetic interaction with Curie-Weiss temperature - 766 K. The astonishing frustration parameter of 3800 is beyond any known iridate thus far. Heat capacity and thermal conductivity are both linear at low temperatures, a familiar feature in metals but here in an insulator pointing to an exotic quantum liquid state. A mere 2% Sr substitution for Ba produces long-range order at 130 K and destroys the linear-T features. Although the Ir4+(5d5) ions in Ba4Ir3O10 appear to form Ir3O12 trimers of face-sharing IrO6 octahedra, we propose that intra-trimer exchange is reduced and the lattice recombines into an array of coupled 1D chains with additional spins. An extreme limit of decoupled 1D chains can explain most but not all of the striking experimental observations, indicating that the inter-chain coupling plays an important role in the novel frustration mechanism leading to this quantum liquid.

cond-mat.str-el↗

arXiv2020

Intervalley quantum interference and measurement of Berry phase in bilayer graphene

Chiral quasiparticles in Bernal-stacked bilayer graphene have valley-contrasting Berry phases of 2π. This nontrival topological structure, associated with the pseudospin winding along a closed Fermi surface, is responsible for various novel electronic properties, such as anti-Klein tunneling, unconventional quantum Hall effect, and valley Hall effect1-6. Here we show that the quantum interference due to intervalley scattering induced by atomic defects/impurities provides further insights into the topological nature of the bilayer graphene. The scattered chiral quasiparticles between distinct valleys with opposite chirality undergoes a rotation of pseudospin that results in the Friedel oscillation with wavefront dislocations. The number of dislocations reflects the information about pseudospin texture and hence can be used to measure the Berry phase7. As demonstrated both experimentally and theoretically, the Friedel oscillation, depending on the atomic defect/impurity at different sublattices, can exhibit N = 4, 2, or 0 additional wavefronts, characterizing the 2π Berry phase of the bilayer graphene. Our results not only provide a comprehensive study of the intervalley quantum interference in bilayer graphene, but also shed lights on the pseudospin physics.

cond-mat.mtrl-sci↗

arXiv2020

A Streaming On-Device End-to-End Model Surpassing Server-Side Conventional Model Quality and Latency

Thus far, end-to-end (E2E) models have not been shown to outperform state-of-the-art conventional models with respect to both quality, i.e., word error rate (WER), and latency, i.e., the time the hypothesis is finalized after the user stops speaking. In this paper, we develop a first-pass Recurrent Neural Network Transducer (RNN-T) model and a second-pass Listen, Attend, Spell (LAS) rescorer that surpasses a conventional model in both quality and latency. On the quality side, we incorporate a large number of utterances across varied domains to increase acoustic diversity and the vocabulary seen by the model. We also train with accented English speech to make the model more robust to different pronunciations. In addition, given the increased amount of training data, we explore a varied learning rate schedule. On the latency front, we explore using the end-of-sentence decision emitted by the RNN-T model to close the microphone, and also introduce various optimizations to improve the speed of LAS rescoring. Overall, we find that RNN-T+LAS offers a better WER and latency tradeoff compared to a conventional model. For example, for the same latency, RNN-T+LAS obtains a 8% relative improvement in WER, while being more than 400-times smaller in model size.

cs.CL↗

arXiv2020

Partially-Typed NER Datasets Integration: Connecting Practice to Theory

While typical named entity recognition (NER) models require the training set to be annotated with all target types, each available datasets may only cover a part of them. Instead of relying on fully-typed NER datasets, many efforts have been made to leverage multiple partially-typed ones for training and allow the resulting model to cover a full type set. However, there is neither guarantee on the quality of integrated datasets, nor guidance on the design of training algorithms. Here, we conduct a systematic analysis and comparison between partially-typed NER datasets and fully-typed ones, in both theoretical and empirical manner. Firstly, we derive a bound to establish that models trained with partially-typed annotations can reach a similar performance with the ones trained with fully-typed annotations, which also provides guidance on the algorithm design. Moreover, we conduct controlled experiments, which shows partially-typed datasets leads to similar performance with the model trained with the same amount of fully-typed annotations

cs.LG↗

arXiv2020

Learning an Adaptive Model for Extreme Low-light Raw Image Processing

Low-light images suffer from severe noise and low illumination. Current deep learning models that are trained with real-world images have excellent noise reduction, but a ratio parameter must be chosen manually to complete the enhancement pipeline. In this work, we propose an adaptive low-light raw image enhancement network to avoid parameter-handcrafting and to improve image quality. The proposed method can be divided into two sub-models: Brightness Prediction (BP) and Exposure Shifting (ES). The former is designed to control the brightness of the resulting image by estimating a guideline exposure time $t_1$. The latter learns to approximate an exposure-shifting operator $ES$, converting a low-light image with real exposure time $t_0$ to a noise-free image with guideline exposure time $t_1$. Additionally, structural similarity (SSIM) loss and Image Enhancement Vector (IEV) are introduced to promote image quality, and a new Campus Image Dataset (CID) is proposed to overcome the limitations of the existing datasets and to supervise the training of the proposed model. Using the proposed model, we can achieve high-quality low-light image enhancement from a single raw image. In quantitative tests, it is shown that the proposed method has the lowest Noise Level Estimation (NLE) score compared with the state-of-the-art low-light algorithms, suggesting a superior denoising performance. Furthermore, those tests illustrate that the proposed method is able to adaptively control the global image brightness according to the content of the image scene. Lastly, the potential application in video processing is briefly discussed.

eess.IV↗

arXiv2020

Nucleons pair shell model in M-scheme

The nucleon pair shell model (NPSM) is casted into the so-called M-scheme for the cases with isospin symmetry and without isospin symmetry. The odd system and even system are treated on the same foot. The uncoupled commutators for nucleon-pairs, which are suitable for M-scheme, are given. Explicit formula of matrix elements in M-scheme for overlap, one-body operators, two-body operators are obtained. It is found that the $cpu$ time used in calculating the matrix elements in M-scheme is much shorter than that in the J-scheme of NPSM.

nucl-th↗

arXiv2020

Spontaneous Surface Collapse and Reconstruction in Antiferromagnetic Topological Insulator MnBi$_2$Te$_4$

MnBi$_2$Te$_4$ is an antiferromagnetic topological insulator which stimulates intense interests due to the exotic quantum phenomena and promising device applications. Surface structure is a determinant factor to understand the novel magnetic and topological behavior of MnBi2Te4, yet its precise atomic structure remains elusive. Here, we discovered a spontaneous surface collapse and reconstruction in few-layer MnBi2Te4 exfoliated under delicate protection. Instead of the ideal septuple-layer structure in the bulk, the collapsed surface is shown to reconstruct as Mn-doped Bi$_2$Te$_3$ quintuple-layer and Mn$_x$Bi$_y$Te double-layer with a clear van der Waals gap in between. Combining with first-principles calculations, such spontaneous surface collapse is attributed to the abundant intrinsic Mn-Bi antisite defects and tellurium vacancy in the exfoliated surface, which is further supported by in-situ annealing and electron irradiation experiments. Our results shed light on the understanding of the intricate surface-bulk correspondence of MnBi$_2$Te$_4$, and provide insightful perspective of the surface-related quantum measurements in MnBi$_2$Te$_4$ few-layer devices.

cond-mat.mtrl-sci↗

arXiv2020

Learning Event-Based Motion Deblurring

Recovering sharp video sequence from a motion-blurred image is highly ill-posed due to the significant loss of motion information in the blurring process. For event-based cameras, however, fast motion can be captured as events at high time rate, raising new opportunities to exploring effective solutions. In this paper, we start from a sequential formulation of event-based motion deblurring, then show how its optimization can be unfolded with a novel end-to-end deep architecture. The proposed architecture is a convolutional recurrent neural network that integrates visual and temporal knowledge of both global and local scales in principled manner. To further improve the reconstruction, we propose a differentiable directional event filtering module to effectively extract rich boundary prior from the stream of events. We conduct extensive experiments on the synthetic GoPro dataset and a large newly introduced dataset captured by a DAVIS240C camera. The proposed approach achieves state-of-the-art reconstruction quality, and generalizes better to handling real-world motion blur.

cs.CV↗