arXiv subjects

Yue Zhang

Publications and source records attributed to Yue Zhang.

At least 757 records · Page 42Linked to original sources

arXiv2018

The high-frequency dynamics of domain walls with strong Dzyaloshinskii-Moriya interaction

Domain walls (DWs) in perpendicularly magnetized nanotracks (PMNTs) with interfacial Dzyaloshinskii-Moriya interaction (DMI) have become the primary objects of theoretical and experimental interest due to their technological suitability in spintronic nanodevices. Chiral DWs in PMNTs can be driven efficiently by the spin-orbit torque. However, the high-frequency dynamic behavior of the chiral DW has not been explored. In this work, using micromagnetic calculation, we have discovered a novel dynamic mode, the sway mode, of DWs under an out-of-plane high-frequency alternating current (AC) magnetic field in a PMNT with strong DMI. This dynamic phenomenon is strictly related with DMI-related boundary effect and can be understood in terms of the propagation of an amplitude-tuned spin wave in the DW plane. The spin wave exhibits some characteristic frequencies due to the space-confinement of DW. This work offers the possibility of a visual route for characterizing DMI.

physics.comp-ph↗

arXiv2018

Subword Encoding in Lattice LSTM for Chinese Word Segmentation

We investigate a lattice LSTM network for Chinese word segmentation (CWS) to utilize words or subwords. It integrates the character sequence features with all subsequences information matched from a lexicon. The matched subsequences serve as information shortcut tunnels which link their start and end characters directly. Gated units are used to control the contribution of multiple input links. Through formula derivation and comparison, we show that the lattice LSTM is an extension of the standard LSTM with the ability to take multiple inputs. Previous lattice LSTM model takes word embeddings as the lexicon input, we prove that subword encoding can give the comparable performance and has the benefit of not relying on any external segmentor. The contribution of lattice LSTM comes from both lexicon and pretrained embeddings information, we find that the lexicon information contributes more than the pretrained embeddings information through controlled experiments. Our experiments show that the lattice structure with subword encoding gives competitive or better results with previous state-of-the-art methods on four segmentation benchmarks. Detailed analyses are conducted to compare the performance of word encoding and subword encoding in lattice LSTM. We also investigate the performance of lattice LSTM structure under different circumstances and when this model works or fails.

cs.CL↗

arXiv2018

Social Vehicle Swarms: A Novel Perspective on Social-aware Vehicular Communication Architecture

Internet of vehicles is a promising area related to D2D communication and internet of things. We present a novel perspective for vehicular communications, social vehicle swarms, to study and analyze socially aware internet of vehicles with the assistance of an agent-based model intended to reveal hidden patterns behind superficial data. After discussing its components, namely its agents, environments, and rules, we introduce supportive technology and methods, deep reinforcement learning, privacy preserving data mining and sub-cloud computing, in order to detect the most significant and interesting information for each individual effectively, which is the key desire. Finally, several relevant research topics and challenges are discussed.

cs.CY↗

arXiv2018

Vehicle Tracking Using Surveillance with Multimodal Data Fusion

Vehicle location prediction or vehicle tracking is a significant topic within connected vehicles. This task, however, is difficult if only a single modal data is available, probably causing bias and impeding the accuracy. With the development of sensor networks in connected vehicles, multimodal data are becoming accessible. Therefore, we propose a framework for vehicle tracking with multimodal data fusion. Specifically, we fuse the results of two modalities, images and velocity, in our vehicle-tracking task. Images, being processed in the module of vehicle detection, provide direct information about the features of vehicles, whereas velocity estimation can further evaluate the possible location of the target vehicles, which reduces the number of features being compared, and decreases the time consumption and computational cost. Vehicle detection is designed with a color-faster R-CNN, which takes both the shape and color of the vehicles into consideration. Meanwhile, velocity estimation is through the Kalman filter, which is a classical method for tracking. Finally, a multimodal data fusion method is applied to integrate these outcomes so that vehicle-tracking tasks can be achieved. Experimental results suggest the efficiency of our methods, which can track vehicles using a series of surveillance cameras in urban areas.

cs.CV↗

arXiv2018

Achieving Efficient and Secure Data Acquisition for Cloud-supported Internet of Things in Smart Grid

Cloud-supported Internet of Things (Cloud-IoT) has been broadly deployed in smart grid systems. The IoT front-ends are responsible for data acquisition and status supervision, while the substantial amount of data is stored and managed in the cloud server. Achieving data security and system efficiency in the data acquisition and transmission process are of great significance and challenging, because the power grid-related data is sensitive and in huge amount. In this paper, we present an efficient and secure data acquisition scheme based on CP-ABE (Ciphertext Policy Attribute Based Encryption). Data acquired from the terminals will be partitioned into blocks and encrypted with its corresponding access sub-tree in sequence, thereby the data encryption and data transmission can be processed in parallel. Furthermore, we protect the information about the access tree with threshold secret sharing method, which can preserve the data privacy and integrity from users with the unauthorized sets of attributes. The formal analysis demonstrates that the proposed scheme can fulfill the security requirements of the Cloud-supported IoT in smart grid. The numerical analysis and experimental results indicate that our scheme can effectively reduce the time cost compared with other popular approaches.

cs.CR↗

arXiv2018

Neural Transition-based Syntactic Linearization

The task of linearization is to find a grammatical order given a set of words. Traditional models use statistical methods. Syntactic linearization systems, which generate a sentence along with its syntactic tree, have shown state-of-the-art performance. Recent work shows that a multi-layer LSTM language model outperforms competitive statistical syntactic linearization systems without using syntax. In this paper, we study neural syntactic linearization, building a transition-based syntactic linearizer leveraging a feed-forward neural network, observing significantly better results compared to LSTM language models on this task.

cs.CL↗

arXiv2018

Prostate Segmentation using 2D Bridged U-net

In this paper, we focus on three problems in deep learning based medical image segmentation. Firstly, U-net, as a popular model for medical image segmentation, is difficult to train when convolutional layers increase even though a deeper network usually has a better generalization ability because of more learnable parameters. Secondly, the exponential ReLU (ELU), as an alternative of ReLU, is not much different from ReLU when the network of interest gets deep. Thirdly, the Dice loss, as one of the pervasive loss functions for medical image segmentation, is not effective when the prediction is close to ground truth and will cause oscillation during training. To address the aforementioned three problems, we propose and validate a deeper network that can fit medical image datasets that are usually small in the sample size. Meanwhile, we propose a new loss function to accelerate the learning process and a combination of different activation functions to improve the network performance. Our experimental results suggest that our network is comparable or superior to state-of-the-art methods.

cs.CV↗

arXiv2018

Dark Tridents at Off-Axis Liquid Argon Neutrino Detectors

We present dark tridents, a new channel for exploring dark sectors in short-baseline neutrino experiments. Dark tridents are clean, distinct events where, like neutrino tridents, the scattering of a very weakly coupled particle leads to the production of a lepton--antilepton pair. Dark trident production occurs in models where long-lived dark-sector particles are produced along with the neutrinos in a beam-dump environment and interact with neutrino detectors downstream, producing an on-shell boson which decays into a pair of charged leptons. We focus on a simple model where the dark matter particle interacts with the standard model exclusively through a dark photon, and concentrate on the region of parameter space where the dark photon mass is smaller than twice that of the dark matter particle and hence decays exclusively into standard-model particles. We compute event rates and discuss search strategies for dark tridents from dark matter at the current and upcoming liquid argon detectors aligned with the Booster beam at Fermilab -- MicroBooNE, SBND, and ICARUS -- assuming the dark sector particles are produced off-axis in the higher energy NuMI beam. We find that MicroBooNE has already recorded enough data to be competitive with existing bounds on this dark sector model, and that new regions of parameter space will be probed with future data and experiments.

hep-ph↗

arXiv2018

Exploring Graph-structured Passage Representation for Multi-hop Reading Comprehension with Graph Neural Networks

Multi-hop reading comprehension focuses on one type of factoid question, where a system needs to properly integrate multiple pieces of evidence to correctly answer a question. Previous work approximates global evidence with local coreference information, encoding coreference chains with DAG-styled GRU layers within a gated-attention reader. However, coreference is limited in providing information for rich inference. We introduce a new method for better connecting global evidence, which forms more complex graphs compared to DAGs. To perform evidence integration on our graphs, we investigate two recent graph neural networks, namely graph convolutional network (GCN) and graph recurrent network (GRN). Experiments on two standard datasets show that richer global information leads to better answers. Our method performs better than all published results on these datasets.

cs.CL↗

arXiv2018

A Decentralized Optimal Control Framework for Connected Automated Vehicles at Urban Intersections with Dynamic Resequencing

Earlier work has established a decentralized framework to optimally control Connected Automated Vehicles (CAVs) crossing an urban intersection without using explicit traffic signaling while following a strict First-In-First-Out (FIFO) queueing structure. The proposed solution minimizes energy consumption subject to a FIFO-based throughput maximization requirement. In this paper, we extend the solution to account for asymmetric intersections by relaxing the FIFO constraint and including a dynamic resequencing process so as to maximize traffic throughput. To investigate the tradeoff between throughput maximization and energy minimization objectives, we exploit several alternative problem formulations. In addition, the computational complexity of the resequencing process is analyzed and proved to be bounded, which makes the online implementation computationally feasible. The effectiveness of the dynamic resequencing process in terms of throughput maximization is illustrated through simulation

math.OC↗

arXiv2018

Two Local Models for Neural Constituent Parsing

Non-local features have been exploited by syntactic parsers for capturing dependencies between sub output structures. Such features have been a key to the success of state-of-the-art statistical parsers. With the rise of deep learning, however, it has been shown that local output decisions can give highly competitive accuracies, thanks to the power of dense neural input representations that embody global syntactic information. We investigate two conceptually simple local neural models for constituent parsing, which make local decisions to constituent spans and CFG rules, respectively. Consistent with previous findings along the line, our best model gives highly competitive results, achieving the labeled bracketing F1 scores of 92.4% on PTB and 87.3% on CTB 5.1.

cs.CL↗

arXiv2018

N-ary Relation Extraction using Graph State LSTM

Cross-sentence $n$-ary relation extraction detects relations among $n$ entities across multiple sentences. Typical methods formulate an input as a \textit{document graph}, integrating various intra-sentential and inter-sentential dependencies. The current state-of-the-art method splits the input graph into two DAGs, adopting a DAG-structured LSTM for each. Though being able to model rich linguistic knowledge by leveraging graph edges, important information can be lost in the splitting procedure. We propose a graph-state LSTM model, which uses a parallel state to model each word, recurrently enriching state values via message passing. Compared with DAG LSTMs, our graph LSTM keeps the original graph structure, and speeds up computation by allowing more parallelization. On a standard benchmark, our model shows the best result in the literature.

cs.CL↗

arXiv2018

A Graph-to-Sequence Model for AMR-to-Text Generation

The problem of AMR-to-text generation is to recover a text representing the same meaning as an input AMR graph. The current state-of-the-art method uses a sequence-to-sequence model, leveraging LSTM for encoding a linearized AMR structure. Although being able to model non-local semantic information, a sequence LSTM can lose information from the AMR graph structure, and thus faces challenges with large graphs, which result in long sequences. We introduce a neural graph-to-sequence model, using a novel LSTM structure for directly encoding graph-level semantics. On a standard benchmark, our model shows superior results to existing methods in the literature.

cs.CL↗

arXiv2018

Readable Racetrack Memory via Ferromagnetically Coupled Chiral Domain Walls

Current-induced motion of domain walls (CIMDW) with interfacial Dzyaloshinskii-Moriya interaction (DMI) in heavy metal (HM)/ferromagnetic (FM) metal multilayers have attracted attention owing to their potential application in novel magnetic memories. In recent years, the CIMDW at ultrahigh speed has been observed in a synthetic antiferromagnetic (SAF) multilayer. However, due to the zero net magnetization, the reading of information from the SAF multilayer is still challenging. In this work, we propose a readable racetrack memory consisting of a synthetic ferromagnetic multilayer composed of two FM layers with an interlayer FM coupling. One FM layer had an isotropic DMI, while the other had an anisotropic DMI. This difference of DMIs resulted in the opposite tilting directions of the DW planes in the two layers. This tilting was inhibited by a strong interlayer FM coupling, resulting in an increase in the DW velocity and the reduction of the minimum allowed spacing between two adjacent DWs. In addition, the FM coupling enhanced the stray field, and the stored information could be read conveniently using a conventional reading head. Therefore, our proposal paves a way for the fabrication of a racetrack memory with high reading speed, large storage density, and good readability.

physics.comp-ph↗

arXiv2018

A post-processing method to improve the white matter hyperintensity segmentation accuracy for randomly-initialized U-net

White matter hyperintensity (WMH) is commonly found in elder individuals and appears to be associated with brain diseases. U-net is a convolutional network that has been widely used for biomedical image segmentation. Recently, U-net has been successfully applied to WMH segmentation. Random initialization is usally used to initialize the model weights in the U-net. However, the model may coverage to different local optima with different randomly initialized weights. We find a combination of thresholding and averaging the outputs of U-nets with different random initializations can largely improve the WMH segmentation accuracy. Based on this observation, we propose a post-processing technique concerning the way how averaging and thresholding are conducted. Specifically, we first transfer the score maps from three U-nets to binary masks via thresholding and then average those binary masks to obtain the final WMH segmentation. Both quantitative analysis (via the Dice similarity coefficient) and qualitative analysis (via visual examinations) reveal the superior performance of the proposed method. This post-processing technique is independent of the model used. As such, it can also be applied to situations where other deep learning models are employed, especially when random initialization is adopted and pre-training is unavailable.

cs.CV↗

arXiv2018

Design Challenges and Misconceptions in Neural Sequence Labeling

We investigate the design challenges of constructing effective and efficient neural sequence labeling systems, by reproducing twelve neural sequence labeling models, which include most of the state-of-the-art structures, and conduct a systematic model comparison on three benchmarks (i.e. NER, Chunking, and POS tagging). Misconceptions and inconsistent conclusions in existing literature are examined and clarified under statistical experiments. In the comparison and analysis process, we reach several practical conclusions which can be useful to practitioners.

cs.CL↗

arXiv2018

Chinese NER Using Lattice LSTM

We investigate a lattice-structured LSTM model for Chinese NER, which encodes a sequence of input characters as well as all potential words that match a lexicon. Compared with character-based methods, our model explicitly leverages word and word sequence information. Compared with word-based methods, lattice LSTM does not suffer from segmentation errors. Gated recurrent cells allow our model to choose the most relevant characters and words from a sentence for better NER results. Experiments on various datasets show that lattice LSTM outperforms both word-based and character-based LSTM baselines, achieving the best results.

cs.CL↗

arXiv2018

NCRF++: An Open-source Neural Sequence Labeling Toolkit

This paper describes NCRF++, a toolkit for neural sequence labeling. NCRF++ is designed for quick implementation of different neural sequence labeling models with a CRF inference layer. It provides users with an inference for building the custom model structure through configuration file with flexible neural feature design and utilization. Built on PyTorch, the core operations are calculated in batch, making the toolkit efficient with the acceleration of GPU. It also includes the implementations of most state-of-the-art neural sequence labeling models such as LSTM-CRF, facilitating reproducing and refinement on those methods.

cs.CL↗