Tsinghua Science and Technology


  • Anti-Interference Low-Power Double-Edge Triggered Flip-Flop Based on C-Elements

    Zhengfeng Huang;Xiao Yang;Tai Song;Haochen Qi;Yiming Ouyang;Tianming Ni;Qi Xu;

    When the input signal has been interfered and glitches occur, the power consumption of Double-Edge Triggered Flip-Flops(DETFFs) will significantly increase. To effectively reduce the power consumption, this paper presents an anti-interference low-power DETFF based on C-elements. The improved C-element is used in this DETFF, which effectively blocks the glitches in the input signal, prevents redundant transitions inside the DETFF,and reduces the charge and discharge frequencies of the transistor. The C-element has also added pull-up and pull-down paths, reducing its latency. Compared with other existing DETFFs, the DETFF proposed in this paper only flips once on the clock edge, which greatly reduces the redundant transitions caused by glitches and effectively reduces power consumption. This paper uses HSPICE to simulate the proposed DETFF and other 10 DETFFs.The findings show that compared with the other 10 types of DETFFs, the proposed DETFF has achieved large performance indexes in the total power consumption, total power consumption with glitches, delays, and power delay product. A detailed analysis of variance indicates that the proposed DETFF features less sensitivity to process,voltage, temperature, and Negative Bias Temperature Instability(NBTI)-induced aging variations.

    2022年01期 v.27 1-12页 [查看摘要][在线阅读][下载 1325K]
    [下载次数:54 ] |[网刊下载次数:0 ] |[引用频次:0 ] |[阅读次数:1 ]
  • I-Neat: An Intelligent Framework for Adaptive Virtual Machine Consolidation

    Yanxin Liu;Yao Zhao;Jian Dong;Lianpeng Li;Chunpei Wang;Decheng Zuo;

    With the increasing use of cloud computing, high energy consumption has become one of the major challenges in cloud data centers. Virtual Machine(VM) consolidation has been proven to be an efficient way to optimize energy consumption in data centers, and many research works have proposed to optimize VM consolidation.However, the performance of different algorithms is related with the characteristics of the workload and system status;some algorithms are suitable for Central Processing Unit(CPU)-intensive workload and some for web application workload. Therefore, an adaptive VM consolidation framework is necessary to fully explore the potential of these algorithms. Neat is an open-source dynamic VM consolidation framework, which is well integrated into OpenStack.However, it cannot conduct dynamic algorithm scheduling, and VM consolidation algorithms in Neat are few and basic, which results in low performance for energy saving and Service-Level Agreement(SLA) avoidance. In this paper, an Intelligent Neat framework(I-Neat) is proposed, which adds an intelligent scheduler using reinforcement learning and a framework manager to improve the usability of the system. The scheduler can select appropriate algorithms for the local manager from an algorithm library with many load detection algorithms. The algorithm library is designed based on a template, and in addition to the algorithms of Neat, I-Neat adds six new algorithms to the algorithm library. Furthermore, the framework manager helps users add self-defined algorithms to I-Neat without modifying the source code. Our experimental results indicate that the intelligent scheduler and these novel algorithms can effectively reduce energy consumption with SLA assurance.

    2022年01期 v.27 13-26页 [查看摘要][在线阅读][下载 7273K]
    [下载次数:21 ] |[网刊下载次数:0 ] |[引用频次:1 ] |[阅读次数:1 ]
  • Mutation Testing for Integer Overflow in Ethereum Smart Contracts

    Jinlei Sun;Song Huang;Changyou Zheng;Tingyong Wang;Cheng Zong;Zhanwei Hui;

    Integer overflow is a common vulnerability in Ethereum Smart Contracts(ESCs) and often causes huge economic losses. Smart contracts cannot be changed once it is deployed on the blockchain and thus demand further testing. Mutation testing is a fault-based testing method that can effectively improve the sufficiency of a test for smart contracts. However, existing methods cannot efficiently perform mutation testing specifically for integer overflow in ESCs. Therefore, by analyzing integer overflow in ESCs, we propose five special mutation operators to address such vulnerability in terms of detecting sufficiency in ESC testing. An empirical study on 40 open-source ESCs is conducted to evaluate the effectiveness of the proposed mutation operators. Results show that(1) our proposed mutation operators can reproduce all 179 integer overflow vulnerabilities in 40 smart contracts, and the generated mutants have high compilation pass rate and integer overflow vulnerability generation rate; moreover,(2) the generated mutants can find the shortcomings of existing testing methods for integer overflow vulnerability,thereby providing effective support to improve the sufficiency of the test.

    2022年01期 v.27 27-40页 [查看摘要][在线阅读][下载 1636K]
    [下载次数:42 ] |[网刊下载次数:0 ] |[引用频次:2 ] |[阅读次数:1 ]
  • A Novel Cross-Project Software Defect Prediction Algorithm Based on Transfer Learning

    Shiqi Tang;Song Huang;Changyou Zheng;Erhu Liu;Cheng Zong;Yixian Ding;

    Software Defect Prediction(SDP) technology is an effective tool for improving software system quality that has attracted much attention in recent years. However, the prediction of cross-project data remains a challenge for the traditional SDP method due to the different distributions of the training and testing datasets. Another major difficulty is the class imbalance issue that must be addressed in Cross-Project Defect Prediction(CPDP). In this work,we propose a transfer-leaning algorithm(TSboostDF) that considers both knowledge transfer and class imbalance for CPDP. The experimental results demonstrate that the performance achieved by TSboostDF is better than those of existing CPDP methods.

    2022年01期 v.27 41-57页 [查看摘要][在线阅读][下载 1148K]
    [下载次数:58 ] |[网刊下载次数:0 ] |[引用频次:1 ] |[阅读次数:1 ]


  • Metabolite-Disease Association Prediction Algorithm Combining DeepWalk and Random Forest

    Jiaojiao Tie;Xiujuan Lei;Yi Pan;

    Identifying the association between metabolites and diseases will help us understand the pathogenesis of diseases, which has great significance in diagnosing and treating diseases. However, traditional biometric methods are time consuming and expensive. Accordingly, we propose a new metabolite-disease association prediction algorithm based on DeepWalk and random forest(DWRF), which consists of the following key steps:First, the semantic similarity and information entropy similarity of diseases are integrated as the final disease similarity. Similarly, molecular fingerprint similarity and information entropy similarity of metabolites are integrated as the final metabolite similarity. Then, DeepWalk is used to extract metabolite features based on the network of metabolite-gene associations. Finally, a random forest algorithm is employed to infer metabolite-disease associations.The experimental results show that DWRF has good performances in terms of the area under the curve value,leave-one-out cross-validation, and five-fold cross-validation. Case studies also indicate that DWRF has a reliable performance in metabolite-disease association prediction.

    2022年01期 v.27 58-67页 [查看摘要][在线阅读][下载 2254K]
    [下载次数:37 ] |[网刊下载次数:0 ] |[引用频次:1 ] |[阅读次数:1 ]
  • Combining Residual Attention Mechanisms and Generative Adversarial Networks for Hippocampus Segmentation

    Hongxia Deng;Yuefang Zhang;Ran Li;Chunxiang Hu;Zijian Feng;Haifang Li;

    This research discussed a deep learning method based on an improved generative adversarial network to segment the hippocampus. Different convolutional configurations were proposed to capture information obtained by a segmentation network. In addition, a generative adversarial network based on Pixel2Pixel was proposed. The generator was a codec structure combining a residual network and an attention mechanism to capture detailed information. The discriminator used a convolutional neural network to discriminate the segmentation results of the generated model and that of the expert. Through the continuously transmitted losses of the generator and discriminator, the generator reached the optimal state of hippocampus segmentation. T1-weighted magnetic resonance imaging scans and related hippocampus labels of 130 healthy subjects from the Alzheimer's disease Neuroimaging Initiative dataset were used as training and test data; similarity coefficient, sensitivity, and positive predictive value were used as evaluation indicators. Results showed that the network model could achieve an efficient automatic segmentation of the hippocampus and thus has practical relevance for the correct diagnosis of diseases, such as Alzheimer's disease.

    2022年01期 v.27 68-78页 [查看摘要][在线阅读][下载 1909K]
    [下载次数:33 ] |[网刊下载次数:0 ] |[引用频次:3 ] |[阅读次数:1 ]
  • Event Temporal Relation Extraction with Attention Mechanism and Graph Neural Network

    Xiaoliang Xu;Tong Gao;Yuxiang Wang;Xinle Xuan;

    Event temporal relation extraction is an important part of natural language processing. Many models are being used in this task with the development of deep learning. However, most of the existing methods cannot accurately obtain the degree of association between different tokens and events, and event-related information cannot be effectively integrated. In this paper, we propose an event information integration model that integrates event information through multilayer bidirectional long short-term memory(Bi-LSTM) and attention mechanism.Although the above scheme can improve the extraction performance, it can still be further optimized. To further improve the performance of the previous scheme, we propose a novel relational graph attention network that incorporates edge attributes. In this approach, we first build a semantic dependency graph through dependency parsing, model a semantic graph that considers the edges' attributes by using top-k attention mechanisms to learn hidden semantic contextual representations, and finally predict event temporal relations. We evaluate proposed models on the TimeBank-Dense dataset. Compared to previous baselines, the Micro-F1 scores obtained by our models improve by 3.9% and 14.5%, respectively.

    2022年01期 v.27 79-90页 [查看摘要][在线阅读][下载 3141K]
    [下载次数:43 ] |[网刊下载次数:0 ] |[引用频次:5 ] |[阅读次数:7 ]
  • A Dynamic and Deadline-Oriented Road Pricing Mechanism for Urban Traffic Management

    Jiahui Jin;Xiaoxuan Zhu;Biwei Wu;Jinghui Zhang;Yuxiang Wang;

    Road pricing is an urban traffic management mechanism to reduce traffic congestion. Currently, most of the road pricing systems based on predefined charging tolls fail to consider the dynamics of urban traffic flows and travelers' demands on the arrival time. In this paper, we propose a method to dynamically adjust online road toll based on traffic conditions and travelers' demands to resolve the above-mentioned problems. The method, based on deep reinforcement learning, automatically allocates the optimal toll for each road during peak hours and guides vehicles to roads with lower toll charges. Moreover, it further considers travelers' demands to ensure that more vehicles arrive at their destinations before their estimated arrival time. Our method can increase the traffic volume effectively, as compared to the existing static mechanisms.

    2022年01期 v.27 91-102页 [查看摘要][在线阅读][下载 6821K]
    [下载次数:48 ] |[网刊下载次数:0 ] |[引用频次:1 ] |[阅读次数:1 ]


  • Two-Stage Lesion Detection Approach Based on Dimension-Decomposition and 3D Context

    Jiacheng Jiao;Haiwei Pan;Chunling Chen;Tao Jin;Yang Dong;Jingyi Chen;

    Lesion detection in Computed Tomography(CT) images is a challenging task in the field of computer-aided diagnosis. An important issue is to locate the area of lesion accurately. As a branch of Convolutional Neural Networks(CNNs), 3D Context-Enhanced(3DCE) frameworks are designed to detect lesions on CT scans. The False Positives(FPs) detected in 3DCE frameworks are usually caused by inaccurate region proposals, which slow down the inference time. To solve the above problems, a new method is proposed, a dimension-decomposition region proposal network is integrated into 3DCE framework to improve the location accuracy in lesion detection.Without the restriction of "anchors" on ratios and scales, anchors are decomposed to independent "anchor strings".Anchor segments are dynamically combined in accordance with probability, and anchor strings with different lengths dynamically compose bounding boxes. Experiments show that the accurate region proposals generated by our model promote the sensitivity of FPs and spend less inference time compared with the current methods.

    2022年01期 v.27 103-113页 [查看摘要][在线阅读][下载 7106K]
    [下载次数:16 ] |[网刊下载次数:0 ] |[引用频次:0 ] |[阅读次数:1 ]
  • Increasing Momentum-Like Factors: A Method for Reducing Training Errors on Multiple GPUs

    Yu Tang;Zhigang Kan;Lujia Yin;Zhiquan Lai;Zhaoning Zhang;Linbo Qiao;Dongsheng Li;

    In distributed training, increasing batch size can improve parallelism, but it can also bring many difficulties to the training process and cause training errors. In this work, we investigate the occurrence of training errors in theory and train ResNet-50 on CIFAR-10 by using Stochastic Gradient Descent(SGD) and Adaptive moment estimation(Adam) while keeping the total batch size in the parameter server constant and lowering the batch size on each Graphics Processing Unit(GPU). A new method that considers momentum to eliminate training errors in distributed training is proposed. We define a Momentum-like Factor(MF) to represent the influence of former gradients on parameter updates in each iteration. Then, we modify the MF values and conduct experiments to explore how different MF values influence the training performance based on SGD, Adam, and Nesterov accelerated gradient.Experimental results reveal that increasing MFs is a reliable method for reducing training errors in distributed training.The analysis of convergent conditions in distributed training with consideration of a large batch size and multiple GPUs is presented in this paper.

    2022年01期 v.27 114-126页 [查看摘要][在线阅读][下载 1694K]
    [下载次数:26 ] |[网刊下载次数:0 ] |[引用频次:0 ] |[阅读次数:1 ]
  • IDEA: A Utility-Enhanced Approach to Incomplete Data Stream Anonymization

    Lu Yang;Xingshu Chen;Yonggang Luo;Xiao Lan;Wei Wang;

    The prevalence of missing values in the data streams collected in real environments makes them impossible to ignore in the privacy preservation of data streams. However, the development of most privacy preservation methods does not consider missing values. A few researches allow them to participate in data anonymization but introduce extra considerable information loss. To balance the utility and privacy preservation of incomplete data streams, we present a utility-enhanced approach for Incomplete Data strEam Anonymization(IDEA). In this approach,a slide-window-based processing framework is introduced to anonymize data streams continuously, in which each tuple can be output with clustering or anonymized clusters. We consider the dimensions of attribute and tuple as the similarity measurement, which enables the clustering between incomplete records and complete records and generates the cluster with minimal information loss. To avoid the missing value pollution, we propose a generalization method that is based on maybe match for generalizing incomplete data. The experiments conducted on real datasets show that the proposed approach can efficiently anonymize incomplete data streams while effectively preserving utility.

    2022年01期 v.27 127-140页 [查看摘要][在线阅读][下载 988K]
    [下载次数:22 ] |[网刊下载次数:0 ] |[引用频次:0 ] |[阅读次数:1 ]
  • Sensitivity of N400 Effect During Speech Comprehension Under the Uni-and Bi-Modality Conditions

    Yanfei Lin;Zhiwen Liu;Xiaorong Gao;

    N400 is an objective electrophysiological index in semantic processing for brain. This study focuses on the sensitivity of N400 effect during speech comprehension under the uni-and bi-modality conditions. Varying the Signal-to-Noise Ratio(SNR) of speech signal under the conditions of Audio-only(A), Visual-only(V, i.e., lip-reading),and Audio-Visual(AV), the semantic priming paradigm is used to evoke N400 effect and measure the speech recognition rate. For the conditions A and high SNR AV, the N400 amplitudes in the central region are larger; for the conditions of V and low SNR AV, the N400 amplitudes in the left-frontal region are larger. The N400 amplitudes of frontal and central regions under the conditions of A, AV, and V are consistent with speech recognition rate of behavioral results. These results indicate that audio-cognition is better than visual-cognition at high SNR, and visual-cognition is better than audio-cognition at low SNR.

    2022年01期 v.27 141-149页 [查看摘要][在线阅读][下载 6662K]
    [下载次数:40 ] |[网刊下载次数:0 ] |[引用频次:0 ] |[阅读次数:1 ]
  • Enriching the Transfer Learning with Pre-Trained Lexicon Embedding for Low-Resource Neural Machine Translation

    Mieradilijiang Maimaiti;Yang Liu;Huanbo Luan;Maosong Sun;

    Most State-Of-The-Art(SOTA) Neural Machine Translation(NMT) systems today achieve outstanding results based only on large parallel corpora. The large-scale parallel corpora for high-resource languages is easily obtainable. However, the translation quality of NMT for morphologically rich languages is still unsatisfactory, mainly because of the data sparsity problem encountered in Low-Resource Languages(LRLs). In the low-resource NMT paradigm, Transfer Learning(TL) has been developed into one of the most efficient methods. It is difficult to train the model on high-resource languages to include the information in both parent and child models, as well as the initially trained model that only contains the lexicon features and word embeddings of the parent model instead of the child languages feature. In this work, we aim to address this issue by proposing the language-independent Hybrid Transfer Learning(HTL) method for LRLs by sharing lexicon embedding between parent and child languages without leveraging back translation or manually injecting noises. First, we train the High-Resource Languages(HRLs) as the parent model with its vocabularies. Then, we combine the parent and child language pairs using the oversampling method to train the hybrid model initialized by the previously parent model. Finally, we fine-tune the morphologically rich child model using a hybrid model. Besides, we explore some exciting discoveries on the original TL approach.Experimental results show that our model consistently outperforms five SOTA methods in two languages Azerbaijani(Az) and Uzbek(Uz). Meanwhile, our approach is practical and significantly better, achieving improvements of up to 4:94 and 4:84 BLEU points for low-resource child languages Az ! Zh and Uz ! Zh, respectively.

    2022年01期 v.27 150-163页 [查看摘要][在线阅读][下载 3346K]
    [下载次数:74 ] |[网刊下载次数:0 ] |[引用频次:4 ] |[阅读次数:1 ]
  • An MPI+OpenACC-Based PRM Scalar Advection Scheme in the GRAPES Model over a Cluster with Multiple CPUs and GPUs

    Huadong Xiao;Yang Lu;Jianqiang Huang;Wei Xue;

    A moisture advection scheme is an essential module of a numerical weather/climate model representing the horizontal transport of water vapor. The Piecewise Rational Method(PRM) scalar advection scheme in the Global/Regional Assimilation and Prediction System(GRAPES) solves the moisture flux advection equation based on PRM. Computation of the scalar advection involves boundary exchange, and computation of higher bandwidth requirements is complicated and time-consuming in GRAPES. Recently, Graphics Processing Units(GPUs) have been widely used to solve scientific and engineering computing problems owing to advancements in GPU hardware and related programming models such as CUDA/OpenCL and Open Accelerator(OpenACC). Herein, we present an accelerated PRM scalar advection scheme with Message Passing Interface(MPI) and OpenACC to fully exploit GPUs' power over a cluster with multiple Central Processing Units(CPUs) and GPUs, together with optimization of various parameters such as minimizing data transfer, memory coalescing, exposing more parallelism, and overlapping computation with data transfers. Results show that about 3.5 times speedup is obtained for the entire model running at medium resolution with double precision when comparing the scheme's elapsed time on a node with two GPUs(NVIDIA P100) and two 16-core CPUs(Intel Gold 6142). Further, results obtained from experiments of a higher resolution model with multiple GPUs show excellent scalability.

    2022年01期 v.27 164-173页 [查看摘要][在线阅读][下载 806K]
    [下载次数:35 ] |[网刊下载次数:0 ] |[引用频次:0 ] |[阅读次数:1 ]
  • SIGNGD with Error Feedback Meets Lazily Aggregated Technique:Communication-Efficient Algorithms for Distributed Learning

    Xiaoge Deng;Tao Sun;Feng Liu;Dongsheng Li;

    The proliferation of massive datasets has led to significant interests in distributed algorithms for solving large-scale machine learning problems. However, the communication overhead is a major bottleneck that hampers the scalability of distributed machine learning systems. In this paper, we design two communication-efficient algorithms for distributed learning tasks. The first one is named EF-SIGNGD, in which we use the 1-bit(sign-based) gradient quantization method to save the communication bits. Moreover, the error feedback technique, i.e., incorporating the error made by the compression operator into the next step, is employed for the convergence guarantee. The second algorithm is called LE-SIGNGD, in which we introduce a well-designed lazy gradient aggregation rule to EF-SIGNGD that can detect the gradients with small changes and reuse the outdated information. LE-SIGNGD saves communication costs both in transmitted bits and communication rounds. Furthermore, we show that LE-SIGNGD is convergent under some mild assumptions. The effectiveness of the two proposed algorithms is demonstrated through experiments on both real and synthetic data.

    2022年01期 v.27 174-185页 [查看摘要][在线阅读][下载 2721K]
    [下载次数:17 ] |[网刊下载次数:0 ] |[引用频次:0 ] |[阅读次数:1 ]
  • CAN: Effective Cross Features by Global Attention Mechanism and Neural Network for Ad Click Prediction

    Wenjie Cai;Yufeng Wang;Jianhua Ma;Qun Jin;

    Online advertising click-through rate(CTR) prediction is aimed at predicting the probability of a user clicking an ad, and it has undergone considerable development in recent years. One of the hot topics in this area is the construction of feature interactions to facilitate accurate prediction. Factorization machine provides second-order feature interactions by linearly multiplying hidden feature factors. However, real-world data present a complex and nonlinear structure. Hence, second-order feature interactions are unable to represent cross information adequately. This drawback has been addressed using deep neural networks(DNNs), which enable high-order nonlinear feature interactions. However, DNN-based feature interactions cannot easily optimize deep structures because of the absence of cross information in the original features. In this study, we propose an effective CTR prediction algorithm called CAN, which explicitly exploits the benefits of attention mechanisms and DNN models.The attention mechanism is used to provide rich and expressive low-order feature interactions and facilitate the optimization of DNN-based predictors that implicitly incorporate high-order nonlinear feature interactions. The experiments using two real datasets demonstrate that our proposed CAN model performs better than other cross feature-and DNN-based predictors.

    2022年01期 v.27 186-195页 [查看摘要][在线阅读][下载 1518K]
    [下载次数:46 ] |[网刊下载次数:0 ] |[引用频次:2 ] |[阅读次数:1 ]
  • Optimal Controller Design for Non-Affine Nonlinear Power Systems with Static Var Compensators for Hybrid UAVs

    Yanchu Li;Qingqing Ding;Shufang Li;Stanimir Valtchev;

    A generalized non-affine nonlinear power system model is presented for a single machine bus power system with a Static Var Compensator(SVC) or State Var System(SVS) for hybrid Unmanned Aerial Vehicles(UAVs).The model is constructed by differential algebraic equations on the MATLAB-Simulink platform with the programming technique of its S-Function. Combining the inverse system method and the Linear Quadratic Regulation(LQR), an optimized SVC controller is designed. The simulations under three fault conditions show that the proposed controller can effectively improve the power system transient performance.

    2022年01期 v.27 196-206页 [查看摘要][在线阅读][下载 8932K]
    [下载次数:35 ] |[网刊下载次数:0 ] |[引用频次:0 ] |[阅读次数:1 ]
  • GGC: Gray-Granger Causality Method for Sensor Correlation Network Structure Mining on High-Speed Train

    Jie Man;Honghui Dong;Limin Jia;Yong Qin;

    Vehicle information on high-speed trains can not only determine whether the various parts of the train are working normally, but also predict the train's future operating status. How to obtain valuable information from massive vehicle data is a difficult point. First, we divide the vehicle data of a high-speed train into 13 subsystem datasets,according to the functions of the collection components. Then, according to the gray theory and the Granger causality test, we propose the Gray-Granger Causality(GGC) model, which can construct a vehicle information network on the basis of the correlation between the collection components. By using the complex network theory to mine vehicle information and its subsystem networks, we find that the vehicle information network and its subsystem networks have the characteristics of a scale-free network. In addition, the vehicle information network is weak against attacks,but the subsystem network is closely connected and strong against attacks.

    2022年01期 v.27 207-222页 [查看摘要][在线阅读][下载 10703K]
    [下载次数:20 ] |[网刊下载次数:0 ] |[引用频次:1 ] |[阅读次数:1 ]
  • 下载本期数据