Yi Li

China United Network Communications Group (China)

Author Statistics

Papers

Citation

H-Index

i-10 index

Research Trends

Author Order

Document Type

Co-Authors

Xiaobin Cao

Southwest Jiaotong University

Yiren Long

Southwest Jiaotong University

Pingzhi Fan

Southwest Jiaotong University

Lejia Liu

Southwest Jiaotong University

Yujing Liang

Southwest Jiaotong University

Hongli Gao

Southwest Jiaotong University

Jun Ye

Xiamen University

Xianfu Lei

Southwest Jiaotong University

Liang Guo

Southwest Jiaotong University

Dageng Chen

Huawei Technologies (China)

Cooperative Institutions

Southwest Jiaotong University

China United Network Communications Group (China)

Beijing University of Posts and Telecommunications

Guizhou Electric Power Design and Research Institute

Chongqing University

Guangzhou Metro Design & Research Institute

Peking University

Lancaster University

Chinese Academy of Sciences

University of Chinese Academy of Sciences

Author Statistics

Papers

Citation

H-Index

i-10 index

Research Field

Research on Alarm Cleaning and Outlier Detection Algorithms of Network Data Analysis

Lecture notes in electrical engineering (2024)

Jingyun Wang Deyi Li Yuchao Jin Yuhui Han Yi Li

False alarm

10.1007/978-981-97-2124-5_51

Cite

Citations (0)

Research on Optimization of Operation Organization of Transship Trains in Railway Hub

Smart innovation, systems and technologies (2023)

Zongying Song Yi Li Mengyuan Yue Kun Liu Miaomiao Lv

10.1007/978-981-99-0848-6_13

Cite

Citations (0)

Convolutional Neural Networks on Apache Storm

Wenyu Zhang Yanfeng Lu Yi Li Hong Qiao

the performance of deep learning largely depends on the size of data. One data source is real-time streaming data, produced from mobile devices, sensors or social media, etc. Streaming data is high-speed and large-scale, which needs real-time processing. However, current mainstream frameworks are mainly designed for off-line data. To suit this, we first propose a deep learning framework based on Apache Storm, which is a distributed stream processing frame, fast and fault-tolerant. Our framework implements the distributed training of CNNs. which is different from MMLSpark or TensorFlowOnSpark that is a pure Java implementation. The design of message passing and synchronization is also suitable to other MapReduce-family distributed computing platforms. To validate our work, MNIST and Cifar -10 datasets are used for evaluation and comparison with similar architectures. The results show our framework, in resource-limited environment, realizes about 10 times speedup.

Speedup

MNIST database

Stream Processing

10.1109/cac48633.2019.8996300

Cite

Citations (1)

UNICAD: A Unified Approach for Attack Detection, Noise Reduction and Novel Class Identification

arXiv (Cornell University) (2024)

Alvaro Lopez Pellicer Kittipos Giatgong Yi Li Neeraj Suri Plamen Angelov

As the use of Deep Neural Networks (DNNs) becomes pervasive, their vulnerability to adversarial attacks and limitations in handling unseen classes poses significant challenges. The state-of-the-art offers discrete solutions aimed to tackle individual issues covering specific adversarial attack scenarios, classification or evolving learning. However, real-world systems need to be able to detect and recover from a wide range of adversarial attacks without sacrificing classification accuracy and to flexibly act in {\bf unseen} scenarios. In this paper, UNICAD, is proposed as a novel framework that integrates a variety of techniques to provide an adaptive solution. For the targeted image classification, UNICAD achieves accurate image classification, detects unseen classes, and recovers from adversarial attacks using Prototype and Similarity-based DNNs with denoising autoencoders. Our experiments performed on the CIFAR-10 dataset highlight UNICAD's effectiveness in adversarial mitigation and unseen class classification, outperforming traditional models.

Identification

10.48550/arxiv.2406.16501

Cite

Citations (0)

On the Spectral and Energy Efficiency of Full-Duplex Small-Cell Wireless Systems With Massive MIMO

IEEE Transactions on Vehicular Technology (2016)

Yi Li Pingzhi Fan Anatolii N. Leukhin Lingjia Liu

The achievable rate of full-duplex (FD) small-cell systems with massive multiple-input multiple-output (MIMO), in which a low-power base station (BS) equipped with large antenna arrays sends/receives data to and from multiple half-duplex (HD) users at the same time on the same frequency, is investigated. The BS uses imperfect channel state information (CSI) obtained from received pilots, nonideal hardware, and a linear transmitter and receiver, i.e., zero-forcing (ZF) or maximum-ratio transmission/maximum-ratio combining (MRT/MRC), to process the signals. The approximate closed-form expressions of the achievable rate for both the ZF and MRT/MRC processing are derived and used to analyze the effect of the number of antennas and the hardware imperfection on the self-interference (SI), which is a bottleneck of the FD systems. To maximize the spectral efficiency (SE) and the energy efficiency (EE) of this system, two nonconvex power-allocation optimization problems are formulated and solved by utilizing the sequential convex approximation technique and the fractional programming technique. Two iterative algorithms are proposed with proved local convergence. Numerical results illustrate that the analytical approximation of achievable rate matches well with the Monte Carlo simulation. It is also shown that the ZF processing has greater ability to suppress SI, compared with MRT/MRC processing. The proposed power-allocation algorithms are shown to increase the SE and EE significantly, compared with the uniform power-allocation scheme when the BS is equipped with moderately large antenna arrays.

Spectral Efficiency

Maximal-ratio combining

Transmitter power output

10.1109/tvt.2016.2577636

Cite

Citations (71)

Application Research of Spectral Response Detection Method in Exploring Topological Structure of Grounding Grid

2022 IEEE International Conference on High Voltage Engineering and Applications (ICHVE) (2022)

Jinyang Xie Qiheng Ma Lejia Liu Yi Li Xiaobin Cao

The grounding grid is the basis for maintaining the safe operation of the power system. Proving the health status of the grounding grid under the soil is the top priority of the maintenance of the power system. When the grounding grid is corroded and fractured, its topology will also change accordingly. In this paper, CDEGS was used for modeling, spectral impedance data of ground grid under different topologies were obtained by spectral method, spectral impedance characteristics of various topologies were analyzed, and the changes of spectral impedance characteristics under different geological conditions were studied by changing soil resistivity. Based on the spectral impedance data obtained from the simulation results, the topological structure of the grounding network in the tested area under the soil is analyzed. This research can further reduce the cost of grounding network status detection and promote social development.

10.1109/ichve53725.2022.9961497

Cite

Citations (0)

Application of spectrum detection technology in rail-ground transition resistance of urban rail transit

2022 IEEE International Conference on High Voltage Engineering and Applications (ICHVE) (2022)

Jun Ye Xiaobin Cao Yiren Long Yi Li Lejia Liu

The harm of stray current brought by the rapid development of urban rail transit is becoming more and more serious. The rail-to-ground transition resistance is the most important factor affecting the size and distribution of stray current, and it is the key to study the problem of stray current. Therefore, it is very meaningful to judge the amount of stray current by detecting the magnitude of rail-ground transition resistance. In this paper, by establishing a simulation model of subway rail-drainage network, the influence of transition resistance, soil resistivity and concrete layer resistivity on the impedance amplitude phase-frequency characteristic curve between the rail and the drainage network is simulated and analyzed. The research finds: The change of soil resistivity has no effect on the impedance amplitude-phase-frequency characteristic curve between the rail and the drainage network. The influence of the resistivity of the concrete layer on the impedance amplitude-frequency characteristic curve between the rail and the drainage network is above 300kHz. The influence frequency range of the phase-frequency characteristic curve of the impedance between the grids is above 70kHz; the frequency range of the influence of the change of the transition resistance on the impedance amplitude and phase-frequency characteristic curve between the rail and the discharge grid is 50Hz-1MHz. The method of detecting whether the rail-to-ground transition resistance meets the standard is of great significance to the evaluation and prevention of stray current by using the amplitude-phase-frequency characteristic curve of the impedance between the rail and the drainage network in the frequency section.

Urban rail transit

Stray voltage

10.1109/ichve53725.2022.9961554

Cite

Citations (0)

Simplified Monocular Ranging Anomalous Behaviour Detection Model for Security Monitoring Applications

Frontiers in artificial intelligence and applications (2024)

Qing Liu Yu Wu Jing Nie Yi Li Xiaohu Zheng

A simplified monocular ranging anomalous behaviour detection model is proposed to address the problems of cumbersome calibration of the traditional monocular ranging camera and the difficulty of selecting anomalous behaviour detection models. The model uses the YOLO target detection algorithm to obtain the pixel position of the target ROI region in the image coordinate system, and then converts the pixel position to the world coordinate system by using the vertical and horizontal field of view angles of the camera and the camera resolution. Based on the camera mounting height and tilt angle, the distance between the target and the camera vertical point is calculated and compared with the preset defense zone distance and the model switching distance to determine whether the target needs to be detected for abnormal human behaviour and which abnormal human behaviour detection model should be used for detection. In addition to ranging the target without camera calibration, the model also divides the anomaly detection network for two different human features by the image distortion distance, which improves the convergence speed and recognition accuracy of the anomaly detection network.

Ranging

Position (finance)

Monocular

Distortion (music)

10.3233/faia231251

Cite

Citations (0)

An SCMA-based uplink inter-cell interference cancellation technique for 5G wireless systems

Yi Li Xianfu Lei Pingzhi Fan Dageng Chen

In future 5G wireless networks, inter-cell interference (ICI) is perceived as one of the most critical performance bottlenecks. Traditionally, the interference is treated as an additional source of noise, recent advances in information theory show that interference is not necessarily an opponent, but might be decoded and cancelled to obtain a larger capacity region. In this paper, a sparse code multiple access (SCMA) based uplink ICI cancellation technique is proposed to enhance the performance of the cell-edge users by jointly decoding the desired signal and the interference signal at the base stations. Unlike the orthogonal transmission schemes such as fractional frequency reuse in 4G, the proposed non-orthogonal SCMA-based scheme allows several cell-edge users to share the same time-frequency resource blocks (RBs). The sparse codewords are designed to make each user spread its data on a small set of RBs to achieve frequency and interference diversity, and make the near-optimal detection feasible with moderate complexity through iterative message passing algorithm (MPA). Then an open-loop MIMO transmission scheme which combines SCMA with Alamouti code is proposed to achieve full space-time-frequency diversity. Simulation results show that when there is no correlation among resource blocks and the antenna configuration is 1×1, the proposed SCMA-based scheme has about 2dB gain for uplink cell-edge users in multi-cell scenario compared with the FFR-based ICI coordination scheme in LTE networks. The gain becomes larger when combined with the existing MIMO technique. In addition, the SCMA-based scheme is also proved to be more robust to the burst interference.

10.1109/wcsp.2015.7341306

Cite

Citations (11)

Research on Chinese Event Extraction Method with No-trigger

Yi Li Chaoyang Geng Dan Yang

Event extraction is a key task in the field of natural language processing, which aims to structure text in preparation for tasks such as subsequent knowledge graph construction. Most of the existing event extraction methods are difficult to handle well the case of containing multiple events within the same document and to distinguish efficiently for negative obfuscation samples. In this paper, we propose a model called RCEENT (A RoBERTa-wwm-based Chinese Event Extraction method with No-Trigger). RCEENT based on RoBERTa-wwm encoder and PTPCG (Pseudo-Trigger-aware Pruned Complete Graph) decoder to address the above problems. The RCEENT model can improve the accuracy of the event detection module without adding too much complexity by using multiple binary classifiers, and improve the results of the event extraction task by using graph pruning and PTPCG model for decoding. RCEENT achieves a 1.4% F1 score improvement in the event detection task and a 3.9% F1 score improvement in the event extraction task on Duee-fin compared to the latest baseline method.

Pruning

10.1109/iccnea60107.2023.00028

Cite

Citations (0)